Last week was Ed Wachenheim and Greenhaven: a double in three years, then the next idea. This is the other temperament. François Rochon’s average holding period is eight years, he has owned Berkshire Hathaway since 2000 and Alphabet since 2011. Despite the record, what he highlights every year is not his returns: it is his mistakes.

The Rochon Global Portfolio — the family account he has run since July 1993 and that every Giverny Capital client mirrors — has compounded at 14.7% a year against 9.9% for its benchmark, 8,596% cumulative. The US sleeve has done 13.9% against 10.8% for the S&P 500.

From a Letter to Omaha to $3 Billion in Montreal

Rochon originally trained as an engineer, but stumbled into investing through a book.

“I think the first book I read about investing was Peter Lynch, One Up On Wall Street. And in that book he explained that when you buy stock you buy shares of a company… And in the book, Peter Lynch talks about Warren Buffett and he names him the best investor of all time. So I said, well, I should learn more about this person. So I read The Intelligent Investor, Ben Graham’s book, and wrote to Warren because that was before the Internet. So I sent a letter to Omaha and he sent me a big book of all the annual reports, I think from 1977 to 1992.” — Meb Faber Show, 2025

He started managing family money in 1993, ran it for five years, and only then opened a firm. The Giverny name was inspired by Monet’s garden:

“I said to myself, well, I’d love to build something [like] that in Quebec, a kind of museum of the art of our time surrounded by beautiful gardens and flowers and water lilies and willow trees. And so I decided that first, obviously, I needed to get rich. So I needed a plan. I just had graduated from engineering. So I thought that probably that was not the the easiest path to wealth.” — A Conversation with François Rochon, 2024

The founding idea has not moved since:

“From the very first days of Giverny, the cornerstone of our portfolio management philosophy was to manage client portfolios in the same way that I was managing my own money… It is crucial to me that clients of Giverny and its portfolio managers are in the same boat! That is why we call our clients ‘partners’.” — Annual Letter 2025

And the founding philosophy:

“I have observed some investors generate returns that were much higher than the indices over the long term (such as Ben Graham, Peter Lynch, Lou Simpson, Warren Buffett, Charlie Munger, Bill Ruane, John Neff, Philip Carret, John Templeton, etc.) So I took the time to study the source of their outperformance and they all had one thing in common: they considered buying a stock as the purchase of a business and were all trying to buy these businesses at a meaningful discount to their intrinsic values.” — Annual Letter 2018

From an old letter (nothing has changed):

“…To have good opportunities of enrichment, we therefore need a climate of pessimism that brings good businesses to prices well bellow their intrinsic value… So the next time someone asks you ‘does the stock correction affects you?’ you can answer: ‘Certainly, it improves my chances of getting richer’.” — Annual Letter 2007

Philosophy & Process

Rochon writes the philosophy out every year in his letters.

1. You own businesses, not pieces of paper.

“The stock market is dominated by participants that perceive stocks as casino chips. With that knowledge, we can then sometimes buy great businesses well below their intrinsic values… There can be quite some time before the market recognizes the true value of our companies. But if we’re right on the business, we will eventually be right on the stock.” — Annual Letter 2024

The test he uses to tell an investment from a bet:

“One way to discern them is to ask yourself the following: if the stock market were closed for 10 years, would I be happy owning this security?… Owning a productive asset (and being able to value it) does not require a short-term stock price; in the same way that the value of a dairy farm is measured by the amount of milk it produces.” — Annual Letter 2021

And the one he uses for the people running it: “when you meet management, you ask yourself, ‘Would I like this man to marry my daughter?’ And if the answer is yes, it’s probably a good person to be partnered with.” (Talks at Google, 2017)

2. The yardstick is 5% above the index over the long term.

It has been on the first page of the letter, in nearly the same words, for at least two decades:

“Our long-term (and ambitious) goal is to maintain an annual return of 5% higher than the indexes. If stocks in general achieve an annual return of 7 to 9% in the future, our long term objective would be to maintain an annual return of 12 to 14%.” — Annual Letter 2007

The corollary is that he is never in cash. His arithmetic for why:

“If we look at the idea that our companies increase their intrinsic value by twelve percent per year, it’s like saying that every month that passes, our portfolio increases its value by one percent. So every month we are not invested, we leave one percent of opportunity cost on the table. So for me, that’s a losing strategy.” — Sommet de l’Investisseur, 2025

3. The Rule of Three.

He wrote it down at the end of 2002, after his best three-year stretch, as a warning to himself:

“At the end of 2002, I then set forth a new rule that was labelled ‘the rule of three’: 1. One of three years, the market will drop at least 10% 2. One of three securities bought will be disappointing 3. One of three years, our performance will be inferior to the indexes. I also added: ‘The rule of three is not linear. We can experience three good years in a row (as was our performance of the last three years) and afterwards have two, even four bad years in a row’.” — Annual Letter 2007

The point of it is psychological, not statistical:

“The best managers usually underperform one year out of three. So it’s still a very ambitious group of rules. But if you accept that from the start, when you have some bad years and some bad investments, you’ll be better prepared psychologically to deal with them.” — Talks at Google, 2017

He checks it in public. In 2016 he counted seven misses in twenty-three years for the US sleeve, “very much in line with our ‘rule of three’” (Annual Letter 2016). Check it against the table. Since 1993 the Rochon Global Portfolio has trailed its benchmark in eleven of thirty-three years. One in three, almost exactly.

4. Over long periods, value and price converge.

Since 1996 the letter has carried an “owner’s earnings” table: the growth in his companies’ earnings per share plus dividends, set against what their stocks actually returned. Thirty years of it now.

“Since 1996 – exactly thirty years – our companies have seen their intrinsic value increase by approximately 3,690%, and their stock market returns have achieved a total return of approximately 3,585%. On an annualized basis, our companies have achieved an intrinsic return of 12.9%, compared to 12.8% for their stock market performance… Over those three decades, the similarity between these two figures is not a coincidence and is significant.” — Annual Letter 2025

He would rather you judged him on the left column than the right, and he explained why fifteen years ago:

“Over 16 years, our portfolio has realized a return that is 5% higher than the S&P 500 for the simple reason that the underlying companies in our portfolio have increased their intrinsic value at a rate that is 5% higher than the average.” — Annual Letter 2011

The same table explains 2025 without excuses. His companies grew intrinsic value 13%; their stocks returned about 7% before currency; the Canadian dollar took another five points.

5. Risk lives in four places, and the fourth is you.

“In my view, an investor’s stock portfolio has four distinct levels of risk: portfolio diversification, the quality of the companies held, the stock market valuations paid, and lastly the investor’s own behavior.” — Annual Letter 2024

On the first: “a portfolio of twenty or so securities considerably reduces the risk of having too much concentration in a few securities.” On the second, the one he calls most important: high returns on equity, low debt, conservative accounting, a durable model. On the third: “a security that trades at 40 times earnings – all other factors being equal – is necessarily riskier than another similar security that trades at 20 times.” And on the fourth:

“In the 1950s, the average holding period for a stock on the New York Stock Exchange was 8 years. It is 8 months today (note: our average holding period at Giverny is 8 years)… In 2023, the average holding period for the S&P 500-based ETF (SPY) was 17 days! The very low holding period of ETF’s holders led John Bogle (the inventor of index funds) to say that concocting ETFs is the equivalent of giving matches to arsonists.” — Annual Letter 2024

He put the fourth risk more bluntly on Montreal radio in the 2022 bear market: “I think 90% of stock market success is not selling during downturns. Most people who have poor results in the stock market do so because they sell at the wrong time.” (Les Affaires, June 2022)

6. Be humble, remember the rule of 3.

“Humility is just as necessary to recognize our mistakes (inevitable in the investment world). From this perspective, humility becomes the catalyst that allows one to become a better investor by constantly trying to learn from one’s mistakes.” — Annual Letter 2022

Which is why the sell rule is borrowed from a book published in 1930. He cited it in 2020 to explain keeping one 2015 purchase and dumping the other:

“Keeping AMETEK and quickly selling Stericycle is in line with the rule stated by the legendary investor Philip Carret in his book published in 1930: ‘Be quick to take your losses, reluctant to take your profits’.” — Annual Letter 2020

The hard part is telling patience from its evil twin. His version, from the Google talk, involves a frog:

“So what you do, you put it in mild water and slowly increase the temperature. The frog won’t get out because it always thinks, ‘Well, things are getting a little hotter, but they’ll cool down eventually.’ And it never cools down, and eventually it gets cooked… There’s a big difference between knowing when to keep your shares because the fundamentals warrant it, and recognizing you made a mistake and just getting out of that stock. That’s a big difference between being patient and being stubborn.” — Talks at Google, 2017

The Podium of Errors

Every Giverny letter ends the same way — not with well-deserved self-congratulation, but with three medals for the year’s best mistakes and a post-mortem of everything bought five years ago. Most managers wouldn’t dare.

“It is with a constructive attitude, to always improve as investors, that we provide this detailed analysis. As is often the case with stocks, errors from omission (non-purchases) are often more costly than errors from commission (purchases)… even if we don’t see those on our statements.” — Annual Letter 2025

He’s been asked why he does it:

“It’s a very popular part of our annual letter. I don’t know why, because it’s very painful! But I think that’s very important if you want to continue to improve yourself.” — Talks at Google, 2017

The 2007 letter gave silver to W.P. Stewart, a fund manager he lost 80% on — “our worst loss ever” — with the line that explains the whole exercise: “I have no problems acknowledging my mistakes; there would be a problem if they would not be recognized and not acted upon.” (Annual Letter 2007)

The 2025 podium.

Bronze: Fiserv. Bought at $118 in 2023 because it was cheaper than Visa; the CEO left for the Social Security Administration in May 2025; he stayed; it deteriorated; sold at $74 in October.

“I should have sold immediately when the CEO left the company to enter politics… We invested in Fiserv, in addition to Visa, primarily because the stock was cheaper (approximately 15x earnings versus 25x for Visa). This was a mistake because I should have simply allocated that capital to Visa.” — Annual Letter 2025

Silver: Netflix, which he studied at 18 times earnings in 2022 after a 74% fall, decided had “several avenues to increase its revenue and, above all, its profit margins” — and didn’t buy. It rose 328%. “Unfortunately, I didn’t convert my enthusiasm into an investment.” He had said as much on the radio a year earlier, watching it add nineteen million subscribers in a quarter: “eventually, if the valuation becomes more reasonable, we might decide to buy it.” (Les Affaires, January 2025)

Gold: Canadian National Railway. Privatized in 1995, listed on the Toronto exchange he was already investing on, ignored. “In Peter Lynch’s book ‘Beating the Street,’ published in 1992, there was this rule: ‘Whatever the Queen is selling, buy it!’ But unfortunately, I ignored Peter’s rule and passed on it.” Seventeen percent a year for thirty years — “a better return than Microsoft‘s for the same period.”

The 2024 gold was worse, and he says so. Cintas, a uniform-rental company he had studied in person twenty-two years earlier, announced an acquisition in 2016 at 26 times earnings. He judged the synergies correctly and passed on price. Earnings quintupled; the stock rose 700%.

“It’s a business that fits exactly my style: not very exciting, hyper stable and recurrent… and also highly profitable… This is a worse mistake than Apollo or Ameriprise. Because in the case of Cintas, I understood the company and its competitive advantages perfectly. To use a baseball analogy: it was a pitch right in the middle of my strike zone.” — Annual Letter 2024

The 2023 gold went further than gold. Novo Nordisk, followed since 1998, visited in Copenhagen in 2014, never bought at scale: “My Novo Nordisk mistake deserves more than a gold medal — it deserves an Oscar for Lifetime Achievement.” (Annual Letter 2023)

Some medals get awarded twice. JPMorgan took bronze in 2012, when it fell to seven times earnings on the London Whale loss and he recommended it to everyone but himself:

“I even went ahead and enthusiastically recommended the stock on television and in the Montreal Gazette newspaper. But with our own money, I stayed on the sidelines. It was difficult to choose which of our other holdings to sell and I thought that we already had a large exposure to financial services in our current portfolio. But such an opportunity is rare and these excuses are regrettable.” — Annual Letter 2012

Twelve years later it was in the 2024 post-mortem again — this time as the bank he sold, having kept Bank of America instead:

“We subsequently sold JP Morgan but kept Bank of America — a mistake since JP Morgan did much better than Bank of America. Over 5 years, Bank of America has only modestly increased its EPS and its share price has risen from $29 to $44, an increase of nearly 50%. During the same period, JP Morgan increased its EPS by 60% and its stock rose from $139 to $240, an increase of 73%.” — Annual Letter 2024

Not every regret is an error. He was asked about Costco‘s results on Montreal radio in 2023 — less a mistake than discipline on valuation:

“Yes, that’s one of my biggest sources of sadness and regret, having sold Costco shares. It was several years ago. I held it for a few years and have continued to follow it. I think it’s one of the best companies in the world, a fantastic company… I would not want to pay more than twenty-four or twenty-five times earnings for this company. So, that’s a significant difference. We’re talking about fifty percent higher than the price I would probably want to pay.” — Les Affaires, September 2023

What Does Giverny Own Today?

Below is the full 13F book with share trends, price performance and fundamentals. Go on Alpha-File.com to view this table for every manager and click through each holding.

A quick note on the numbers. The 13F is $2.97 billion across 51 names at June 30, 2026, and it is not the Rochon Global Portfolio. It is every US-listed line across every account the firm manages — Rochon’s Montreal book and David Poppe’s New York book, run separately under the same roof since 2020.

Poppe, who ran Sequoia for thirteen years before this, described the arrangement last October:

“He doesn’t own anything I would not own, and I don’t own anything he would not own. We operate as one research team with a lot of communication… The portfolios are also probably two-thirds overlapping. We’re not running two wildly different enterprises under one roof.” — David Poppe, Graham & Doddsville, Fall 2025

Two consequences. Constellation Software, the letter’s largest holding and the stock that cost him 2% in 2025, is not in it — it trades in Toronto. Neither is Dollarama, which rose 46%. And the two books are legible if you know the letters: Heico (2016), Ametek (2015), NVR (2018) and Keysight (2019) are Rochon purchases he has written up; Medpace, Installed Building Products, Kinsale, Arista and Watsco are Poppe’s, and he has written those up himself. Some of those (Heico and Ametek) likely overlap.

I would describe Rochon as a “quality” investor, but one who can discover unfollowed or misunderstood names. He owns fairly little tech compared to most quality investors we’ve followed, is heavier in financials, and has more valuation discipline. He is trying to compound by owning “boring” businesses wherever possible.

His top five are 29%; nothing is above 7.2%. He tends to trim winners over time and add to losers. The June quarter’s biggest adds were to the names that had fallen — ResMed +69%, ADP +56%, Bright Horizons +46% — while Berkshire was cut 34%, Keysight 56% and Markel 28%. Buy the disappointment, trim the winner.

The spine — Alphabet, Meta, Visa, Berkshire (~25%)

Alphabet is the largest position across both share classes and the most profitable in the entity’s disclosed history, at more than $300 million. The tip came from Charlie Munger:

“A few years back, I asked a question to Charlie Munger about moats and all those great companies that seem to see their moats shrinking. He said, ‘Well, I know all those moats are getting filled with sand.’ So I asked him: which company do you think has an expanding moat? That was six or seven years ago, and he said, ‘Google. I think that’s an incredible company.’… And we bought shares in 2011 after that comment. We were lucky because at that time the P/E ratio was 15 times. If it had been a few points higher, probably we would have passed and missed an increase of 400% since then.” — Talks at Google, 2017

The letter that year explained the exception to his rule about technology:

“We are generally not fond of technology companies. We have observed over the course of several years a number of technological shifts that rendered obsolete once outrageously dominant companies. But in our eyes, Google isn’t really a tech company. It is first and foremost a service provider with the type of dominant brand we have seldom seen in our careers.” — Annual Letter 2011

It was a small position for years — “Google is a small position in our portfolio because we believe that the company could be a better allocator of excess cash flows” (Annual Letter 2012); “about 2-3%” in 2015 because of “its stock option program” (Annual Letter 2015). The capital-allocation complaint has become a $185 billion question, and this February he answered it:

“The good news is that the company has the means to pay one hundred and eighty-five billion in capital expenditures in a year. In the last quarter, they generated fifty-two billion in cash flow, so that’s over two hundred billion on an annualized basis. So they have the money. Now, is it an intelligent decision to make with that one hundred and eighty-five billion? Personally, I think they are doing this primarily to protect their core business, their search and advertising business, and so far, it’s yielding results.” — Les Affaires, February 2026

Meta is the 2018 purchase, bought “when the stock was very unpopular after the Cambridge Analytica scandal (which today the vast majority of investors have probably completely forgotten about)” (Annual Letter 2023). EPS $7.57 to $14.87 in five years; the stock $131 to $354. The better test was November 2022, with the stock at $93 and Wall Street calling it a metaverse money pit. He went on the radio the week of the layoffs:

“The key, the most important thing, is the number of monthly and daily users across all platforms, primarily Facebook and Instagram. And that is up 4%. So, people are not only staying on Facebook, but it continues to grow. That’s the key because as long as people spend time on Facebook and Instagram, the company will find a way to monetize it.” — Les Affaires, November 2022

Eighteen months later it had quintupled and he was calling it “a money-generating machine” (Les Affaires, April 2024). Poppe’s version of the AI case is one sentence: “It’s not a revolution at Meta, it’s an evolution.” (Graham & Doddsville, Fall 2025)

Visa was bought in 2010 into the Durbin-amendment panic, and the letter that year quoted a thirty-year-old Buffett interview to explain the attraction:

“Something he said always stuck in my mind: ‘The best types of businesses to be acquired are those which have the equivalent of a royalty on the growth of others.’ We believe that this applies to credit card companies… Visa’s balance sheet is immaculate, with no debt and $3.5 billion in cash ($5 per share). So, we are paying $66 for about $5 in estimated EPS for 2011—about 13 times profits.” — Annual Letter 2010

Fifteen years later it is “the strongest business model in the ecosystem and, above all, with the widest economic moat” (Annual Letter 2025) — which is exactly why Fiserv was the wrong way to spend the money. Asked at Google why he owned $35 million of Visa and $210,000 of Mastercard: “We prefer Visa, obviously. MasterCard we own a few shares just to follow the company… Visa has a little something more.” (Talks at Google, 2017)

Berkshire since 2000. The first big position was bought below book value in March 2000, when “Warren Buffett was seen as a has-been” — “That was a gift.” (A Conversation with François Rochon, 2024)

“During last September, the stock was trading near the level of $70. At that time, we increased our position in Berkshire by 50% and we consider this perhaps one of the best occasions to have bought this stock in our lifetime. If Mr. Buffett was a few years younger, we probably would have bought even more.” — Annual Letter 2011

It was cut by a third this quarter, and he told Meb Faber why before the filing did:

“The company has now more than $300 billion in cash, which is probably 28% of assets. So it’s going to be hard to compound at, let’s say, more than 10% a year when you have 28% of the assets in something that yields 3% or 4% a year… It’s not a question of quality, it’s a question of size.” — Meb Faber Show, 2025

The industrials — Heico, Ametek, NVR, Keysight, Analog Devices (~17%)

“Boring” industrial compounders have always been core to Rochon’s book.

Heico is 6.4%, the second-largest position, initiated September 2016. The 2021 post-mortem explains the moat in two sentences:

“Heico sells alternative (non-original) aeronautical parts to aviation companies. It is very difficult to obtain Federal Aviation Administration (FAA) approvals to supply aircraft parts. With its vast number of approved products, developed and commercialized over several years, Heico has a very significant competitive advantage… One of the reasons for our purchase in 2016 was our admiration for the members of the Mendelson family who run the company — this admiration has only increased since.” — Annual Letter 2021

Ametek was one of two purchases in 2015. The other was Stericycle.

“Ametek… is one of the oldest companies listed on the New York Stock Exchange (since its inception in 1930). Yet the company remains largely unknown and isn’t closely followed… As is often the case with exceptional companies, it’s the corporate culture that really stands out.” — Annual Letter 2015

Eleven years, +$152 million, the third-most profitable position in the entity’s history behind Alphabet and Berkshire.

NVR is a 2018 purchase, bought into a rate scare after twenty years on the watch list.

“It doesn’t own land on its balance sheet, which means it has a balance sheet with very few assets and therefore little debt. It has options on land and also builds houses much more uniformly and much faster than the industry average. I think it builds a house in three months, whereas the industry average is four months. It might not seem like much, but that’s a 33% difference… It’s a truly exceptional company, and I think it’s, in my opinion, the best in the industry.” — Les Affaires, March 2021

EPS grew 19% a year for five years and the stock tripled: Giverny sold out of NVR entirely in early 2023 and bought it back in March 2026 at a higher price. NVR is the same asset-light homebuilder Ed Wachenheim and Greenhaven use as the benchmark for his entire housing thesis, and Rochon’s version of that thesis is in the 2022 letter: “the number of new homes needed to return to equilibrium is over two million annually (compared to an annual construction level of one million).”

Keysight was bought in 2019 — “the stock still ended the year at a 92% higher level than when we first bought shares” (Annual Letter 2024) — and is +$90 million in the filing. It was cut 56% this quarter, and Analog Devices 45%. He explained both before the filing showed them, in a January interview about semiconductors reaching 20% of the S&P 500:

“I think we have three companies that could be considered directly or indirectly related to the semiconductor industry. And the three together represent barely 4% of the portfolio. So they are not large positions; they were larger positions before, but when their valuations became high recently, we reduced them. So we’re talking about Analog Devices, a company we’ve held in the portfolio for three years… Right now, it’s trading at 30 times earnings, so those are optimistic ratios.” — Les Affaires, January 2026

The financials — Schwab, Progressive, ADP, Markel, Bank OZK (~18%)

Schwab at 5.2% is the one being bought — up 9% this quarter:

“Charles Schwab was a more disappointing investment. EPS only increased at 5% annually, despite the large acquisition of TD Ameritrade. We believed Schwab would perform well during an interest rate hike and that was not the case… In our opinion, however, Schwab’s competitive advantages have increased over the past five years and the long-term outlook remains strong. This is the why we’ve decided to remain shareholders.” — Annual Letter 2023

He had cheered the TD Ameritrade deal the week it was announced — “I think it’s excellent news. I think Charles Schwab is the company with the lowest operating costs in the industry” (Les Affaires, December 2019). The 2025 letter adds the second reason: Schwab has been “penalized in the stock market not because of poor results, but due to fears that AI could affect them in the future. Wall Street’s typical approach is to sell first and examine the facts later.” (Annual Letter 2025) Rochon is still adding.

Progressive is another stock he sold and bought back — owned 1999 to 2007, back in 2019, EPS $5.65 to $14.04 since (Annual Letter 2024). Both managers own it, and Poppe’s description:

“One of the former CEOs who succeeded Peter Lewis, Glenn Renwick, liked to describe Progressive as a statistics factory. And it really is… Progressive today has about an 8-10 point underwriting advantage over the rest of the industry, which is unheard of. If anything, the margin is widening.” — David Poppe, Graham & Doddsville, Fall 2025

Poppe trimmed a third of his in 2025 on market share — “I think we’re somewhere around the sixth inning with Progressive” — and calls it “a melting iceberg, but it’s a very large iceberg.” Rochon’s reason for holding Progressive at 14-15 times earnings is simpler: “it’s known but it’s not like a darling.” (Sommet de l’Investisseur, 2025)

ADP is new this year and took the quarter’s second-largest add, to 4.9%. No letter has covered it yet.

Markel has been in the letters since 2013: “We first bought shares in 2013 when Markel announced its large acquisition of Alterra Reinsurance Company. Wall Street was skeptical of the benefits of this acquisition and the stock was trading at a compelling valuation. We seized the opportunity to become shareholders.” (Annual Letter 2014) Same pattern as NVR in the filing: gone in early 2023, back in March 2026, and then cut 28% in June.

Bank OZK is the longest story in the book. Bought in 2006 at half a normal weight because the price looked high — “refusing to invest in an exceptional business because the price seems a little too high is often a mistake” (Annual Letter 2007). He went to Little Rock to meet George Gleason, and this is the kind of bank he likes:

“I had the feeling that Mr. Gleason knew each loan on the bank’s books. Nothing exotic (or esoteric), no subprime mortgages, no incomprehensible derivative products: only traditional banking activities… Despite the widespread stereotype, risk and reward don’t necessarily always go hand in hand.” — Annual Letter 2011

Then he sold most of it in 2018 to pay for Schwab, and graded himself five years later — the filing shows the position rebuilt to 2% in the quarter after the grade was published:

“This postmortem would be incomplete without adding that we used a portion of the Bank OZK sale to finance the Schwab purchase. We kept a small weight in OZK, but, looking back five years, it is clear that this transaction was a mistake.” — Annual Letter 2023

Twenty years in, it’s 2.1%.

The New York book and the podium alumni — Medpace, Installed Building Products, Arista, Brown & Brown, Lululemon, Floor & Decor

This is the part of the book with significant overlap with David Poppe’s Giverny Capital Asset Management, and it can be hard to parse out the positions exactly from the 13F. A few of the names that likely overlap:

Medpace is 4.5% and has done exceptionally well:

“Medpace was founded in 1992 by August Troendle, a former medical review officer for the U.S. Food and Drug Administration. He owns 17% of the company… Medpace has doubled its profit over the past three years.” — David Poppe, GCAM Q4 2024 Letter

Installed Building Products, bought in 2022, is a scale story told in one line:

“As a result, IBP can say it can install insulation in a new home, including labor, for less than a homebuilder can buy the fiberglass.” — David Poppe, GCAM Q4 2024 Letter

Arista was Poppe’s largest position at the end of 2024 and has been trimmed since, another 20% in June.

“With Arista, I wish I could say the original thesis was AI, but it was not. The thesis years ago was cloud computing data centers, and the idea that demand for bandwidth would grow for the rest of my career… I wrestle with this with Arista now. The growth looks great, but I can see a path where earnings double in a few years, yet the stock goes nowhere because you started from such a high multiple.” — David Poppe, Graham & Doddsville, Fall 2025

Then there are the medals he converted into positions.

Brown & Brown he owned from 2004 to 2009 — “the best business in the sector: its margins highly exceed its competitors” (Annual Letter 2007) — sold, and gave the 2023 bronze to what came next:

“I don’t think selling the stock in 2009 was a mistake. What was a mistake was to stop following the company. I could have noticed the improvement in Brown & Brown’s business fundamentals… One other conclusion is that you should always continue to follow the companies you sell. Things can change for the worse, but also sometimes for the better.” — Annual Letter 2023

He started following again, bought in December 2024 at about $100, doubled the position at $78 a year later. It’s $72. He now cites it as the model of a company that compounds through acquisitions rather than margins (Sommet de l’Investisseur, 2025).

Lululemon is interesting given where the shares are. In 2018:

“But the day we were about to buy, the CEO resigned for ‘lack of leadership’. So I opted to wait to see who the new appointee would be. Meanwhile, financial results came out and were very strong. The share price then climbed quickly and I decided to wait for a better price. At the time of writing this, the stock is 88% higher.” — Annual Letter 2018

The filing shows a 0.1% placeholder from late 2017, and then the real position: built in late 2022 and 2023 at an average cost of $302, after the stock had quadrupled from where he passed. In September 2023, with the position at its 2.5% peak, he was still enthusiastic: “it’s a company that truly has the wind in its sails, with very strong growth” and, next to Costco, “a lower earnings multiple and a considerably higher growth rate” (Les Affaires, September 2023).

In August 2024, after the first halving: “I think Lululemon’s business is intact… at nineteen times earnings, I think it’s very reasonable… So I think this could be an opportunity.” (Les Affaires, August 2024) By 2025, the tone had changed:

“When we acquired it, it was about three years ago, I think we paid $300. You’re right, it’s $160 today, it even went up to $500 at one point… I would say another question to ask is, if we didn’t own the stock in our portfolio, would we buy it again? When the answer to that is no, perhaps not, it usually starts to look like a mistake… I’m not convinced enough to buy more. So I would say it’s on hold for the moment.” — Sommet de l’Investisseur, 2025

It’s $119 now, trimmed to 0.5%, the second-largest loss in the ledger at −$26 million.

Floor & Decor was the 2020 bronze, missed by a day: “The stock then climbed more than 10% on the very day we wanted to buy it and we hence decided to wait for a return to ‘our’ price. Results improved quickly and the stock never returned to ‘our’ price.” (Annual Letter 2020) It arrived in the filing two years later at 0.5%, cost $72, price $59.

Summary

The P&L chart covers the entity’s own filings, 2015 to now — not the thirty-three-year record, but the part you can check.

CarMax, −$38 million — the biggest loss in the ledger, and the most-documented position in the archive. Bought in 2007 at $21; “During its low of November 2008, Carmax was trading at $7… we should have bought more stock during that low had we had foresight in addition to patience.” (Annual Letter 2010) At Google in 2017 it was his example of what patience looks like:

“CarMax stock over the last decade has been quite volatile. It went down 25% a few times. But if you look at the earnings over those 10 years, it’s been a pretty stable business. If you just focus on what’s happened to CarMax over the last 10 years, besides 2009, earnings went up every year, and it grew earnings by more than 16% annually. Just focus on that and forget about the stock price movements; you’ll be okay.” — Talks at Google, 2017

In October 2022, with the stock under $70: “I think that within, let’s say, five years, this is a company that can earn almost $12 per share in profits… So I think someone who is patient and acquires with a long-term horizon will be rewarded.” (Les Affaires, October 2022) Then Carvana, omnichannel, a CEO change, and 2025:

“We had to recognize that we had been patient enough and decided to sell. Ultimately, we achieved a positive return on this investment over the 18 years we were shareholders; however, it was a very low number and well below our targets.” — Annual Letter 2025

Eighteen years, positive return, still a mistake by his own scoring. Poppe files it under the one reason to sell that isn’t valuation: “in a case such as Carmax, where a strong competitor emerges.” (Graham & Doddsville, Fall 2025)

For me, the most interesting thing will be how Rochon deals with AI. He makes a good point on Alphabet but his thesis rests on Google trying to protect its existing moat. Does that thesis still hold as we see astronomical increases in capex? What happens if search revenue starts to decline?

Another interesting thing will be seeing how he handles Constellation Software. According to his letters, he has trimmed it above 10% before and sold the spin-offs. He kept the core through a 26% fall and a CEO’s near-death, on the argument that “it seems unrealistic to believe that an operator of a dozen mini-golf courses is going to start coding their own AI-powered reservation software to save something like $5,000 in annual fees!” Will this point of view change?

I’ll be tracking every filing on Alpha-File.com, where you can pull this same view — position history, initiations and exits, fundamentals, cumulative P&L, and every Giverny letter back to 1995 — for Giverny or any manager, free. If you enjoyed this, subscribe: the next manager profile in the series is already in the works.

Nothing here is investment advice. Position data comes from SEC 13F filings as of June 30, 2026, as aggregated on Alpha-File.com; the 13F is long U.S.-listed equity only, aggregates all accounts the firm manages, and does not show Canadian or other foreign holdings, cash, or shorts. Quotes from Les Affaires podcasts are Alpha-File’s English translations of the French originals.

Edgar Wachenheim III founded Greenhaven Associates in 1987 in Purchase, New York. Saurabh Madaan, introducing him at Latticework in 2018 as the best investor you’ve never heard of. He wrote Common Stocks and Common Sense about his own process and the record on the book flap is 18% a year from 1990 to 2021. He is 88.

The June 2026 13F is $8.66 billion across 28 names. Lennar, Toll Brothers, Pulte and D.R. Horton are 37.8% of it before you get to General Motors. In October 2025 he told a conference the market was at 28 times and Greenhaven was 45% in cash — “historic for us.” In April 2026 he said it was 48%, and he had only held more once in his life. Then the June quarter printed, and in it he had opened an $860 million position in Becton Dickinson, $421m in Accenture and $148m in IQVIA and $47m in EPAM: $1.5bn deployed in a single quarter to two sectors battered by AI fears.

Background

Born 1937, New Rochelle. Two years at MIT, then Williams, a short stint at IBM, Harvard Business School, and Goldman Sachs as an analyst. He left Goldman because the hours were eating the family, and joined the investment arm of Central National-Gottesman — his in-laws’ paper company — ran by Arthur Ross. When Ross retired he took it over; in 1987 he spun it out as Greenhaven. For the first few years the only money was the family’s. Endowments came later.

At Central National, Ross told a thirty-year-old Ed to get out from behind his desk: “You’re not going to learn a lot sitting at your desk and looking at annual reports. Go out and become a director of companies.” He did — three boards in fact.

As it turned out, he did learn something that would prove significant throughout his career. On the board of a bakery in Kansas City he sat next to Jim Grosfeld, then the CEO of Pulte:

“I used him as my professor. I called him Professor Jim. And I said, ‘Jim, would you teach me something about the homebuilding business?’ And I bought some beers, and I bought him some coffee, and I got a good understanding of the fundamentals of the industry from the professor.” — Business Breakdowns

He has been in that industry, in one form or another, ever since. He bought into U.S. Home in the mid-1990s at 0.6 times book and under six times earnings — “adrenaline flows for value investors.” He bought heavily around 2000, sold in 2005 because prices and starts both looked above normal, sat out the levered builders after the crash and owned Home Depot, Lowe’s, Masco and Whirlpool instead, then went back into the builders themselves once the debt was gone.

Philosophy & Process

I would describe Ed as a “classic value” investor. He’s looking for areas of the market ignored by the market and always paying a sensible valuation. Ed explicitly outlined his process in Common Stocks and Common Sense.

1. Buy undervalued stocks. Ignore the market. Cash is what’s left over.

“It’s a very simple strategy. We do not pay attention to the big picture. If we can find enough stocks that are attractive to us, even though the stock market may be very high — which it is today at about 28 times earnings — we will buy the stocks and be fully invested. If we can’t find enough stocks that are attractive, we’ll hold cash.” — Latticework 2025

The version at Columbia University version six months later: “I have no view on where the market is going and I do not think anybody else does either… The height of the stock market does not directly set how much cash we hold, but the cash position is the residual of that.”

2. Use a common sense approach to valuation

Ed looks two or three years out, projects earnings under normal conditions, and puts a normal multiple on them. Since the S&P has averaged about 16.5 times over sixty-five years, an average company is worth 16.5 times.

Then the buy rule: he needs to see a double. “If we can buy the stock cheaply enough — and usually that means we want to double our money in two or three years to get a return close to 20%, and we make plenty of mistakes — we need to be able to see a double.” A double in three years is a 30% CAGR; he expects to be wrong often enough that it works out to 20%.

Why that horizon?

“This is why we look out 2–3 years, because the outlook for 2–3 years is a pond that very few others are fishing in. Most investors are fishing in the pond of whether the stock is going to appreciate in the next 6 months to one year, and that’s a very overfished pond.” — Graham & Doddsville, Spring 2026

Someone at Latticework asked whether AI has retired the 16.5. “I don’t want to be the skunk at the picnic, but the last time stocks sold at this price because of a concept — the internet, the new economy — it ended up very badly. From top to bottom, the Nasdaq went down about 75%.”

The book’s fundamentals over time. Cheap on earnings, clean on debt — the two things he screens for, in that order.

3. Balance sheet first. Then make money on the income statement.

“We are sticklers on the balance sheet. We are risk-averse, and you protect your net worth on the balance sheet, and then you make money on the income statement.” At Columbia: “The people who jumped out of windows in 1929 had debt. The companies that went bankrupt in the Depression had debt.” This is why he would not own the builders in 2010 — “the balance sheets were too weak” — and why he owns them now. “If a company has more cash than debt, and some of the homebuilders do today, as does General Motors, that matters a great deal.”

4. Be contrarian

Per a 2016 office visit: “Most people in the investment business are followers and they don’t have the ability to go out with confidence and say, ‘Yes, everybody thinks this, but I’m going to do that.’” At Columbia: “I do not think you can make money unless you are a contrarian.” Ideas don’t come from the mainstream — “by the time I read something on Bloomberg it has been seen by the general public”.

He has a name for the bias he’s fighting. “Excessive attention to current news, particularly negative news: what I call proximity bias. The Iran war will come and go. Spiking gasoline will come and go. But as investors we over-focus on short-term negative news, and it makes us make poor, emotional decisions.” And he will not take management’s word for anything.

5. Fifteen names

“If we owned 25 stocks instead of 15, the 16th through 25th names would almost by definition be riskier. Portfolio managers generally do not want to buy low-return stocks, so they compromise on risk to try to make money — which is why over-diversification can end up giving you a riskier portfolio, not a safer one.” — Graham & Doddsville, Spring 2026

The policy: no position above 12% at cost, and in practice the biggest have been 8–10%. He owns close to 6% of Toll Brothers and 5% of Lennar and does not want more, because liquidity is a weapon. The 13F has 28 lines, but the bottom six are $1 million stubs and the top ten are the fund.

6. Reverse when you’re wrong

“When you’re wrong, you need to recognize you’re wrong, and be willing to reverse yourself and sell the stock before it goes down more. It’s very bad to let your mistakes fester.” Whirlpool is the example: “They made an absolutely dumb acquisition, paying a very high price, leveraging their balance sheet. As soon as I saw that management made a major mistake, we had to sell our stock.” And the exit rule on the other side: “We buy stocks that we think are undervalued for a particular reason. When that reason happens, we tend to sell. We hold typically for three or four years and then go on and do something else. I don’t think we could achieve 15% to 20% per year if we held stocks for long periods of time.”

Regular readers will hear the contrast. TCI holds Moody’s for seventeen years. Hibbert wants duration measured in decades. Ed wants a double and then the next idea. The reason? Ed is not shopping for compounders. He’s looking for businesses with strong balance sheets that are underpriced but once they hit fair value, he’s gone. There is no point owning a homebuilder for 20 years.

Historic Cash

Here is what he said on stage in October 2025:

“Because the market is so high, many of our stocks reached levels where they were no longer attractive on a risk-reward ratio, and we sold them. With the market this high, we are having great difficulty finding new ideas. We’re 45% in cash. And that is historic for us.” — Latticework 2025

Six months later, to Columbia: cash was about 48%. He had only held more once. “We do the analysis to determine what a company’s going to be worth, and we can’t justify purchasing many given the height of the stock market.”

Then the June 30 filing printed. A new $860 million Becton Dickinson line at 9.9%, the fourth-largest position in the fund, built from nothing in one quarter. Accenture new at $421 million. IQVIA new. ICON raised 55%. He added to Lennar and to GM on the way in. And the sells that quarter total only $17 million. Ed is finding value!

The June quarter in one picture. Almost nothing sold; healthcare, IT services and more housing bought.

He has not mentioned BDX on any of the six transcripts we hold. But it is not a new idea either. Pull the position chart and there is a prior Greenhaven holding, about three million shares, from late 2010 to early 2012 — bought around $55, sold around $50, a small loss. Fifteen years later he came back with $860 million at an average cost of $151. At the end of August the stock was $188, and the position summary reads +$172.9 million on the round trip.

Becton Dickinson, with Greenhaven’s buys and sells on the price line. The 2010–12 round is on the left; the June 2026 entry is the green marker at the right, at a price last seen i

What Does Greenhaven Own Today?

Below is the full book with share trends, price performance and fundamentals. Go on Alpha-File.com to view this table for every manager and click through each holding, all exportable for free.

A quick note on the numbers: weights are against the $8.66 billion disclosed equity book, not the firm’s total assets, so everything here is roughly a third bigger than it would be against the ~$11.7 billion he actually manages. The 13F shows no cash and no foreign lines.

The Homebuilders — Lennar, Toll, Pulte, Horton, Meritage (~39%)

Ed’s thesis? Demand you can count on: “We need about 1.5 million housing units per year to satisfy the needs of our growing population, and about 400,000 houses per year are torn down.” Cumulative underbuild of two to four million homes. At 1.7 million starts a year it takes twenty years to close.

The business used to be bad. Owning the house was fine; owning the dirt was not. “Owning and building the house is a very good business. Owning the land is a poor business, because land typically appreciates at 3–4% a year.” He and Grosfeld argued for optioning land with Centex’s CEO in 2005 and were told they didn’t understand the business. NVR already had the model. The rest of the industry eventually got religion, and that is the variant perception he thinks is still not in the price:

“The analysts tend to be very short-term oriented, and I think they have completely underestimated the change of the business. They consider it still a real estate business. It’s not. It’s a manufacturing business. The ownership of land is incidental.” — Business Breakdowns

Horton’s option land is 75% of what it controls; net debt of $2.3 billion in 2013 became $600 million of net cash; ROE went from 10% to 22%. The large builders’ share went from 39% in 2018 to 58% in 2025, and because they’re asset-light and buying back stock, Horton’s revenue per share compounded at about 15%. “That is approaching the growth rate of a growth stock, and yet they trade at depressed multiples.” The thesis in one sentence: if NVR has averaged 16 times for a decade, and the others are now asset-light too, there is no reason they should trade at ten.

Lennar. The green markers are four consecutive quarters of adds into a falling price.

Lennar is the favorite because it stumbled. “They made a temporary mistake this year. They left the stock behind the other large home builders, so I focus on Lennar, even though we own Horton, Toll, and Pulte, and all the stocks are cheap.” On stage in October he built the model out loud: 100,000 homes in two years at $420,000, $42 billion of revenue, 13.3% margins, $19.75 of normal 2027 earnings, times 16, equals $315 against a then-$127 stock. The June 13F has it at $90. He added shares.

D.R. Horton — the position he walked through for an hour on Business Breakdowns, and the one he uses to explain the whole industry.

Toll, Pulte and Meritage.

He knows the near term is ugly. “They were doing fine until November 2024, at which point mortgage rates went up by about 1%. Buyers experienced sticker shock and activity has remained subdued since.” And he is willing to wait: “People have to live somewhere. They can defer for six or nine months; they cannot defer indefinitely.” Eight years after he first told CNBC these companies deserved 16 times, the multiple still hasn’t agreed.

The “truck company” — GM, Lear (~14%)

Not a car company. He corrected Jim Cramer on stage in 2018 — “You said the autos. I consider it a truck company” — and he is still saying it at Columbia:

“Automobiles are a lousy business and you can put that exact word in your newsletter. About 40% of the vehicles that GM makes are the on-frame cars, but they account for about 60% of the revenues and virtually 100% of the profits. There are only three major producers in the United States. Margins are typically 17% and the business throws off a lot of cash — most investors do not appreciate any of this.” — Graham & Doddsville, Spring 2026

The change he thinks is still not in the price is the buyback. GM’s share count fell 33% in three years; revenue per share went from $108 to $190, a 21% CAGR. “If you had a 21% CAGR in revenue per share for a company, you would call it a growth company.” He has owned it for 31 quarters and speaks regularly with the CFO. He cut a third of the shares in Q4 2025 and another 17% in Q1 as it got too big, then added back in Q2. Lear makes the seats.

GM — the single most profitable position in Greenhaven’s disclosed history at +$942 million, and still 11.5% of the book.

Lear — the seats and wiring inside the trucks, 2.3%.

Healthcare — Becton Dickinson, ICON, Baxter, Avantor, IQVIA (~24%)

BDX was above. ICON — a clinical-trials contractor, the same name Glenn Greenberg just made his largest position — was already in the book and got a 55% add to $539 million. IQVIA is new at $148 million. Baxter came in during 2025. Avantor keeps getting bought and keeps going down. Ed hasn’t discussed these publicly but its not hard to see the thesis: a beaten down sector post the COVID boom trading at decades-low valuation.

ICON. Two concentrated value investors, same name, same quarter — Greenberg at 12%, Wachenheim at 6%.

Baxter, Avantor and IQVIA. Baxter arrived in 2025; Avantor has been bought steadily into a falling price and is the fourth-largest loss in the firm’s history; IQVIA is new this quarter.

Oshkosh, Arrow, Avnet (~13%)

Oshkosh he did walk through: fire engines, garbage trucks, access equipment, the Postal Service van. Long contracts with no inflation protection crushed margins; the new contracts have it. “The company is projecting for 2028 between $18 and $22 per share. Use $18, put a 16 multiple on it, you get something like a $300 stock. The stock has been selling at about $135.” Arrow and Avnet are the electronics distributors — he’s been in both for years and trimmed both this quarter.

Oshkosh. Added in the June quarter; the stated target is roughly double.

Arrow and Avnet — the electronics distributors, held for years, both trimmed this quarter.

The drawer

Accenture is new at $421 million and unexplained. The oilfield sleeve — SLB, NOV, DNOW — is the one he did explain, and it’s shrinking: the rusty-iron thesis “has not played out yet,” they never got the full position, and SLB was trimmed 18%. VBIL.

Accenture — $421 million, new, and unexplained on any tape.

The oilfield sleeve: SLB trimmed, NOV held, DNOW a rounding error. The rusty-iron thesis, dying of neglect.

The Ghosts

Here is where the P&L chart looks like. Investors focused on compounders often have 1 name be the largest tile by far: Apple for Berkshire Hathaway or SpaceX for Atreides.

Every position since 2010, sized by cumulative P&L. The banks he no longer owns are three of the top four tiles.

Ed is a value investor. He doesn’t let things run. He is disciplined on valuation and once something reaches intrinsic value, its out of portfolio. What is impressive is how he’s achieved is record: consistency. Most of the boxes are the same size.

It’s also worth pointing out how much money he’s made in financials. Citigroup made $847.6 million. JPMorgan made $841.9 million. Goldman made $818.6 million. Three of the four most profitable positions in the firm’s disclosed history are banks he no longer owns. They didn’t die. They doubled.

Goldman Sachs. Pitched at $230 with a $400 target in 2018; sold down through 2025 for +$818.6 million.

Citigroup and JPMorgan — +$847.6 million and +$841.9 million. The 2018 Delivering Alpha bank book, harvested.

Ed moves slowly. Then, as the June quarter shows, he doesn’t. For me, it will be interesting how Ed navigates the AI disruption volatility. If he’s comfortable enough to buy Accenture and EPAM, will we actually see a SaaS company in the portfolio? Will he ultimately be proven correct on his mega homebuilders bet? How will healthcare play out?

I’ll be tracking every filing on Alpha-File.com, where you can pull this same view — position history, initiations and exits, fundamentals, and cumulative P&L — for Greenhaven or any manager, free. If you enjoyed this, subscribe: the next manager profile in the series is already in the works.

A note on the data: All portfolio data and charts in this piece are sourced from Alpha-File. The same position-history, trading, composition, industry, P&L, and fundamentals exhibits are available for every tracked manager on Alpha-File, including 13F filers, US N-PORT mutual funds and ETFs, and supported European, Canadian, and other global funds. Public-company and manager names throughout link directly to their specific Alpha-File pages.

The broken arc

Most managers profiled in this series have a clean shape: build the record, compound the reputation, scale the franchise. TCI still runs the book its founder started. AKO‘s founder handed his firm over on his own terms, with eight years of preparation. Even Lone Pine‘s succession was planned.

Francisco García Paramés is the first with a broken arc — a career cut in half by a boardroom war, with a two-year exile in the middle and a five-year performance struggle thereafter followed by a spectacular recovery.

The Investor: an engineer of process

Paramés was born 1963 in El Ferrol, Galicia, studied economics at the Complutense, did an MBA at IESE in 1989, and then went straight into Bestinver — the in-house asset manager of the Entrecanales family’s Acciona group. In his own words:

Francisco García Paramés, Ben Graham Centre / Ivey Business School value investing conference, Lisbon, 2025:

“I didn’t know that I wanted to invest in the stock exchange — actually I explained in the book that I thought that the stock market was a thing of speculators… they made some mistakes, so we stopped doing this private-equity-style investment for these families. There was a small brokerage operation. So I just sat there and started to analyze the Spanish companies from A to Z… it has 150 companies. In one year — I mean, they left me alone, so I was happy on that — I could more or less have an opinion on most of the companies. Then the team left… and I asked, can I do that?”

He taught himself, out of books, on a boat:

“Exactly at the same time, 1990, I was going to Mallorca on the ship and I was reading Peter Lynch’s book… I was starting to invest in 1990 in Spain, so you can understand that there was not value investor or anything that was related to that. The language didn’t even exist at that time. So coming across Peter Lynch, Warren Buffett and all the guys was a kind of a life saving for me — when all the bad times came, precisely for instance in ‘98, ‘99, you knew Warren Buffett was there and was having the same issues. We all remember the front page — ‘where are you going, Warren?’ ‘Where are you going, Francisco’ was exactly the same thing.”

Many managers in this series had a mentor to help shape their philosophy: Armitage at Egerton helped mentor Nicolai Tangen at AKO whilst Steve Mandel at Lone Pine helped mentor Mala Gaonkar at SurgoCap.

Paramés developed his philosophy purely from books, thousands of miles from Omaha or New York City, in a country where the value investing vocabulary did not yet exist.

The second formative layer is the Austrian school, absorbed in the mid-2000s through Jesús Huerta de Soto’s Madrid circle — not as a forecasting tool but as a filter. Money is continuously debased; real assets beat monetary assets over any long horizon.

“As I say in my book — the last thing I say, if you learn something: inflation is going to kill your portfolio. That was 10 years ago. And I still think the same. With electoral cycles, governments have to print paper — paper money — and that’s a fact… there is no other option than owning assets, and that’s what we do… That’s what happened in Germany in the 20s of the last century, or in Argentina at the beginning of this century. We will suffer some hiccups here and there, but we will do fine as equity investors. I would recommend not to have anything in cash — both in the bank, or bonds, or whatever — because you will lose most of the money.”

Another major life event: the impact on investing, if any, unclear. On 7 March 2006, the light aircraft carrying Paramés from Madrid to Pamplona crashed. The pilot and Bestinver’s CFO died. Paramés walked away.

Act I — Bestinver: the record (1993–2014)

Bestinfond compounded at 15.74% annually from January 1993 to Paramés’s exit in September 2014 — a cumulative +2,384%, taking the NAV from €6.06 to €150.58, a 24.8× over twenty-two years. Bestinver Internacional, launched in 1998, returned 10.28% annualized over its sixteen years under him.

What the headline hides is how violent the ride was. The 2008 drawdown took the funds down more than 40% before the 2009 recovery — a dress rehearsal, as it turned out, for what was coming a decade later. AUM grew from tens of millions to roughly €8–10 billion by 2014, making Bestinver comfortably Spain’s largest independent equity manager and Paramés “the Warren Buffett of Europe” in every lazy headline, an epithet he visibly dislikes.

Paramés is best described as an old-school value investor and on the deeper end of value. He isn’t on the hunt for compounders and isn’t a connoisseur of company “culture”. Rather, his approach is about finding undervalued assets, buying them for 50 cents, and selling when they approach fair value and that intrinsic value may not be compounding to the upside. It’s an old school approach that has largely been forgotten by the quality-obsessed approach we’ve seen take over investing circles over the last 2 decades.

He doesn’t necessarily own the world’s greatest businesses and several things about his process follow from that.

1. Publish the intrinsic value, and trade to it. Every position carries an intrinsic-value estimate, and the portfolio’s aggregate target value and implied upside is published, quarter after quarter. It expands in drawdowns, contracts in rallies, and he manages to it. There is no point letting these businesses compound and deliver to the upside as in many cases, these are asset plays with just one puff.

“You use the target price of the portfolio and our shares, and depending on how the company is behaving vis-à-vis the rest of the portfolio, we act — we increase the position, or we decrease the position [if] the stock price is going up… we have a family of 50 stocks that we love, we like them a lot, and if one stock goes up 20% and another stock goes down 20%, it’s a 40% change. It doesn’t make any sense not to do anything about that. This is a probability game.”

Paramés describes it as: “in terms of names it is not very large, but in terms of the trading that we do, actually it is about 100%.” Names persist; weights churn constantly. European taxation rules help — “in Europe we have the good thing that we don’t pay taxes when we sell stocks [inside the fund]. So it’s almost friction-free.”

2. Broad diversification and valuation discipline. When you’re not owning the world’s best businesses, you need to diversify. Paramés runs 40–70 names with position caps but a weightier top ten; concentration lives at the thesis level — shipping, energy infrastructure, family holdings — rather than the individual stocks. This diversification is accompanied by strong valuation discipline. The filter is: family-owned, ~30% return on capital, a business he understands, around seven times free cash flow, then a 1% starter position built slowly. “It’s a long time since we don’t do that… it’s very rare that we go directly into a 3–4% position.”

3. The benchmark is noise; the clients are not. “All my savings are in the fund. So I don’t give a damn about the benchmark.” But also: “we have 35,000 clients — anyone can be a client for €100, so we are very democratic on that… you have to do it right and do it properly.” Passive flows are the “frenemy”: “the more the passive investing trend develops, the better the market for the ones that are thinking before investing… the enemy part is that over the short term you have issues with performance versus the market, the clients, and all that noise.”

4. Patience, cubed. “We have to be patient in buying, patient in holding, and patient [in selling]… three years, which may be the holding period of most people — but three years is a lot for most people.”

5. Management meetings are optional; the circle of competence is not. On large caps: “meeting the management of BMW or Mercedes-Benz — what can they add in that sense?” On mid caps: “of course, this is key.”

He is the most “traditional value” investor of everyone we’ve profiled but even he has admitted towards tilting to quality.

Francisco García Paramés, Ben Graham Centre / Ivey Business School value investing conference, Lisbon, 2025:

“The catalyst was Joel Greenblatt — the small book about quality and performance with low price versus other stock. We went into that, we analyzed that, and we said we had been making mistakes going into the crisis — the financial crisis — with perhaps a lower quality than it should be in the portfolio. So from that moment on we increased the degree of quality in the portfolio, and we started to focus on high return on capital companies. And that’s the same thing we do today.”

The rupture — September 2014

The Bloomberg scoop landed mid-morning on 23 September 2014: Paramés was leaving the firm he had built, effective immediately. Not for retirement — over control. The Entrecanales family owned Bestinver outright; Paramés wanted equity, independence and institutional permanence for his team; the family said no. Bestinver bled billions in redemptions within months.

Then the strangest chapter: a two-year non-compete, spent partly in London, during which he wrote Invirtiendo a largo plazo (2016; published in English as Investing for the Long Term, 2018) — part memoir, part Austrian primer, part process manual.

The Bestinver engine from the mid-2000s was a trio: Paramés, Álvaro Guzmán de Lázaro (2003) and Fernando Bernad (2007).

Bestinver fissioned into two competing Madrid value shops. Guzmán de Lázaro and Bernad left Bestinver in January 2015 and founded azValor in December 2015 with Beltrán Parages.

Paramés did not join them. When his non-compete expired he launched Cobas instead — and in October 2016 five of azValor’s six analysts, plus its head of major clients and its head of trading, left to join him.

Act II — Cobas: the five bad years

Cobas launched Selección on 31 December 2016 and the flagships in March–April 2017. The four main funds held €1,442M by end-2017. What followed was a public stress test.

The per-fund record (clase C, the original 2017 series continued):

The shape of the hole: launch → December 2018, −27.5% against −4.8% for the index; launch → the COVID print on 31 March 2020, −52.7% against −7.1%. Seventeen consecutive losing months at one point. Citywire ran a “Paramés has ‘no answer’” piece.

What he actually owned while it happened

The composition history for Cobas Internacional runs across 27 reporting periods: Golar LNG, the Teekay complex (Corp, LNG Partners, Tankers), International Seaways, Babcock International, Currys, Renault, Danieli, CIR/Cofide, Atalaya Mining, Israel Chemicals, Maire Tecnimont, Wilh. Wilhelmsen, Samsung Electronics, Hyundai Motor. Shipping, energy infrastructure, Italian family holdings, UK electricals, Korean preference shares. Deep value, globally sourced, and almost perfectly designed to be destroyed by 2018–2020 growth/quality run.

Aryzta — the one that hurt. The Swiss-Irish baker was the top holding at 8.9% of Internacional at end-2017 and collapsed through 2018. Then Cobas did the thing that makes this position worth a section of its own: after fighting management to an EGM over the CHF 800M capital increase in October–November 2018, it underwrote the raise. The weight went to 9.7% at December 2018 and peaked at 10.2% in March 2019 — which is to say the position was at its largest after the collapse, not before it.

The de-risking then took four and a half years: 8.5% by June 2019, with the public promise “aumentaré mi inversión en Cobas” as personal penance; cut to 4.8% in the COVID quarter; 5.7% at end-2020; 4.3% at end-2021; 1.9% at June 2023, and gone from the book by December. The outcome, on Alpha-File’s reconstruction: −€38.7M at 0.6× MOIC over five and a half years, an IRR of −18.3% and a total return of −37%. Grandes Compañías held it too, for a further −€2M at the same multiple.

That makes Aryzta, by a distance, the most expensive position in Cobas’s history — nearly four times the next-largest loss. His verdict, from Lisbon, without naming it:

“Probably the main mistake I’ve done in my investing life — which was six years ago — was in terms of the sizing: the position sizing was too big for the quality of the company.”

The quote is precise about which mistake it was. Not the thesis, not the price paid — the size.

Throughout this whole time, Cobas kept publishing their intrinsic value estimates and kept seeing the discount widen.

The vindication (2021–2026)

The same book that halved then compounded: +34.5% (2021), +10.8% (2022 against −9.6% for the index), +11.6%, +23.4%, +25.8%, and +22% YTD 2026.

Bloomberg, in July 2026, noted that Cobas Internacional had achieved an average annual return of 20% over the past five years and was up 21% on the year, in the 95th percentile. Firm AUM has quadrupled off the trough: €1,150M (Q1 2020) → €2,309M (2024) → €3,450M+ (2025) → €4,674M (June 2026).

The reconstructed P&L for Internacional across the eight years is +€310.3M of gains against −€64.2M of losses, and the top of the list tells you exactly which thesis paid: Golar LNG +€57.1M, Danieli +€40.1M, Teva +€27.4M, Kosmos Energy +€16.7M and Bayer +€15.3M. Liquefied natural gas, an Italian steel-plant builder, a generics business nobody wanted, an African oil explorer, and a German chemicals company bought in the middle of its litigation crisis.

Behind them: BW Offshore +€12.8M, International Seaways +€13.5M, Academedia +€12.3M, Avio +€12.2M, Israel Chemicals +€12.2M, Babcock +€11.6M, CK Hutchison +€10.2M, Exmar +€9.5M and Cofide +€9.4M. Every one of them was uninvestable at the moment he bought it.

In March 2026 Cobas soft-closed Internacional and Selección — on reaching €3,000M combined. The reason given: room to keep trading small caps and less liquid assets without compromising returns.

The two-act structure: Bestinver proved the process works; Cobas put that process to an extreme stress test but emerged stronger.

What Does Cobas Own Today?

Cobas manages four funds, all on Alpha-File.com. These are semi-annual snapshots with an unnamed residual bucket (23–30% of each fund). The figures below are the December 2025 reports.

Cobas Internacional — €919M, the flagship

By sector: Energy & Materials 36.2%, Industrials 24.9%, Consumer 14.5%, Healthcare 9.3%, Real Estate 2.8%.

Parames talked about the oil and gas concentration: “oil and gas is our largest investment, with 10% in the oil exploration and 10% in services… we don’t have competition buying free cash flow at three times.”

The second-half 2025 activity is a deep-value manager doing exactly what he says he does — selling what worked, buying what hurts. Out: Kosmos Energy, Avio, Academedia, Befesa, Caltagirone, Energean, Panoro. In: Borr Drilling, SAF-Holland, PetroReconcavo, Meren Energy, Douglas, Ibstock. Trimmed: Currys, Teva, Danieli, Atalaya — four of the biggest winners.

Cobas Selección — €1.2B, the largest fund

Essentially Internacional plus an Iberian sleeve, and it shows: the same Golar/CK Hutchison/Danieli/BW spine, with Atalaya Mining 4.3% at the top of the named lines, plus Spanish positions in Meliá Hotels, Sacyr, Grifols and Almirall. 162 distinct issuers across the eight years — the widest net of the four.

Cobas Iberia — €81M, and the best number in the family

The smallest fund returned +52.3% in 2025, and its December book explains why it is a different animal: Meliá Hotels 9.1% (up 152% in six months), Sacyr 7.7% (doubled), Técnicas Reunidas 7.6% (held eight years, +37%), Grifols 7.1%, Almirall 7.0%, Gestamp 4.8%, Global Dominion 4.6%, Atalaya 4.5%, Miquel y Costas 4.3%, Tubacex 3.9% (new), Iberpapel 3.0%, Vocento 2.9% (new), Inmocemento 2.6%. Out in the half: Repsol and Semapa.

This is the portfolio that could not exist anywhere but Madrid — Spanish mid-caps, several of them family-controlled, most of them uncovered by anyone outside Iberia. It is also the clearest expression of the edge he claims and the one he disclaims in the same breath:

“In Europe — I mean, Europe, man, you know… you go to New York and you had a thousand guys that know more than you. And I see that when people ask me about the Spanish companies — US guys, very bright guys that know a lot about the US market — they don’t have a clue about the Spanish market. So, I don’t have a clue about the US market.”

What Happens Next

Paramés is the stubborn value investor, like the ones that refused to buy tech in the tech bubble only to have to close shop weeks before the bubble popped and value goes on its historic run. He survived the miserable 5+ years right after launching Cobas and now has the record to prove that his process works.

Don’t expect Paramés to start entering semiconductor or AI stocks. He’s someone that has proven that he is willing to stick to his discipline whatever the market brings. As the AI hype cycle continues, it will be interesting to see how his portfolios perform and if a similar tech-bubble dynamic plays out where value investors will be left for dead, see assets dwindle, only to see vindication once the bubble pops.

Paramés’s willingness to stick to his process is why Cobas earns its place. TCI showed concentration plus permanence. AKO showed a founder who engineered his own replaceability. Valley Forge showed a punch card with three holes left. Cobas is the only firm in this series where the process was stripped of every external support — the firm, the team, the record’s momentum, the clients’ faith — and had to prove itself twice, from scratch, in public.

I’ll be tracking every report on Alpha-File.com, where you can now pull the full CNMV holdings history for all four Cobas funds — position history, composition, cumulative P&L and every semi-annual letter — free. If you enjoyed this, subscribe: the next manager profile in the series is already in the works.

Capital One

Capital One got a fresh vote of confidence from the value crowd this quarter: Ariel Investments initiated a position while Third Point scaling up. Steadfast Capital added 22% and Junto Capital added 30%. The other side of the trade was heavyweight: Berkshire Hathaway cut its long-held stake by 58%, Viking Global slashed 76%, and Lone Pine, CAS Investment Partners and Fremen Capital all exited.

Capital One Financial operates as a digital-first bank and leading credit card issuer. The company’s moat is rooted in underwriting and cost advantages and superior data analytics. More recently, the moat has been bolstered by the network scale and payment integration unlocked by its acquisition of Discover. From a long-term buy-and-hold perspective, the bull case centers on the transformational Discover transaction driving network synergies (getting off V/MA rails), shifting the loan book toward higher-quality credit profiles, and scaling its upmarket premium travel ecosystem. Conversely, the bear case highlights macroeconomic cyclicality, subprime credit quality deterioration during recessions, and potential friction from merging the complex card networks. Capital One’s stock trades at a P/E of 11.9x with projected revenue growth over the next three years post-Discovery of MSD/HSD.

Sequoia Fund (Ruane, Cunniff & Goldfarb) — Year-End 2025 Letter (Alpha-File)
“Shares of Capital One Financial Corp (“Capital One”) returned 38% in 2025, capping a year in which revenues grew approximately 37% and earnings per share increased approximately 40%. […] Discover Financial Services also strengthens Capital One’s position in the low-to-mid prime credit segment, where the company has struggled to replicate the success that it has long had in subprime. Further, the transaction tilts Capital One’s revenue and profit base even more toward the card business, which earns higher returns than the company’s other lending activities.”

D.R. Horton

Housing is where the value managers went shopping. For DHI, Berkshire Hathaway initiated a (small) position alongside Duquesne Family Office, and Greenhaven Associates — the biggest homebuilder bull added again. Third Avenue, Burgundy Asset Management and Brave Warrior all trimmed.

D.R. Horton is the largest homebuilder by volume in the United States, primarily focusing on the construction and sale of single-family homes. Its economic moat is built on its unmatched national scale and a capital-light land strategy where it controls a vast portion of its lot supply through purchase contracts and developer options to maximize returns and mitigate balance sheet risks. The long-term bull case relies on sustained market share consolidation, the operational flexibility of its option-focused land strategy, and robust demand from first-time buyers who stand to benefit as housing affordability improves. The bear case is defined by margin compression driven by elevated sales incentives (such as mortgage rate buydowns), rising cancellation rates, and persistent affordability constraints that have pressured near-term volumes. The company trades at a P/E of 13x.

Heartland Mid Cap Value Fund — Q3 2025 Letter (Alpha-File)
“Our best-performing holding in the quarter, D.R. Horton (DHI), came from our Deep Value bucket. The largest homebuilder in the country, DHI enjoys around a 10% market share with scale advantages in a highly fragmented industry. The company has a particularly strong position in entry-level homes. To produce affordable housing, D.R. Horton runs the business with speculative inventory, meaning it builds homes before buyer contracts are signed. This allows the company to operate the business more like a manufacturer thereby reducing unit costs with most savings passed to the homebuyer.”

CRH

CRH keeps quietly accumulating fans. Hound Partners initiated a 6.0% position, while Third Point (+16%), Duquesne Family Office (+47%), Egerton Capital (+59%) and Durable Capital (+80%) all added. Only Gates Capital (−32%) and CDAM (−54%) went the other way.

CRH Plc is a global building materials leader that generates approximately 75% of its EBITDA in North America supplying primary materials such as cement, aggregates, and road solutions. Its economic moat is supported by a localized network of quarry and extractive properties that present high barriers to entry, alongside a vertically integrated “connected portfolio” strategy that enables collaborative downstream solution selling. The bull case rests on direct exposure to massive, long-term US infrastructure programs (such as IIJA, IRA, and the CHIPS Act), re-rating upside following its primary listing transition to the US, S&P 500 inclusion, and its highly active, value-accretive M&A engine. The primary bear case stems from potential slowdowns in global construction demand, elevated energy cost inflation impacting intensive cement operations, and strict government transition mandates on decarbonization. CRH trades at a P/E of 15.6x with consensus projecting revenue to expand from $39.67 billion in FY1 to $44.90 billion in FY3, and EPS to grow from $5.89 in FY1 to $7.35 in FY3.

YCG, LLC — Q1 2025 Investment Letter (aggregates initiation) (Alpha-File)
“Aggregates are raw materials such as gravel, crushed stone, and sand that are extracted from rock pits and quarries. They are mission-critical ingredients in concrete, asphalt, and other construction materials, and, as such, they are used in almost all commercial real estate, residential real estate, and public infrastructure projects. […] giving us what we believe was an attractive opportunity to diversify into the three leading North American aggregates businesses.”

Lennar

The second homebuilder on the list. For LEN, Berkshire Hathaway added 30% to its position — the same quarter it initiated DHI — while Greenhaven Associates (+6%) and Third Avenue (+5%) added. First Manhattan cut nearly half its position, and Eagle Capital and Brave Warrior trimmed at the margin.

Lennar Corporation (LEN) operates as a leading builder of quality homes, constructing affordable, move-up, and active adult residences primarily under the Lennar brand name. The company’s economic moat is anchored by its “Everything’s Included” approach, which maximizes purchasing power and simplifies homebuilding, alongside its completed asset-light transition that uses land banks and developer agreements to control 98% of its homesites. The long-term bull case centers on its transition to an asset-light model that reduces balance sheet risk, mitigates land-holding impairments, and generates significant free cash flow to accelerate returns to shareholders. On the other hand, the long-term bear case focuses on soft demand, falling net prices, and rising costs that create ongoing margin risks, as Lennar must heavily incentivize specs to reach its volume targets in an affordability-constrained consumer environment. Lennar currently trades at an estimated P/E of 15.5x.

GoodHaven Fund — Q2 2026 Semi-Annual Letter (Alpha-File)
“In addition, Builders was our biggest material addition of capital in the period and Lennar was our third biggest. […] A recent Presidential Economic Report - which we are not endorsing in aggregate - implies that we are domestically underbuilt by approximately 10 million homes. […] Clearly, we see much upside from recent levels for this sector in aggregate.”

New York Times

A quiet-compounder pick with a fresh sponsor: Giverny Capital initiated a 1.1% position in NYT this quarter and Berkshire Hathaway added 4%. Maverick Capital and Egerton Capital exited, while Burgundy, Abrams Bison (−6%) and Darsana trimmed.

The New York Times Company (NYT) is a global media organization focused on creating and distributing high-quality independent journalism, news, and information. The company’s economic moat lies in its direct audience relationships, trusted brand, and multiproduct bundle of news, games, cooking, newsletters, and sports via The Athletic, which drive deep user engagement, retention, and competitive resilience. The long-term bull case rests on the company’s strong pricing power as it successfully graduates bundle subscribers from promotional pricing to higher rates, alongside scaling high-margin digital advertising and licensing partnerships. Conversely, the long-term bear case is driven by slower subscriber adoption that limits the paid news market opportunity, accelerating declines in the legacy print business, and rising competition from consumer-facing artificial intelligence platforms that pressure top-of-funnel traffic. The New York Times currently trades at an estimated forward P/E of 23.6x.

Greenhaven Road Capital — Q4 2020 Letter (NYT initiation) (Alpha-File)
“While the top line quantitative data is underwhelming, the business is transitioning away from an advertising-supported to a subscription-supported model. In 2000, The New York Times’ revenue was made up of approximately 25% subscription and 75% advertising/other. Today those numbers are flipped, with approximately 75% of revenue earned from subscriptions.”

Union Pacific

The transcontinental merger has the value crowd picking sides. Viking Global initiated a 1.6% position in UNP, Third Point added a massive +250%, and Markel Group (+7%) and Park Presidio (+74%) added. On the other side, Baupost Group cut 23%, EdgePoint cut 40%, and First Manhattan, Gardner Russo & Quinn and Skye Global trimmed.

Union Pacific Corporation (UNP) operates a premier North American rail franchise encompassing 32,889 route miles in 23 states with 7 border crossings, transporting bulk, industrial, and premium freight. The company’s economic moat is built on limited and rational competition, high capital barriers to entry, captive rail lanes, and its unique geographic network that acts as an indispensable inland freight oligopoly. The long-term bull case focuses on operational turnarounds under CEO Jim Vena and the proposed Norfolk Southern acquisition, which would create the nation’s first transcontinental railroad to unlock vast network reach, single-line service efficiencies, and multi-year synergy-driven EPS growth. Conversely, the long-term bear case is dominated by merger-related execution risks, regulatory pushback and concession demands from the Surface Transportation Board (STB), and pricing power constraints from shippers or regulatory inquiries. Union Pacific trades at a forward P/E of 22x, with management projecting a high-single digit to low-double digit EPS growth CAGR through 2027.

Semper Augustus — 2025 Annual Letter (Alpha-File)
“Union Pacific and Norfolk Southern announced a merger, creating the first true transcontinental railroad in the United States. I never fathomed a merger like this one, and odds of STB and FTC regulatory approval are surely long. Only under the current administration would this merger be contemplated. In response to the announcement, in August BNSF and CSX announced an unprecedented coast-to-coast “partnership” with the same transcontinental benefits as the Union Pacific and Norfolk Southern deal but without the same regulatory hurdle.”

CarMax

KMX is the classic falling-knife value debate. Hawk Ridge Capital initiated a 4.9% position and Burgundy added 63%, but the long-time owners headed for the exits: Ruane, Cunniff & Goldfarb (Sequoia) exited in 2024, Giverny Capital sold out after 18 years (see below), Ariel Investments cut 38%, Starboard Value cut 48%, and First Manhattan trimmed 16%.

CarMax, Inc. (KMX) is a major used-car retailer engaged in the purchasing, selling, and financing of used vehicles through an omnichannel platform that integrates physical dealerships with digital channels. The company’s economic moat rests on its irreplaceable national physical footprint, beloved brand trust, robust digital platform, and its highly profitable in-house financing division, CarMax Auto Finance (CAF). The long-term bull case focuses on the company’s strategic turnaround under CEO Keith Barr, aimed at restoring a positive sales trajectory, enhancing reconditioning and logistics cost efficiencies, and achieving operating leverage on same-store sales growth. On the other hand, the long-term bear case highlights intense competition from digitally native disruptors like Carvana, a structurally higher cost burden from maintaining duplicate physical and digital operations, and consumer macroeconomic pressures that impact vehicle affordability and credit performance. CarMax currently trades at a forward P/E of 20.9x.

Giverny Capital — 2025 Annual Letter (sold the position) (Alpha-File)
“We first had two stocks that fell sharply in 2025: Carmax and Fiserv. Both represented approximately 4% of the portfolio at the beginning of the year and both declined by more than 50% during the year. Carmax was first acquired in 2007.”

Colliers

CIGI is a small-cap compounder the value list quietly owns. Spruce House added 9% this quarter, while Durable Capital exited and Turtle Creek (−24%), CDAM (−19%), BloombergSen and Burgundy trimmed.

Colliers International Group Inc. (CIGI) is a global diversified professional services and investment management company operating through Commercial Real Estate, Engineering, and Investment Management segments. The company’s economic moat lies in its top-tier commercial real estate services platform—the largest globally after CBRE and JLL—reinforced by high-touch, specialized client relationships and deep structural barriers in its Engineering segment, such as complex regulatory approvals and longstanding local authority connections. The long-term bull case centers on its robust cash flow compounding model (targeting 5% to 6% organic growth plus 10% from acquisitions for 15%+ earnings growth) and the secular tailwinds of infrastructure renewal, urbanization, and energy transition driving its resilient Engineering business. The bear case is defined by sensitivity to the commercial real estate transaction cycle, rising labor costs, and elevated near-term leverage of 2.5x post-acquisition. From a valuation perspective, CIGI trades at an attractive P/E multiple of approximately 13x, with expected earnings per share growth of over 15% in the near term and low-teens growth potential over the next several years.

Royce Small-Cap Opportunity — Q3 2024 Commentary (Alpha-File)
“Colliers International Group, a leading global professional services firm that operates in real estate leasing, advisory, outsourcing, capital markets, and investment management, along with growing activities in engineering services. Colliers has a high recurring revenue, asset light business model and serves large and highly fragmented markets.”

Microsoft

Yes, MSFT now shows up in a Value review. Pershing Square initiated an 11.9% position (its largest new buy of the quarter), joined by ShawSpring Partners (7.3%), BloombergSen (2.4%) and Maplelane Capital (2.5%), with Coatue (+18%) and Maverick Capital (+15%) adding. The exit column was just as crowded: Stockbridge Partners, Two Creeks, BlueSpruce, Ariel Investments, Whale Rock and Junto Capital all sold out, and Eagle Capital (−7%) and Sands Capital (−14%) trimmed.

Microsoft Corp. (MSFT) is a global technology leader that develops, licenses, and supports software, services, and devices across its Intelligent Cloud, Productivity and Business Processes, and More Personal Computing segments. The company’s massive economic moat is underpinned by its deeply entrenched enterprise customer base, significant economies of scale in its global datacenter infrastructure, and an early mover advantage in Generative AI that drives substantial customer switching costs. The long-term bull case is driven by capacity-led acceleration in Azure (which recently reached 43% constant-currency growth), rapid enterprise monetization of Microsoft 365 Copilot (surpassing 30 million paid seats), and a hybrid licensing model that layers high-margin, consumption-based billing on top of standard seat pricing. Conversely, the bear case is constrained by severe data center capacity limitations that restrict near-term demand capture, alongside massive capital expenditure requirements of approximately $175 billion for calendar year 2026 that pressure gross margins. Reflecting its exceptional earnings durability, Microsoft trades at a P/E multiple of 24.4x, with consensus long-term expectations forecasting a 14% revenue CAGR and 16% EPS CAGR through fiscal year 2029.

Pershing Square Holdings — Q2 2026 Semi-Annual Letter (new position) (Alpha-File)
“The company’s share price trades at one of its lowest earnings multiples of the past decade, reflecting investor concerns around AI disruption to the company’s core M365 productivity suite and the growth prospects of its Azure cloud business as well as skepticism about the company’s dramatically increased investment in compute infrastructure. We believe M365’s deep embeddedness in daily workflows, superior price-to-value proposition and proprietary Copilot AI agent make it far more resilient to AI disruption than point solution software.”

Brink’s

BCO is the kind of misunderstood physical-infrastructure story value managers love. Turtle Creek re-initiated and Simcoe Capital added 12%; only First Manhattan and Ariel Investments trimmed, both marginally.

Brink’s Co. (BCO) is a leading global provider of cash and valuables management, digital retail solutions, and ATM managed services. The company’s economic moat is built on its irreplaceable global physical logistics and security infrastructure, spanning 51 countries, which creates massive scale barriers and deep customer integration. The long-term bull case focuses on its multi-year business model transformation, which shifts the revenue mix toward higher-margin, technology-enabled ATM managed services and digital retail solutions (which grew 14% organically in Q2 2026), alongside the highly accretive acquisition of NCR Atleos that is expected to unlock $200 million in run-rate synergies and generate a combined ~$1 billion in free cash flow. The bear case is driven by significant integration and regulatory risks associated with the NCR Atleos deal, high debt levels of approximately $4.24 billion, and restricted financial flexibility due to strict debt covenants. Brink’s currently trades at a forward P/E multiple of 12.39, with its high-margin tech services poised to drive continued mid-to-high teens organic growth over the coming years.

SouthernSun Small Cap — Q3 2025 Commentary (Alpha-File)
“BCO, a leading global provider of cash and valuables management, digital retail solutions (DRS), and ATM managed services (AMS), was a top contributor for the quarter. In the second quarter, the company delivered 16% organic growth in AMS and DRS […] These higher-margin, recurring revenue businesses now represent over 25% of total company revenue and are expected to continue delivering double digit organic growth for the foreseeable future. Free cash flow continues to improve with over $100 million generated in the quarter on EBITDA growth, continued capital efficiency and strong working capital performance.”

UnitedHealth

Bridger Management initiated a 6.6% position in UNH and Steadfast, Brave Warrior and TenCore opened starter positions, with Tweedy, Browne adding 12%. Meanwhile Markel and Appaloosa exited, Hound Partners cut 32%, Longview Partners cut 25%, and First Manhattan (−16%) and Tiger Global (−15%) trimmed. Everyone agrees the earnings power is depressed — they disagree on how long it stays that way.

UnitedHealth Group Inc. (UNH) is a highly diversified healthcare enterprise operating through its UnitedHealthcare medical benefits segment and its Optum health services platform (encompassing care delivery, pharmacy services, and technology solutions). The company’s insurmountable economic moat is built on its leading scale across both health insurance and integrated care delivery, which allows it to drive superior cost management, capture extensive data, and leverage a highly scalable platform. The long-term buy-and-hold bull case focuses on the robust recovery of Optum Health’s operating margins (which expanded to 5.1% in Q2 2026), conservative pricing strategies that position its Medicare Advantage book ahead of cost trends, and durable cash flow supporting at least $5 billion in share repurchases. Conversely, the bear case is highlighted by near-term margin pressure within government programs, including negative Medicaid margins in 2026 due to high member acuity and insufficient state funding. Supported by strong long-term visibility, UNH currently trades at a P/E of 19.7x, with consensus forecasting a highly durable 13% to 16% annual adjusted EPS growth rate over the next three years.

Davis Funds — Q2 2026 Review (Global Fund) (Alpha-File)
“Within healthcare, we were very opportunistic throughout 2025 and the prior year, investing in a decisive way when shares of leading managed care insurers stumbled as operating costs surged rather unexpectedly for the group. […] Seemingly small margin improvements on low profit margins can translate into very large percentage increases in structural earnings power. We believe the market’s valuations for these portfolio companies underestimated the potential for a rebound in their bottom-line economics. UnitedHealth is a representative investment in this category.”

Progressive

Best-in-class gets cheap occasionally. Longview Partners initiated a 4.3% position in PGR, Ariel Investments effectively rebuilt its position (+5,732%), and Fairholme (+9%) and Giverny Capital (+3%) added. Viking Global exited and First Manhattan trimmed 37%.

The Progressive Corp. (PGR) is a leading provider of personal auto, commercial auto, and specialty property and casualty insurance products. The company’s formidable economic moat is defined by its best-in-class auto underwriting prowess, disciplined technology-driven pricing, and a flexible multi-channel distribution network that consistently yields superior combined ratios and underwriting profitability compared to peers. The long-term buy-and-hold bull case centers on its structural growth advantages over competitors, aggressive capital deployment to capture market share during soft markets, and sustained underwriting outperformance relative to personal auto peers. Conversely, the bear case is highlighted by a slowing personal auto policy-in-force (PIF) growth trend, potential inflationary headwinds on loss costs, and near-term EPS compression as exceptionally high underwriting margins normalize. Reflecting these dynamics, Progressive trades at an NTM P/E multiple of 12.6x, with expected earnings projected to compress slightly as the auto insurance cycle softens.

Middle Coast Investing — Q1 2026 Letter (Alpha-File)
“I think Progressive is a good bet despite that. The management team seems long-term oriented, even as the company publishes monthly results. Those results show incredible profitability since Progressive raised rates to deal with used car price inflation in 2022. Through February, Progressive has reaccelerated PIF growth and is still improving profitability (if barely). The stock trades at 10x earnings, which is its lowest since 2020.”

Apple

AAPL cuts both ways on the watchlist. BlueSpruce opened a small starter and Sands Capital added aggressively (+1,317%), but Viking Global, Steadfast, Junto Capital and Skye Global all exited, ICONIQ Capital halved its position (−54%), GFI Investment Counsel cut 42%, and Gardner Russo & Quinn, Giverny Capital and H&H International trimmed. Notably, Berkshire Hathaway — Apple’s most famous holder — did not touch the position.

Apple Inc. designs, manufactures, and markets consumer electronics, software, and digital services globally, with its flagship products including the iPhone, Mac, iPad, Apple Watch, and AirPods, alongside an expansive Services platform that monetizes its massive installed base of over 2.5 billion active devices. The company’s powerful economic moat is built on high customer switching costs and intense brand loyalty, which seamlessly integrate its hardware, software, and services to lock in consumers. Long-term bulls point to an accelerating iPhone replacement cycle driven by the global rollout of Apple Intelligence and foldable devices, alongside highly profitable Services growth. Conversely, the bear case centers on rising component prices—specifically unprecedented memory cost inflation—app-layer regulatory challenges, and potential demand elasticity from hardware price hikes. From a valuation and growth standpoint, Apple trades at a trailing P/E multiple of 33.7x, and is projected to deliver revenue growth of 14.6% in FY26, 13.9% in FY27, and 5.5% in FY28, alongside earnings growth of 18.4%, 13.1%, and 8.3%.

Wedgewood Partners — Q2 2026 Client Letter (Alpha-File)
“Apple was also a top contributor to performance during the quarter. Revenues grew 17%, driven by 22% growth in iPhone and 16% growth in services. The iPhone 17 family has catalyzed a solid upgrade cycle ahead of what we expect to be another strong launch later this year, featuring a new foldable form factor. […] As one of the largest single purchasers of DRAM, the Company has strong negotiating leverage, but it can also implement hardware and software innovations to reduce its dependence on memory.”

Simpson Manufacturing

A sleepy structural-connectors monopoly with one loud dissenter: Ariel Investments added 9% to SSD, while ValueAct cut its position 41% and Burgundy trimmed at the margin.

Simpson Manufacturing Co., Inc. is a leading manufacturer of engineered structural connectors and building solutions, primarily operating through its flagship Simpson Strong-Tie brand to design wood and concrete construction products for residential and commercial markets. The company’s durable economic moat is anchored by decades of product innovation, high brand recognition, and extensive field support, with its connectors frequently specified directly into building codes and delivered to distributors within 24 to 48 hours. Long-term bulls are encouraged by Simpson’s track record of outperforming the broader market relative to U.S. housing starts, its strategic expansion into mass timber and digital software solutions, and its highly efficient cash conversion cycle. Conversely, the bear case is centered on the company’s heavy exposure to the cyclical residential housing market, potential margin compression from labor and material cost inflation, and recent pricing pressures from fastener competitors in undifferentiated product categories. Simpson trades at a P/E of 22.6x, and the company has historically compounded its earnings per share at a 19% CAGR over the last 10 years.

Conestoga Capital Advisors — Q4 2023 Commentary (Alpha-File)
“Simpson Manufacturing Co. Inc. (SSD), a maker of structural connector systems, was the leader in the sector. We believe SSD’s management team has guided the company very well through a challenging home construction environment.”

Amrize

The fresh Holcim spin has already split the room. D1 Capital initiated a small position in AMRZ and Permian Investment Partners (+38%) and First Sabrepoint (+6%) added, while First Manhattan (−20%), Gates Capital (−35%) and Alpha Wave (−23%) reduced.

Amrize Ltd. is a building solutions company focused on the North American market, offering customers a comprehensive range of building products from foundation to rooftop, including cement, aggregates, ready-mix concrete, asphalt, and commercial roofing systems. Spun out from Holcim in June 2025, the company possesses a strong economic moat driven by highly localized market dynamics and high barriers to entry, such as steep capital costs and permitting timelines of up to 15 years for new cement plants, complemented by an efficient distribution network where over 60% of its volume is shipped via barge and rail. The long-term bull case is bolstered by structural demand tailwinds from federal infrastructure investments, industrial onshoring, and resilient, non-discretionary re-roofing spend, along with growth from high-synergy aggregates bolt-on acquisitions. On the other hand, the bear case centers on severe margin pressure from oil-price-driven freight and raw material inflation, cyclical economic downturns, and near-term execution and credibility risks as an independent SEC filer.

Bestinver — Q3 2025 Quarterly Letter (Alpha-File)
“[Amrize] is the undisputed leader in cement in the US and Canada, with a dominant position in local markets and a production capacity double that of the country’s second-largest operator. […] One of the most important aspects for us, however, is who runs it.”

WEX

WEX is a small-cap payments orphan the value managers are quietly accumulating: Turtle Creek added 9% and Hawk Ridge added 50%, with only marginal trims from Ariel Investments and Sunriver.

WEX Inc. operates a global commerce platform that delivers seamlessly embedded, personalized payment and technology solutions across three segments: Mobility, Benefits, and Corporate Payments. The company’s robust economic moat is sustained by its proprietary closed-loop fleet network covering approximately 95% of U.S. fueling locations, alongside cost-efficient funding advantages provided by its banking subsidiary, WEX Bank. Long-term bulls are attracted to WEX’s secular tailwinds in B2B payment capture, durable SaaS account expansion in health benefits, and high-conviction capital return strategies via aggressive share buybacks. Conversely, the bear case is pressured by structural threats from electric vehicle transition risks to fleet economics, transaction volatility from fuel price fluctuations, a highly competitive B2B market, and recent market anxiety regarding potential artificial intelligence disruption to financial service platforms. Currently, WEX trades at a forward P/E multiple of 14.80x, with estimates of 4% revenue CAGR and an 11% adjusted EPS CAGR over the next 3 years.

Norwegian Cruise Line

The deep-value leisure trade. Baupost Group more than doubled its NCLH position (+113%), with Ariel Investments (+5%) and Hound Partners (+17%) adding — and, unusually for this list, zero trims or exits on the watchlist.

Norwegian Cruise Line Holdings Ltd. (NCLH) operates as a leading global cruise line company across three distinct brands—Norwegian Cruise Line, Oceania Cruises, and Regent Seven Seas Cruises—serving approximately 700 destinations worldwide. The company’s economic moat is supported by efficient scale and massive capital barriers to entry, as the cruise industry is highly capital intensive, requiring long-lead ship construction contracts and specialized shipyard availability to build and deploy competitive vessels. From a long-term standpoint, the bull case is centered on a management turnaround strategy featuring a refreshed board of directors and new senior commercial leaders who are optimizing revenue management, executing shoreside cost-cutting initiatives, and leveraging private island expansions like the grand opening of the Great Tides Waterpark at Great Stirrup Cay to accelerate yield growth. Conversely, the bear case focuses on NCLH’s elevated net leverage ratio, which is projected to end fiscal 2026 above 6x, a substantial debt burden, and sensitivity to macroeconomic downturns or fuel price spikes that could disrupt consumer travel spending and strain liquidity. The stock trades at a P/E of 16.0x, and the company expects its near-term capacity growth to moderate to a 2.5% CAGR from 2026 to 2029 as capital expenditure levels decline.

Patient Capital Management — Q2 2025 Opportunity Equity Letter (Alpha-File)
“NCLH got hit hard during the selloff this year due to concerns about cyclicality and debt. We believe NCLH can grow earnings power double digits for the next decade or more. The other large cruise lines have already made it back to post-pandemic record prices, we believe NCLH will too.”

Kenvue

Merger-arb meets deep value. Tweedy, Browne added 78% to its KVUE position and Maverick Capital added 89% ahead of the Kimberly-Clark deal close, while Skye Global cut 42% and Burgundy trimmed.

Kenvue Inc. (KVUE) is the world’s largest pure-play consumer health company by revenue, operating through three segments—Self Care, Skin Health and Beauty, and Essential Health—with a portfolio of globally trusted brands such as Tylenol, Listerine, Neutrogena, and BAND-AID. Its economic moat is anchored by its powerful portfolio of science-backed, doctor-recommended consumer brands that command premium pricing, high consumer loyalty, and dominant retail shelf space. From the perspective of a 5+ year investor, the primary bull case is the pending value-creating merger with Kimberly-Clark Corporation (expected to close in the fourth quarter of 2026), which will combine world-class portfolios to create a global health and wellness leader and is projected to deliver approximately $1.9 billion in cost synergies and $500 million in revenue synergies within the first few years. The bear case rests on persistent litigation headwinds—most notably product liability claims surrounding acetaminophen (Tylenol)—as well as severe retail competition in categories like oral care, market share losses in skincare that require heavy brand reinvestment, and an estimated $80 million annualized drag from tariffs. Kenvue trades at a next-twelve-months (NTM) P/E multiple of 17.28, and while the company is not providing forward guidance due to the pending merger, Kimberly-Clark targets a long-run constant currency net sales growth rate of approximately 3% for the combined entity.

TD SYNNEX

The AI supply chain’s cheapest expression. First Manhattan and Maverick Capital opened token starter positions in SNX, while Brave Warrior cut its position by more than half (−53%), Abrams Bison cut 19%, and Burgundy trimmed.

TD SYNNEX Corporation (SNX) is the largest IT distributor in the world, playing a critical role in the technology supply chain by aggregating and distributing hardware, software, and systems, while also designing and delivering cloud and AI infrastructure solutions through its fast-growing Hyve Solutions subsidiary. The company’s economic moat is defined by its massive global scale, specialized go-to-market enablement capabilities, and deep integrations with over 2,500 vendor partners that make it an indispensable intermediary within an increasingly complex IT ecosystem. From a long-term investment horizon, the bull case is centered on accelerating cloud capital expenditure and enterprise datacenter modernization, which are driving massive demand for Hyve’s accelerated compute and storage solutions, supported by a 1-million-square-foot manufacturing capacity expansion to accommodate new hyperscaler client wins. Conversely, the bear case centers on the structurally low-margin profile inherent to IT distribution, short-term working capital swings that consume cash during growth ramps, and the risk of PC demand elasticity or component inflation pinching margins. TD SYNNEX trades at a current P/E multiple of 16.17, with broker estimates projecting revenue and non-GAAP gross billings to compound at a solid 6% annual rate over the next three years in a base case scenario.

Middle Coast Investing — Q1 2024 Letter (Alpha-File)
“TD Synnex is a technology distributor. Classically, that means it buys computers and phones from Apple and Hewlett Packard and Dell and so on, and then re-sells them to small businesses and other companies. SNX is focused on adding more software customization and value-added IT support to its customers, as a way to improve margins.”

SiriusXM

A Berkshire-blessed melting ice cube priced for permafrost. Tweedy, Browne added 42% to its SIRI position and Gates Capital added 66%, while Maverick Capital exited a token position and First Manhattan trimmed 25%.

Sirius XM Holdings Inc. (SIRI) is the leading audio entertainment company in North America, operating the proprietary SiriusXM subscription satellite radio network and the ad-supported and premium Pandora streaming platforms. The company’s economic moat is anchored by its unique contiguous FCC spectrum licenses, its dominant position as the preeminent factory-installed in-vehicle audio service, and long-term exclusive talent relationships that create a highly sticky listener base. The long-term bull case centers on its expanding digital advertising and off-platform monetization strategy—including premium podcasting networks and an exclusive ad representative partnership with YouTube Audio—alongside record-low self-pay churn of 1.4% and the potential monetization of its unvalued excess spectrum assets, which are estimated to have a present value of $2 billion to $5 billion. The bear case centers on persistent underlying self-paid subscriber losses, high dependency on highly cyclical auto industry sales, rising memory costs that reduce hardware profitability, and intensifying competition from digital music streaming services. As of August 18, 2026, SiriusXM trades at a P/E of 9.2x, and while subscription revenue is expected to remain relatively flat as ARPU gains offset average subscriber declines, broker forecasts project a steady EPS expansion over the next three years.

Kraft Heinz

The deepest of deep value. Fairfax Financial added 172% to its KHC position and Fairholme added 92%, with Maverick Capital opening a token position — and no one on the watchlist selling.

The Kraft Heinz Company (KHC) is one of the largest food and beverage manufacturers in the world, producing and marketing a massive portfolio of iconic brands like Heinz, Kraft, and Philadelphia across platform categories such as condiments, cheese, dairy, and meals. The company’s economic moat rests on its vast scale and immense brand saliency, with a portfolio of eight brands that each generate over $1 billion in annual retail sales, allowing it to maintain a presence in nearly every household. From the perspective of a 5+ year investor, the bull case centers on management’s decision to pause the planned business separation and invest $700 million in 2026 into marketing, product innovation (like lactose-free Philadelphia and kids’ beverages), and price competitiveness to drive a return to sustainable, volume-led growth, while expanding rapidly in Emerging Markets. Conversely, the bear case emphasizes structural category headwind vulnerabilities, as Kraft Heinz has high exposure to price-sensitive center-store categories where private label brands are rapidly gaining volume share and consumer trade-down behavior continues to depress volumes. KHC trades at a P/E of 14.1x, and while management targets a long-term return to volume-led growth in 2027 and beyond, near-term organic net sales are projected to decline between 0.5% and 2% in fiscal 2026.

Southeastern Asset Management — Q4 2025 Letter (Alpha-File)
“While Kraft Heinz was a detractor this year, we were thrilled when Steve Cahillane, a great CEO partner from our successful Kellogg’s investment, was announced as Kraft’s CEO at the end of the year.”

Chevron

The lone energy major on the list. Discovery Capital initiated a 2.4% position in CVX and First Manhattan added 8%, while Ariel Investments — whose Q1 letter praised the position — trimmed 17% into strength.

Chevron Corporation (CVX) is a major U.S.-based integrated energy company with global operations spanning upstream exploration and production of crude oil and natural gas, as well as downstream refining, marketing, transportation, and chemical manufacturing. From the perspective of a long-term buy-and-hold investor with a 5+ year horizon, Chevron’s economic moat is built on its premier, low-cost global asset base, high-return integration across the value chain, and robust capital discipline that prioritizes free cash flow efficiency over volume growth, further reinforced by its successful integration of Hess Corporation. The long-term bull case is underpinned by high-margin production expansion in premier growth areas like the Permian Basin, Guyana, and Argentina, alongside the realization of substantial structural cost synergies and a novel, repeatable expansion into behind-the-meter clean energy solutions, highlighted by a massive 20-year power purchase agreement with Microsoft for Project Kilby. Conversely, the bear case centers on the company’s structural exposure to volatile global commodity prices, execution risks in complex international mega-projects, and geopolitical vulnerabilities, such as recent loading disruptions at the Caspian Pipeline Consortium (CPC) terminal. Financially, Chevron trades at a P/E of 13.8xwhile targeting over 10% average annual growth in both EPS and free cash flow through 2030.

Ariel Investments — Q1 2026 Commentary (Alpha-File)
“Chevron’s +37.1% first quarter return was similarly boosted by elevated oil prices brought by the recent conflict. Chevron also benefited from shifting political leadership in Venezuela. The company has operated in Venezuela since 1923. When Hugo Chavez nationalized the last privately run oil fields in the country in 2007, Chevron was one of the few U.S. majors to continue to operate as a minority investor.”

The Value Managers

The cohort behind this post — the 40 manager profiles and 8 tagged funds that make up alpha-file’s “Value Managers” list:

Manager profiles

• Abrams Bison Investments

• Appaloosa

• Ariel Investments

• Baupost Group

• Berkshire Hathaway

• Brave Warrior Advisors

• Burgundy Asset Management

• CAS Investment Partners

• CDAM (UK)

• Dalal Street

• Discovery Capital Management

• Fairfax Financial Holdings

• Fairholme Capital Management

• First Sabrepoint Capital

• Gates Capital Management

• Great Hill Capital

• Greenhaven Associates

• Guardian Point Capital

• Hawk Ridge Capital

• Himalaya Capital

• Hound Partners

• Lansdowne Partners

• Luxor Capital Group

• MFN Partners

• P2 Capital Partners

• Packer & Co

• Permian Investment Partners

• Praetorian PR

• Prescott General Partners

• Route One Investment Company

• Sachem Head Capital

• Simcoe Capital Management

• Sprucegrove Investment Management

• Starboard Value

• Third Avenue Management

• Third Point

• Tiger Pacific Capital

• Trian Fund Management

• Turtle Creek Asset Management

• Tweedy, Browne

Tagged fund profiles

• Fidelity Low-Priced Stock Fund

• First Eagle Global Fund

• FPA Crescent Fund

• Longleaf Partners Fund

• Longleaf Partners Global Fund

• Longleaf Partners Small Cap Fund

• Oakmark Fund

• Oakmark Global Fund

A note on the data: All portfolio data and charts in this piece are sourced from Alpha-File. The same position-history, trading, composition, industry, P&L, and fundamentals exhibits are available for every tracked manager on Alpha-File, including 13F filers, US N-PORT mutual funds and ETFs, and supported European, Canadian, and other global funds. Public-company and manager names throughout link directly to their specific Alpha-File pages.

Background

How concentrated should you be? TCI runs 10–15. WindAcre has put 48% of a fund into a single stock. Valley Forge owns seven names and has made seventeen investments in nineteen years. The idea that good investments are rare, so if you find one, why diversify?

Glenn Greenberg has been running a ten-to-twelve-stock book since 1984, first out of the firm he co-founded with John Shapiro — Chieftain Capital Management — and now with his firm Brave Warrior. He runs his portfolio while blocking out the entire apparatus of the industry: no marketing, no broker calls, no tracking positions, no sell-side analysts feeding him ideas.

He also had his concentration doctrine tested, almost to destruction: in 2013 he had 29.4% of the portfolio in Valeant Pharmaceuticals.

The Investor: an English major who told the brokers not to call

Glenn Greenberg was born in New York in 1947, the son of Hank Greenberg — Detroit Tigers first baseman, two-time MVP, the first Jewish superstar in American sport, and a man who lost four seasons to military service. Glenn’s early career had far less superstardom relative to his father’s. His account of his own path:

“I was an English major in college and taught school during the Vietnam War era. I went to business school at the recommendation of one of my bosses. When I was at business school, the only thing that appealed to me was a portion of one class that dealt with security analysis. It seemed like a lot of fun to study a company and figure out whether its stock would be a good investment. I didn’t understand what investment bankers did. I didn’t think I’d be interested in being a management consultant, which attracted all the best and brightest. So I took a job at Morgan Guaranty, now JP Morgan Chase, in their trust and investments department.” — Graham & Doddsville, Spring 2010

Five years at Morgan Guaranty, then five more as the analyst for a single successful private investor — “I spent five years crunching his numbers and analyzing his investments” — and then, in 1984, Chieftain. He is clear that the pre-spreadsheet training was not a handicap:

“One thing that’s changed a lot is that we were much more qualitative, we didn’t have PCs to do these fabulous, complex multi-variable models, but I don’t think we lost anything by not having that available.” — Graham & Doddsville, Spring 2010

The founding rules:

“When we founded Chieftain, I felt that since my partner and I were going to be doing everything, all the research and picking stocks, we wanted to spend 100% of our time on that. So we made up some rules. Number one, we didn’t want brokers to call us with their ideas. We told them we would call when we wanted information. We decided not to market. We did accept accounts over the years, but we didn’t spend any time marketing. It was people who selected us who we thought would be good clients. Often they were CEOs of companies we met through our research.” — Graham & Doddsville, Spring 2010

Read that against every other firm in this series. AKO built a research department of forensic accountants and trained interrogators. Whale Rock does 2,500 management meetings a year. Greenberg’s answer to the same problem: just two people in a room researching — the edge was time and he bought it by refusing everything that consumes it.

Chieftain’s 1984–2000 stretch is reported at 25% compounded before fees against 16% for the S&P; the full 1984–2009 run is put at roughly 18% a year against about 12% for the market, with assets compounding from $43 million to $3 billion.

In 2009, after twenty-five years, Greenberg and Shapiro separated. The accounts describe a personality conflict rather than a philosophical one. Shapiro kept the Chieftain name; Greenberg took the clients who came with him and renamed his operation Brave Warrior.

Post-split, Brave Warrior is reported at around 16% annually from 2009 to 2016 with no more recent figures than that. What is not in dispute is a quarter-century of double-digit outperformance.

Philosophy & Process

1. The 5% floor

Greenberg will not own a position he is unwilling to size at 5% of the fund. A 1% holding contributes almost nothing to the result while consuming the same research and maintenance as a large one. The consequence is arithmetic. If the minimum is 5%, the maximum number of ideas is twenty, and in practice he has run ten to twelve, with individual names frequently above 20%. If you can’t justify 5%, you have not done enough work to justify anything.

2. The valuation fits on a legal pad

I’d describe Brave Warrior as Value Investing 2.0. Not balance sheet focused, not looking for net-nets, but looking for a double digit FCF yield with some optionality on top typically manifesting as a high free-cash-flow yield, a conservative growth estimate, which hurdles to mid-teens on a high probability judgment of what the business looks like in three to five years. Quotes— mostly from his Columbia Business School lectures:

“There are these really simple things that you don’t have to be a genius to figure out.”

“Many companies are what I refer to as geodes — ordinary rock outside, beautiful crystals within.”

“If you are going to be an investor, you have to do the numbers yourself.”

He calls the target a “two-inch putt” — high return, low risk, no cleverness required. That is a very different ambition from Whale Rock’s exponential S-curves. It is also, on the evidence of the Valeant position, harder to hold to than it sounds.

3. The edge is duration

Asked in 2010 whether the modern market had eliminated the informational advantage, his answer conceded the point then reframed it:

“I think it’s a harder business… regulation FD makes it more difficult to get insights into businesses by meeting with management and gleaning information before the crowd does. Basically, these companies undress themselves four times a year, plus every conference they attend. People are focusing on every little number and change in business direction and it’s harder to get a differentiated view… There’s not that much that you’re going to think of that no one has thought to ask.”

“I would say our edge is the willingness to take a longer view of a business. Maybe the willingness to take advantage of somebody getting jittery about something that’s a short-term [issue].” — Graham & Doddsville, Spring 2010

Greenberg is not claiming to see further. He is claiming only to be willing to wait, and to have arranged his firm so that nobody can make him stop.

Analysts model cost-per-click next quarter and EPS this year. Almost nobody underwrites the quality of a business over five years — which is why, in his account, mispricing survives even among the largest and most-covered companies in America.

4. Macro...kind of

From his rare 2017 television appearance:

“We made a big bet that normal interest rates would not stay at zero. It was that simple.” — CNBC, March 2017

That bet — JPMorgan, Primerica, Schwab, Capital One — is still visible in the book nine years later.

Valeant

By the end of 2013, Valeant was 29.4% of Brave Warrior’s portfolio — nearly a third of the fund in a serial acquirer of drug assets, financed with debt, run by a management team promising that R&D was a waste of shareholders’ money.

It got bigger than that. The filings put the position at 36.6% of the disclosed book on 30 June 2015 — $1.27 billion of a $3.46 billion portfolio — which is the true peak, and more than a third of the fund in one levered pharmaceutical roll-up.

Then came the congressional hearings, the price-gouging headlines and the collapse. Alpha-File’s reconstruction puts the outcome at −$297.4 million at 0.61× over 3 years — the largest realized loss in the fund’s disclosed history.

Asked about it on CNBC in 2017, he named it his biggest career mistake without qualification.

The orange band is Valeant — roughly $1.4 billion of a $3.5 billion book at its 2015 peak. Note what replaces it: no single position since has come close.

And the second scar is almost as big. Antero Resources: −$278 million at 0.15× MOIC, held for under a year. A natural-gas producer, bought big, and 85% of the money gone almost immediately, although sold for potentially better opportunities near the COVID lows.

What Does Brave Warrior Own Today?

The Q2 2026 filing shows $4.6 billion across 40 disclosed positions: eighteen positions of $40 million or more account for 99.5% of the money. The remaining twenty-two lines account for $22.8 million and are likely a start-up portfolio for an analyst. Turnover runs 36.7%, an implied hold of about 2.7 years, and Brave Warrior has made 109 investments since the 13F record begins in 2013.

The mini-portfolio looks like a Glenn Greenberg portfolio but with a slightly higher quality-tilt with shades of TCI. Most are a few hundred thousand dollars — Visa, Meta, Moody’s, UnitedHealth, Texas Instruments, Analog Devices, Marriott, Progressive, Mastercard, Amazon, Microsoft, Comcast, Bank of America, Applied Materials.

ICON plc — 11.9%, and the newest idea

ICON at 11.9% ($548M) was initiated in the March quarter and added to in June, and it is already up +$189M at 1.5× on a cost basis of $113.8 against $173.7.

ICON is a contract research organization (CRO) — a provider of outsourced clinical-trials to the biopharma industry—the sector de-rated brutally as biotech funding dried up and trials started slowing, and ICON was then hit with accounting irregularities in Q1 2026 which have since been cleared up.

At a 10%+ FCF Yield, this looks like a prototypical investment: relatively simple business to understand 3-years out dislocated due to medium-term industry issues and a temporary accounting issue, since resolved.

OneMain, SLM, Capital One, Primerica, FNF, Arthur J Gallagher (~37%)

OneMain Holdings is the second-largest position at 10.1% ($462M), held four years, +$170M at 1.5× — a subprime personal lender. SLM at 7.6% is the student lender and one of three positions currently underwater, at −$39.7M, 0.9×. Capital One sits at 4.3%. And then Primerica at 5.4% has been held thirteen years — for +$529.4 million at 2.26×. Compounded for thirteen years without interruption. Alongside them sits F&G Annuities & Life at 1.4% ($66.0M), trimmed this quarter.

New this quarter: Arthur J Gallagher at 6.4% ($295M), straight in at the fifth-largest position — insurance broking, a business with no underwriting risk, recurring commissions and pricing that ratchets with premiums, a bread and butter business sold off due to fears AI will replace insurance brokers but like the broader theme, a beneficiary of higher inflation and interest rates.

The housing complex: Lennar, DR Horton, Builders FirstSource, Millrose (~15.6%)

Lennar 4.7%, Builders FirstSource 4.6%, Millrose Properties 3.9% and DR Horton 2.4% — and Millrose is Lennar’s own land-banking spin-off. DR Horton has been the winner (+$163M, held four years); Builders FirstSource has been the drag (−$67M at 0.76×, and it was trimmed). It actually feels like the opposite macro bet to the financials: a beneficiary of lower rather than higher rates. Perhaps an intentional hedge or perhaps cheap enough to offset any macro headwinds.

The rest: TD Synnex, Elevance, AutoNation, Ryanair, MPLX

TD Synnex was the largest position in March; the share count was cut 53% in June, taking it to 8.4% — after a +$507.5 million gain at 2.07× in two years and two months, the second-largest gain in the fund’s history. An IT distributor, bought as a low-multiple logistics business, sold down likely due to AI fueled growth in one of its subsidiaries which, despite the markets excitement, may be harder to underwrite over 3-years.

Elevance Health at 8.2% has been held nearly seven years and trimmed 13%. AutoNation 6.2%, Ryanair 5.6% — added 30% this quarter, the only meaningful add besides Gallagher, and the rare non-US name — and MPLX 3.9%, a midstream partnership. Out entirely: Ares Management and First American Financial.

The Ghosts

The reconstruction (June 2013 to June 2026 — the full disclosed record) makes the shape unusually legible, because there are so few positions.

The wins are patient and financial: Primerica +$529.4M (2.26×, 13.0y) · TD Synnex +$507.5M (2.07×, 2.2y) · Apollo Global +$431.3M · HCA Healthcare +$394.9M (1.92×, 6.2y) · JPMorgan +$427.7M (11.7y), and behind them Charles Schwab +$271.6M and Discover +$249.9M. Two of the top five were held for more than a decade; none of them required a technology view.

The losses are rarer but include two standouts: Valeant −$297.4M (0.61×) · Antero Resources −$278.0M (0.15×, under a year).

If I were to summarize the pattern: Greenberg makes money on unglamourous businesses trading at cheap valuations. Where it has run into trouble, the concentrated book has meant relatively large losses on a handful of positions but over 13 years, there are really only 2 mistakes of note.

What Remains Unknown

What we we can glean, Brave Warrior has a phenomenal record assembled from meat and potato businesses bought at high conviction weights. After the Valeant debacle, we’ve seen concentration come down, relatively speaking (no more 20%+ positions). A lot remains unknown as there are few interviews, podcasts episodes, or letters available to the public. e

I’ll be tracking every filing on Alpha-File.com, where you can pull this same view — position history, initiations and exits, fundamentals, and cumulative P&L — for Brave Warrior or any manager, free. If you enjoyed this, subscribe: the next manager profile in the series is already in the works

It’s also worth noting that alpha-file.com collects data outside of 13F’s including US mutual fund data as well as individual mutual funds all over the world. Don’t see a fund you want? Request the feature and we’ll try and get it done.

Now that 13F season is over, I wanted to review some of the top buys from alpha-files list of “Quality Managers”. Investing style is a matter of debate and the best managers can own a 4x P/E stock and a 40x P/E stock at the same time (think Appaloosa).

With that said, this list is generally comprised of managers focused on high-returning, wide-moat businesses run by capable management.

The table below is a look at what the super-investor portfolio looks like. The list rank weighs the number of managers along with their median weight. It’s interesting to see the blend of initiations, exits, adds and trims and see which stocks are true battleground stocks versus consensus longs or shorts.

This is the same table but with a fundamental overlay.

In subsequent posts, I’ll discuss what some of the top portfolio holdings of “Value Managers” looks like as well as the quirkiest and most interesting ideas that surfaced from 13F season.

Amazon ($AMZN)

Amazon is a major position for several superinvestors and when a diverse group of investors spanning Dragoneer Investment Group, Bill Ackman’s Pershing Square, and Stan Druckenmiller’s Duquesne Family Office all initiate or add, you pay attention. For AMZN, Dragoneer and Pershing each established large new positions (10.7% and 10.5% of portfolio), while Duquesne increased its stake by more than tenfold. The other side of Amazon was more selective: ICONIQ Capital and D1 Capital Partners made deep trims, while Peconic Partners exited.

The key driver of Amazon will clearly be AWS but its worth spending some time on the e-commerce business. After admitting to wildly overbuilding during the COVID fueled e-comm mania, Amazon is now in a spot to fill-out excess capacity and has even launched Amazon Supply Chain Services, an effort to expand beyond to full-service logistics to bigger businesses like P&G and American Eagle. However, e-commerce and advertising are just 40% of EBIT and far lower portion of future growth. The prize at the end of the day is AWS.

AWS has several advantages in the AI race. First, relative to Microsoft and Alphabet, AWS is the most “neutral” inference cloud with the least amount of competing solutions. If inference is like a fab, AWS is the closest to TSMC whereas GCP is arguably Intel. Second, AWS looks to, per Gavin Baker’s own words, have one of the best custom ASICs teams with Graviton and Tranium both looking like real differentiators.

Invest Like the Best, EP.451 — Gavin Baker - Nvidia v. Google, Scaling Laws, and the Economics of AI (Dec 9, 2025).

From my perspective, the best ASIC team at any semiconductor company is actually the Amazon ASIC team. They were the first one to make the Gravitron CPU. They have this Nitro, it was called SuperNIC. They’ve been extremely innovative, really clever, and like Trainium and Inferentia 1, maybe they’re a little better than the TPU v1, but only a little. Trainium2, you get a little better. Trainium3 is I think the first time it’s like, “Okay.” And then I think Trainium4 will probably be good.<br><br>I will be surprised if there are a lot of ASICs other than Trainium and TPU. And, by the way, Trainium and TPU will both run on customer-owned tooling at some point. We can debate when that will happen, but the economics of success that I just described mean it’s inevitable. Like, no matter what the companies say, just the economics make it, and Reasoning from first principles make it absolutely inevitable

While there was a freak-out about hyperscaler capex, AWS has been through heavy capex cycles before when cloud first emerged and Andy Jassy explicitly addressed the AI ROIC question on their call and spelled out ambitions to make AWS a $1tn revenue business.

Earlier this year, we said we plan to invest approximately $200 billion in cash CapEx in 2026, the majority of which is to support AI and AWS. At this level of spend and higher, we have clear line of sight to strong financial returns. I’ll explain why. There are two major parts of the investment: the data centers and the servers and networking equipment that go into them. These have different capital cycles. Data-center capital is spent starting two years before we can put servers into them to start monetizing. Once a data center opens with servers plugged in, we start generating significant revenue right away and then get to monetize these data centers for 30 plus years without having to spend that startup capital again.<br><br>Servers and networking equipment operate on a shorter cycle. We typically purchase these a few months before putting them into service, so we have strong visibility into customer demand before we trigger the spend. If the demand isn’t there, we won’t spend the capital. For servers and networking equipment, on average, it takes a little less than three years to break even on that investment. The servers currently have a useful life of at least five to six years. And most of our AI capacity these days is being contracted for at least five-year terms. That means that we’re driving significant free cash flow on the servers and networking equipment in the two to three years after we break even.

And later

We long believed AWS could become a few hundred billion dollar revenue business and now believe it’ll be at least double that, and very possibly be $1 trillion annual revenue business for us in time with very appealing accompanying free cash flow and return on invested capital.

The stock trades at ~28x earnings (~23x 2027E vs. a 3-year average NTM P/E of 34x). Consensus has revenue compounding ~15% a year over the next 3 years, with AWS growing 30%+.

Pershing Square Holdings — Q2 2026 Semi-Annual Letter

“Amazon’s share price increased 3% year-to-date as of June 30, 2026. As with our other hyperscaler investments, we believe investor concerns around the magnitude of capital expenditures behind the datacenter buildout at Amazon Web Services (AWS) continue to weigh on the stock. These concerns understate the resiliency of Amazon’s business and its significant growth runway, as evidenced by the company’s robust operating momentum. Increasing AI adoption has materially accelerated AWS’s revenue growth profile from 20% growth in 2024 and 2025 to more than 30% this year. Likewise, Amazon’s retail segment continues to take market share, with unit volumes up 15% in Q1 2026, the fastest pace since 2021. Longer term, we believe Amazon can compound earnings at a 20%-plus rate, driven by secular tailwinds from AI and rising e-commerce penetration alongside substantial retail margin expansion.”

Alphabet ($GOOG)

Alphabet is a major position for several superinvestors and when a diverse group of investors spanning Duquesne Family Office, Altimeter Capital Management, and Berkshire Hathaway all initiate or add, you pay attention. For GOOG, Duquesne and Altimeter initiated positions while Berkshire increased its stake by 83% from the shocking equity raise. The selling camp was equally notable: Tiger Global Management cut 45%, while Cat Rock Capital Management and Avala Global exited.

Alphabet or Google is a true battleground stock when it comes to AI. Altimeter, Stan Druckenmiller and Berkshire Hathaway all in whilst Sands Capital, Tiger Global, Divya Nettimi’s Avala, and Viking all trimming or out.

Alphabet has a similar thesis to Amazon but with a melting ice-cube in search. Everyone I talk to says their search usage is 90% lower but the evidence points to search actually accelerating. Is Google milking everything they can from search to put everything into GCP or is search actually growing with success from AI Overviews? And by-the-way, Buffett just participated in recent equity raise with him confirming on CNBC that it was his idea (not Greg, not Ted)!

GCP grew at an astronomical 82%+ last quarter and there are no signs of slowdown. While Amazon has Graviton and Tranium, GCP has by far the most mature AI Accelerator program with their Tensor Processing Unit (TPU) which is already on their 8th generation. GCP is now selling the physical chips to end-customers meaning GCP has a mini-NVIDIA embedded within the business which is arguably getting zero credit.

Over a 3 to 5-year period, the bull case relies on search continuing to generate enough cash to sustain hypergrowth in GCP as AI adoption accelerates. It looks like Google has lost the frontier race but perhaps it doesn’t matter, as long as OpenAI and Anthropic continue to be hungry for compute? After the quarter, there was major news with leadership changes at the top of Deep Mind including Deep Mind founder and Head stapping to Chair and Jeff Dean, Google’s highly regarded Chief Scientist departing.

Again, just like Amazon, Alphabet had guided towards a massive capex ramp ($205bn+) and the key question for the whole sector will be what ROIC will be on that capex spend. Perhaps analysts are underestimating the ability of Alphabet to leverage AI for internal initiatives including search, YouTube, and Waymo.

The stock trades at ~26x earnings, near the top of its historical 12-29x range. Consensus has revenue growing ~17-18% a year with ~20% operating income growth over the next 3-5 years.

Right Tail Capital — Q4 2025 Investor Letter

“I have invested in Alphabet (parent company of Google) for more than six years, and it has been a significant holding for Right Tail since inception. Throughout that period, I’ve often felt that the market has undervalued the quality and breadth of Alphabet’s business portfolio, even as the stock itself has performed reasonably well. […] Earlier this year, I added to our position when shares declined to roughly $150 amid concerns around tariffs, skepticism about Alphabet’s AI positioning, and an ongoing DOJ case. I believed the market was underappreciating Alphabet’s ability to compete in AI and the strength of businesses such as Google Cloud, YouTube, and Waymo.”

TSMC ($TSM)

TSMC is a major position for several superinvestors and when a diverse group of investors spanning Long Walk Management, Gavilan Investment Partners, and Third Point all initiate or add, you pay attention. For TSM, Long Walk opened a 15.3% position, Gavilan started at 8.8%, and Third Point added 67%. On the other side, Lone Pine Capital cut 93%, Blue Whale Capital exited, and Cantillon Capital Management nearly eliminated its position.

Taiwan Semiconductor Manufacturing (TSMC) is the world’s leading dedicated pure-play semiconductor foundry. They are the unquestioned leader in leading edge chips effectively manufacturing 100% of Nvidia’s and even a large share of Intel’s leading chips. With the huge supply constraint and TSMC’s refusal to overbuild (or even build), there is a huge runway of secured in-place demand for years or even decades, but their reluctance to build capacity has potentially left an opening for Intel or Samsung to re-emerge or even brand new onshore initiatives from the likes of SpaceX.

The bull case is obvious: TSCM has a technological lead, all their capacity is sold out several years into the future, they are expanding capacity into the US with a GIGAFAB cluster in Arizona and new capacity in Japan, and they are the trusted partner to all the leading semiconductor designers. The bear case has to be geopolitics combined with a re-emergence of onshore manufacturing. The stock trades at ~19x earnings (16-17x 2027E, roughly in line with its historical average). Consensus has revenue growing 30-35% a year through 2028, on top of 40%+ guided growth in 2026.

Wedgewood Partners — Q2 2026 Client Letter

“Another beneficiary of the AI spending boom has been Taiwan Semiconductor Manufacturing. Revenues grew by more than 40% (in USD), on top of 40% growth last year. Its leading-edge fabs and packaging capacity are fully booked, driving margins to all-time highs. Much of this capacity was put in place a few years ago, before generative AI was a household and business-wide term. More recent demand signals from customers - including Nvidia, Broadcom, and even Micron - indicate AI-related growth of over 50% per annum through 2029. Whereas the Company used to have demand visibility only a few quarters out, it now has visibility a few years out.”

Mercado libre ($MELI)

MercadoLibre is a major position for several superinvestors and when a diverse group of investors spanning Light Street Capital Management, Markel Group, and Skye Global Management all initiate or add, you pay attention. For MELI, Light Street initiated a 2.5% position while Markel and Skye increased their holdings by 340% and 126%. Conversely, Ancient Art and Soma Equity Partners exited, and D1 Capital Partners trimmed 26%.

MELI, unlike the tech behemoths, has less direct tie-in with AI albeit AI will clearly impact their business.

MercadoLibre, Inc. is the premier online commerce and digital financial services ecosystem in Latin America, with major operations spanning Brazil, Mexico, and Argentina. Its integrated business model combines the Mercado Libre online marketplace, Mercado Envios fulfillment and logistics network, Mercado Pago digital fintech services, Mercado Credito consumer/merchant lending, and Mercado Ads retail media solutions. The bull case rests on emerging market purchase frequency converging to developer market levels and the flywheel of owning e-commerce, payments, logistics, and banking cranking away driving a wider moat, growth, and operating leverage all at the same time. The stock trades at ~51x trailing earnings but only ~22x 2026E EV/EBITDA and ~12x NTM EV/FCF — near historical lows and a ~44% discount to peers. Consensus has EBITDA and EPS compounding 25-30% over the next 3 years.

Lakehouse Capital — Global Growth Fund, May 2026 Monthly Letter

“Bigger picture, we remain firm supporters of the business and view the recent share-price weakness as overdone. With e-commerce penetration across Latin America still relatively low and a large underbanked population yet to be served, we believe MercadoLibre is investing from a position of strength into a long and highly profitable runway.”

Visa ($V) & Mastercard ($MA)

Visa is a major position for several superinvestors and when a diverse group of investors spanning Pershing Square, Soma Equity Partners, and Junto Capital Management all initiate or add, you pay attention. For V, Pershing opened a 5.8% position, Soma began at 3.5%, and Junto initiated at 2.9%; the counterweight was Coatue Management and Atreides Management exiting, alongside a 97% reduction by DSM Capital Partners.

Mastercard is a major position for several superinvestors and when a diverse group of investors spanning Pershing Square, Route One Investment Company, and Eagle Capital Management all initiate or add, you pay attention. For MA, Pershing and Route One opened positions of 5.6% and 1.5%, while Eagle increased its stake by 330%. In contrast, Two Creeks Capital Management and Junto Capital Management exited, and DSM Capital Partners reduced its holding by 84%.

Moaty businesses like Visa and Mastercard are reaching historically low multiples on crazy fears that AI agents will somehow use new payment rails. Those concerned ignore the fact that the network effects and trust that Visa & Mastercard have established takes literally decades to build. Visa & Mastercard work seamlessly with hundreds of millions of merchants in 210 countries worldwide and they have proven be a beneficiary of trends like stable-coins where they can layer on Value-Added Services and Solutions (VAS) spanning cybersecurity, fraud management, digital authentication, and data analytics. The bull case is unchanged: the Visa & Mastercard duopoly will continue to compound earnings at double digits and at a 20’s multiple, there is room for durable, predictable growth. Bears can always point to regulations or macro, but the real reason the rails have taken a hit are due to fears and misunderstandings around AI.

Mastercard trades at ~27x earnings, a ~17% discount to its 3-year average multiple. Consensus has EPS compounding ~15-16% a year over the next 3 years. Visa is slightly more mature so trades 24x earnings with consensus expecting low-teens growth over the next 3 years.

Pershing Square Holdings — Q2 2026 Semi-Annual Letter (new position)

“Earlier this year, we initiated positions in Visa and Mastercard, two businesses we have long admired, which provide the dominant global networks for consumer and commercial payments, with an increasing share of revenue growth coming from value-added services. In our view, Visa and Mastercard are among the highest-quality businesses in the world. Both are capital-light “toll-takers” that earn a nominal fee on each transaction without taking any material risk and are natural beneficiaries of higher inflation. Their networks, built over decades, connect billions of consumers with hundreds of millions of merchants and thousands of financial institutions. […] Despite these attributes, Visa and Mastercard recently de-rated to 22 times next twelve months’ earnings. We attribute this to investor concerns around stablecoin disruption, agentic commerce, and proposed U.S. regulation, each of which we believe is misplaced. We believe stablecoins represent an opportunity for the card networks rather than a threat.”

Meta ($META)

Meta is a major position for several superinvestors and when a diverse group of investors spanning Hound Partners, Pershing Square, and Meritage Group all initiate or add, you pay attention. For META, Hound initiated at 11.8%, Pershing at 9.2%, and Meritage at 6.2%. The rotation was sharp on the sell side: Cat Rock Capital Management, Dragoneer Investment Group, and Avala Global all exited.

Meta operates Facebook, Instagram, WhatsApp, Messenger, and Threads—and the Reality Labs (RL) segment focused on virtual/augmented reality and AI hardware. Meta generates substantially all of its revenue from digital advertising and has seen its ROIC increase as its spent heavily on datacenters and GPUs entirely for its own use cases.

The key debate now is whether Meta needs to be at the frontier. All of the other hyperscalers have established customers to sell their capacity to if it all goes wrong but Meta only has one customers (although they may try to become a hyperscaler) so are all their recent initiatives to try and develop an in-house model worth it?

Over the next 3 to 5 years, Meta’s bull case hinges on compounding 15% to 20%+ earnings growth as its massive capital expenditures in AI infrastructure—reaching 5+ gigawatts of compute capacity—power next-generation generative AI products, full-stack personal and business AI agents, LLM-native recommendation engines, and potential direct compute monetization.

The stock trades at ~19x forward earnings vs. a 3-year average of ~24x and peers at ~25x. Consensus has revenue growing ~20% a year through 2028.

Pershing Square Holdings — Q2 2026 Semi-Annual Letter

“We view Meta as one of the clearest beneficiaries of AI, which is driving higher engagement through better content recommendation, greater ad monetization through improved targeting, and meaningful product innovation in consumer and business agents. Though still in the early days, these benefits are already visible in greater revenue acceleration in Meta’s core advertising business, which is growing at its fastest quarterly pace since 2021. Meta’s compute investment also affords downside protection as capacity not consumed internally can be profitably monetized in a highly supply-constrained compute market. Following this year’s elevated investment, we expect Meta to grow earnings at a 20%-plus rate over the next few years. Against this growth backdrop, we view Meta’s 18 times forward earnings multiple as a highly discounted valuation for a dominant consumer internet franchise with a stable user base of 3.6 billion daily active users.”

Booking ($BKNG)

Booking is a major position for several superinvestors and when a diverse group of investors spanning Sunriver Management, DSM Capital Partners, and Dorsey Asset Management all initiate or add, you pay attention. For BKNG, Sunriver initiated a 5.4% position, DSM established a 1.6% position, and Dorsey made the largest relative add. There were no reported trims, but Ako Capital, Lone Pine Capital, and Maverick Capital exited.

Booking Holdings Inc. is a global leader in online travel facilitating reservations across 220 countries through Booking.com, Priceline, Agoda, KAYAK, and OpenTable. One of the flagship initiatives is the “Connected Trip” strategy connecting hotels with cars to flights to restaurants.

The key question on everyone’s minds is AI. Pre-AI, the bull case rested on Booking’s strong market share in Europe combined with a surge of US travelers into Europe. AI changes the story: do we even need to go on Booking.com? Will Google/Booking still be relevant when ChatGPT can book your entire trip end-to-end?

Over the 1-year horizon, the bear case is shaped by potential macroeconomic headwinds and geopolitical travel disruptions—such as ongoing conflict in the Middle East causing elevated flight ticket prices and route capacity cuts.

The stock trades at ~20x earnings (17x 2027E vs. a ~21x historical NTM average). Consensus has revenue growing ~8% a year with low-to-mid-teens EPS growth.

Antipodes Partners — Q2 2026 Global Fund Commentary (new position)

“Booking Holdings is the leading online travel agency, with room nights and bookings running 2-3x those of its nearest peers, Expedia and Airbnb. The stock has fallen out of favour on concerns that AI agents will divert consumers away from Booking’s profitable direct-traffic model. We think the market underappreciates the continued underlying drivers of its no. 1 position, including its deep supply-side relationships with small hotels, paid ad flywheel, loyalty program and metadata. The business is also run by a tenured management team that has navigated it through multiple industry crises. Lingering travel fears tied to the Iran conflict, together with the stock’s underperformance versus hotel operators, offer additional margin of safety in a long-term GDP+ travel sector.”

S&P Global ($SPGI)

S&P Global is a major position for several superinvestors and when a diverse group of investors spanning Pershing Square, TenCore Partners, and Triple Frond Partners all initiate or add, you pay attention. For SPGI, Pershing initiated at 5.4%, TenCore at 4.5%, and Triple Frond at 4.2%. The other side saw BlueSpruce Investments, Fremen Capital Management, and Himalaya Capital Management exit.

S&P Global Inc. is a global provider of credit ratings, benchmarks, financial data, analytics, and workflow solutions serving capital, energy, and commodity markets through four core reportable segments: S&P Global Ratings, S&P Dow Jones Indices, S&P Global Energy, and S&P Global Market Intelligence (following the July 2026 spin-off of Mobility Global, the Carfax business).

S&P is basically a monopoly or duopoly on three of the most important businesses: ratings, indices, and Platts (commodities). Sure, Capital IQ might face some pricing pressure from the likes of alpha-file.com but those software businesses are a small part of operating income.

Over a 3 to 5-year period, S&P Global’s bull case is anchored in its very wide moat and high-margin benchmark franchises, benefiting from a massive multi-year global debt refinancing wall, accelerating AI data center and infrastructure financing, rapid expansion into private credit ratings and indices, and proprietary cloud and AI data tools that drive low-teens earnings-per-share compounding and strong capital returns.

Its frankly hard to see a major bear case: potential cyclical pressure on transaction-based revenue with macro volatility or high interest rates dampening corporate debt issuance alongside near-term geopolitical headwinds in the Energy segment.

The stock trades at ~24x forward earnings, a discount to history. Consensus has 6-8% organic revenue growth driving low-teens EPS growth over the next 3 years (Pershing argues closer to 20%).

White Brook Capital — Q1 2026 Investor Letter (new position)

“S&P Global ($SPGI) is one of the financial market’s greatest toll businesses. […] In reality growth and tumult in private credit, data center financing, and AI driven financial engineering products reinforce the necessity of the company’s services as a provider of ratings, analytics, and insight. […] We believe the company is a rare combination of defensive resilience and high-growth potential and are thrilled by the price that we were able to buy it.”

MSCI ($MSCI)

MSCI is a major position for several superinvestors and when a diverse group of investors spanning Viking Global Investors, BlueSpruce Investments, and Steadfast Capital Management all initiate or add, you pay attention. For MSCI, Viking and BlueSpruce initiated positions, while Steadfast increased its stake by nearly 300%. Ownership Capital B.V. cut 89%, Meritage Group trimmed 32%, and Himalaya Capital Management exited.

MSCI is a pureplay indices business with S&P Global being the nearest comp. MSCI’s businesses include the well-known MSCI indices but also high quality assets in data, analytics, and ESG including owning the Barra risk factor model used by institutional investors globally. These are standards that are hard to switch away from.

Over a 3-to-5-year horizon, the bull case rests on sustained low-double-digit organic revenue growth driven by the secular global shift to passive investing, expanding adoption of custom indexing, rapid growth in private asset benchmarking across a multi-billion-dollar addressable market, AI-enabled product monetization, and high operational leverage supporting long-term adjusted EBITDA margins above 60%.

The bear case feels superficial: flat-to-negative growth in the ESG business from U.S. political backlash and European regulatory uncertainty, some mix shift potentially compressing fees, and general market volatility impacting asset-based fees.

The stock trades at ~28x earnings and ~20x 2027E EV/EBITDA, roughly two standard deviations below its 5-year average. Consensus has ~9-11% revenue growth driving mid-teens EPS growth over the next 3 years.

GFI Investment Counsel — GoodQuarter, Q4 2025

“MSCI licenses indexes, data, and software solutions to a broad set of financial customers. While growth in the traditional asset manager segment has moderated due to industry maturity and slowing demand in ESG solutions, the remaining ~70% of the business, comprising asset-based fees and subscriptions sold to banks, hedge funds, and wealth managers, should continue to grow at a double-digit rate for many years to come. […] As private markets continue to grow, we believe this represents a large and durable growth opportunity for MSCI.”

Medline ($MDLN)

Medline is a major position for several superinvestors and when a diverse group of investors spanning Pelham Capital, Egerton Capital (UK), and Lone Pine Capital all initiate or add, you pay attention. For MDLN, Pelham initiated a 3.4% position, Egerton increased its holding by 277%, and Lone Pine added 78%. Sellers were more decisive: Fremen Capital Management and Durable Capital Partners made large reductions, while Maverick Capital exited.

Medline has been described as the Costco of healthcare distribution. Medline is the largest provider of medical-surgical products and supply chain solutions in the United States. The company generates revenue by manufacturing and distributing its own branded Class I and Class II consumable medical products—including surgical procedural trays, gloves, protective apparel, and wound care—and by distributing third-party national brand products through its logistics network. Medline is known to be very cost conscious, pass savings on to end-customers, and has their in-house “Kirkland signature” equivalent allowing customers to save massively on every day consumables.

The stock trades at ~27x 2026E earnings and ~19x 2027E EBITDA, a premium to distribution peers at ~14x. Consensus has revenue growing ~8-10% a year through 2028.

Artisan Global Discovery Fund — Q1 2026 Quarterly Commentary (added on weakness)

“Medline provides medical-surgical products and supply chain solutions through its Medline brand and supply chain solutions segments, combining vertically integrated manufacturing across core clinical categories with third-party product distribution and customized logistics services for health care customers globally. This integrated model benefits from significant scale and long-standing customer relationships, supporting a large, recurring revenue base and what we believe to be attractive long-term economics. […] We viewed the pullback as an attractive opportunity to increase our position, given Medline’s improving growth profile, expanding profitability and a valuation that we believe does not yet fully reflect its competitive positioning and longer term earnings potential.”

TransDigm ($TDG)

TransDigm is a major position for several superinvestors and when a diverse group of investors spanning Third Point, Cat Rock Capital Management, and Windacre Partnership all initiate or add, you pay attention. For TDG, Third Point increased its position by 216%, Cat Rock by 37%, and Windacre by 34%. On the sell side, Lone Pine Capital, Long Path Partners, and Sands Capital Management all trimmed.

TransDigm is a leading commercial aerospace and defense component supplier focused on proprietary, sole-source parts with substantial aftermarket exposure. Their reputation is as a price gouger but the reality is that their products are captive for the life of an aircraft giving them regulatory monopoly status. Customers simply can’t switch. The company’s portfolio is 90% proprietary and 75% are sole-sourced.

Over a 3-to-5-year horizon, the bull case rests on sustained earnings compounding driven by global air traffic recovery, expanded commercial OEM build rates, M&A, and of-course pricing. Over the next year, the bear case focuses on macro and potentially a less favorable fleet age (older planes need more maintenance).

The stock trades at ~29x earnings and ~21x EV/EBITDA, a 30%+ discount to aerospace aftermarket peers. Consensus has EPS compounding in the low-to-mid teens through FY28.

YCG, LLC — Q4 2025 Investment Letter (new position)

“TransDigm has spent the last 30 years building a portfolio of hundreds of thousands of parts with these characteristics. And while each part breaks unpredictably and infrequently, by assembling this vast portfolio, they have essentially created a collection of near monopolies that create an incredibly smooth and high pricing power annuity stream. Moreover, this annuity is renewable and growing over time.”

Spotify ($SPOT)

Spotify is a major position for several superinvestors and when a diverse group of investors spanning Greenlea Lane Capital Management, Ownership Capital B.V., and Dorsal Capital Management all initiate or add, you pay attention. For SPOT, Greenlea Lane initiated at 5.0%, Ownership Capital began at 1.7%, and Dorsal increased its stake by 71%. The counterweight was Cantillon Capital Management cutting 95%, plus exits by Junto Capital Management and Lone Pine Capital.

Spotify is the global leader in audio subscription streaming, operating across Premium subscription and Ad-Supported tiers spanning music, podcasts, and audiobooks. We all use it.

Over a 3-to-5-year horizon, the bull case projects a low-to-mid-teens revenue compound annual growth rate and a 25% to 30% EBITDA CAGR, anchored by expansion toward 1 billion monthly active users, sustained pricing power, two-sided marketplace growth, and margin-accretive non-music content scaling that pushes gross margins toward 35% to 40% and EBIT margins to 20% or higher by 2028–2030.

The bear case emphasizes near temporary margin pressure from elevated high-teens operating expense growth driven by AI compute and marketing investments, lingering softness in advertising monetization, near-term monthly active user deceleration from free-tier optimization in emerging markets, and intense pricing or licensing pressure from concentrated record label suppliers and tech competitors.

The stock trades at ~33x earnings, falling to ~22x on 2028E numbers and well below its (short) post-profitability average of ~60x. Consensus has mid-teens revenue growth and a ~20% FCF CAGR over the next 3-5 years.

Artisan Global Discovery Fund — Q1 2026 Quarterly Commentary

“Spotify, a leading audio streaming platform, is well positioned to further monetize its dominant role in global music distribution and discovery through subscription pricing, advertising and tiered subscription offerings, in our view. We believe the upcoming super-premium tier represents a meaningful catalyst. In its most recent quarter, Spotify delivered solid user growth and operating leverage, with revenue and operating income exceeding guidance.”

PDD ($PDD)

PDD is a major position for several superinvestors and when a diverse group of investors spanning Himalaya Capital Management (Li Lu) and H&H International Investment (Duan Yongping) all initiate or add, you pay attention. For PDD, Himalaya increased its position by 134% and H&H added 27%; there were no reported trims. The sell signal came entirely through exits by Tiger Pacific Capital, Cat Rock Capital Management, and Appaloosa.

PDD is a leading Chinese e-commerce company operating domestic marketplace Pinduoduo and global cross-border platform Temu. Key operations focus on driving high purchase frequency, empowering small-to-medium enterprise (SME) merchants, and expanding Temu across North America, Europe, and emerging global markets through fully-managed and semi-managed supply chain models.

Over a 3-to-5-year horizon, the bull case centers on PDD continuing to gain domestic e-commerce market share due to consumer preference for value-oriented products, scaling Temu toward breakeven and high transaction volume across international markets, and successfully establishing its three-year, RMB 100 billion “Xinpinmu” first-party brand initiative to orchestrate global supply chains.

The bear case focuses on near-term earnings volatility and margin compression driven by a soft Chinese macro, stricter tax enforcement impacting SME merchants’ advertising budgets and GMV, heavy ongoing commitments to its RMB 100 billion merchant support initiative, and front-loaded investments in first-party brand building.

The stock trades at just ~8.5x forward earnings (~4x ex-cash). Consensus has revenue growing ~8-10% a year with earnings roughly flat in 2026 before returning to low-double-digit growth. This is owned by quality investors but is trading near deep value territory.

RV Capital — 2023 Annual Letter (PDD deep dive)

“In contrast to many fast-growing companies, PDD’s rapid growth has not come at the expense of profitability. At the 9-month mark, PDD had a GAAP operating profit margin of 23%. Its free cash flow margin is higher still as merchants are required to put down deposits to sell on the platform and purchase advertising, and only receive payment on sales after the order is completed. PDD has a fortress balance sheet thanks to its strong cash generation.”

I hope this article gives you a sense of what a superinvestor portfolio looks like today. Of course, feel free to explore, create your own lists, read more letters and deep dives, all available on alpha-file.com

Subscribe now

Quality Managers

Alpha-File currently lists 55 entities with its Quality tag. The 46 manager profiles are listed first; the nine tagged fund profiles follow separately so the list is complete.

Manager profiles

• H&H International Investment

• First Manhattan

• Pershing Square

• ValueAct Holdings

• Akre Capital

• Eagle Capital

• Gates Foundation Trust

• Cat Rock Capital

• Windacre Partnership

• BlueSpruce Investments

• Steadfast Capital

• Viking Global Investors

• Valley Forge Capital

• Park Presidio Capital

• Gobi Capital

• Junto Capital

• Giverny Capital

• XN

• Dorsal Capital

• TCI Fund Management

• Dorsey Asset

• Fremen Capital

• Anomaly Capital

• Skye Global Management

• RV Capital

• Greenlea Lane Capital

• Zeno Equity Partners

• Stockbridge Partners

• Ruane, Cunniff & Goldfarb

• GFI Investment Counsel

• Sunriver Management

• Meritage Group

• Longview Partners

• Triple Frond Partners

• Gardner Russo & Quinn

• Oakcliff Capital Partners

• Pelham Capital

• Ako Capital

• Alfreton Capital

• BloombergSen

• Markel Group

• Long Path Partners

• Cantillon Capital

• Egerton Capital

• Eleva Capital

• Ownership Capital

Tagged fund profiles

• Akre Focus ETF

• Eagle Capital Select Equity ETF

• Dodge & Cox Stock Fund

• Fidelity Contrafund

• Fidelity Magellan Fund

• Sequoia Fund

• FM Focus Equity ETF

• FM Compounders Equity ETF

• Growth Fund of America

A note on the data: All portfolio data and charts in this piece are sourced from Alpha-File. The same position-history, trading, composition, industry, P&L, and fundamentals exhibits are available for every tracked manager on Alpha-File, including 13F filers, US N-PORT mutual funds and ETFs, and supported European, Canadian, and other global funds. Public-company and manager names throughout link directly to their specific Alpha-File pages.

Regular readers have already met Nicolai Tangen from behind the microphone.

When Chris Hohn explained in the TCI profile why he takes 10–15% positions, and confessed to selling Moody’s at $100 and buying it back at $150, he was answering questions on In Good Company. When Mala Gaonkar laid out SurgoCap’s entire philosophy — the telescope and the microscope — that was the same podcast.

This profile is about the man asking the questions, and it turns out his interviewing skills are not just a hobby he picked up — it is part of the edge of the firm he built.

AKO Capital also completes a bloodline. The Egerton profile introduced John Armitage. Tangen worked for Armitage. So the chain runs Armitage → Tangen, and Armitage → (admiring) Hohn, and Tangen → (interviewing) Hohn, and the three of them between them run or built something like $100 billion of some of the most similar portfolios in Europe. Moody’s, Visa, Mastercard, CPKC, FICO, Linde — you will find these names in all three books.

The Investor

Nicolai Tangen was born in Kristiansand, on Norway’s southern coast, in 1966. Before university there was the Norwegian Intelligence Service — Russian language, and something that in the 1980s was called by its proper name:

“The first time I started to think about interviewing was when I was in the Norwegian intelligence service. And as part of the program there, we did some training in interviewing. I guess it was called interrogation at that stage, but now it’s not such a popular word. You just thought what you could get out of people if you asked the right questions and the right follow-up questions and right preparation level.” — Capital Allocators, EP.354

Then the Norwegian School of Economics, then Wharton (class of 1992), then five years at the London broker Cazenove, where he credits a single mentor with everything technical he knows:

“I worked with a guy called David Croft who ran the European Department, a super guy who sat down with me, taught me to build models, taught me to ask questions, really taught me everything I knew about analysis.” — Capital Allocators, EP.354

And then the apprenticeship that made him — Egerton Capital, at a time when it was one of the first hedge funds in Europe, working for John Armitage and Bill Bollinger:

“We talk about AAA in financial markets. I think if you talk about people, they are really triple A plus. And I think we underestimate how important apprenticeship is for investing. And starting a firm is not something you can just do. You need to learn from something.” — Capital Allocators, EP.354

From Armitage he took the business-quality lens:

“What is a good business, what are important moats, what is operational leverage, what does it really mean, how much of your incremental revenue drops down to the bottom line.” — Capital Allocators, EP.354

From Bollinger he took something rarer:

“You have to make hay when the sun is shining. The easiest thing is just to hedge away all types of risks. It’s to hide. It’s to spend all your money on put options to just protect your own set. But there are times when you just have to run naked. You have to take away the protection… And that’s a real skill set. When to take away the brakes and go for it.” — Capital Allocators, EP.354

And the trigger is fear:

“You need to have max exposure when the world looks the worst. When you are at the bottom and you are the most afraid and you can hardly sleep, that’s when you need to take your protection away.” — Capital Allocators, EP.354

Then he did something unusual. He stopped, for three years. He learned to cook. And he spent two of those years at the Courtauld Institute reading art history, because he had been collecting for years and “just felt that need to put it into some kind of system.”

In 2005 he set up AKO Capital:

“Managed to hire some really good people. First person I hired was the secretary I had at Egerton, and she was phenomenal… And then Thomassen, a Norwegian fellow. And then we built up a team. We were like 7 people. 7 people and a dog. And we raised $550,000,000. It was a very good launch.” — Capital Allocators, EP.354

Fifteen years later it was north of $20 billion. Along the way he did a second master’s — social psychology at the LSE — on corporate culture, because he could not stop noticing that two companies doing the same thing produce opposite outcomes. “You see it in the elevator industry. You see it in the footwear industry… Some companies are doing well. Some companies don’t. What is it?”

And the art collection became the largest private holding of Nordic modernist art in the world, which he gave away to his home town; the Kunstsilo museum opened in Kristiansand in 2024, in a converted grain silo.

Philosophy & Process

1. Interrogation is a research edge.

Most funds say they do good channel checks. AKO hired the people who train police:

“When I set up AKO, I thought, you know what, we probably can get more out of companies than other people. And we started to train the analysts in interview technique or interrogation or whatever you call it. Got the leading interrogator from the Norwegian police force to set up a course. We hired an English person who was superb and who had helped develop the whole interrogation framework for the English police force. And together, they made a program. And it was the best program that money could not buy because there wasn’t anything really available.” — Capital Allocators, EP.354

The technique itself is anti-charismatic, which is why so few analysts do it: “Ask simple questions. Don’t build too much into the question. Don’t try to show the other person that you are an intelligent guy… They’ll realize soon enough that you have done your homework. Don’t ask double questions and just keep it short and drill down.” Anyone who has listened to In Good Company has heard this method running live.

2. Quality was a conclusion, not a starting position.

AKO did not launch as a quality shop. It backed into it by noticing an asymmetry in the surprises:

“It started off as a mid cap deep research firm, gradually moved a bit larger caps because we thought that the research methodology worked for large companies as well. Then increasingly became quality investing focused… we just saw that over time, the quality companies just surprised on the upside. The bad companies just always had profit warnings, bad surprises, bad management changes. But the good ones — hey, you know what? Always something great. They pull some kind of rabbit out of the hat.” — Capital Allocators, EP.354

Thomassen tells the same story with a date attached. The pivot came out of the financial crisis, in an off-site where the choice was posed starkly:

“We had an off-site during the financial crisis, and we thought either we have to retrain as plumbers and electricians, the whole financial system is gonna come crashing down. Or this could be an amazing opportunity… blank piece of paper. You can buy whatever you like because everything is cheap. So we thought, well, let’s be a bit optimistic, and let’s do the latter. So we just thought, okay — what are the best companies in Europe? We started buying those.” — Gorm Thomassen, Accelerating Excellence, 2023

And then AKO did the thing that makes it unusual in this series: it wrote the book. Quality Investing: Owning the Best Companies for the Long Term (2016) was written by AKO’s own Torkell Eide and Patrick Hargreaves with Lawrence Cunningham — Warren Buffett’s essayist — and has sold over 26,000 copies in four translations. Most managers guard the framework. AKO published it.

3. Disaggregate the analyst.

This is AKO’s most transferable idea, and Thomassen states it more bluntly than any manager in this series would dare:

“We think, for instance, that channel checking is much better done by professional market research team who can make 15,000 calls in 20 different languages than 1 guy… He’s invested his own ego in making an investment recommendation. Then you make channel checks, you just hear what you want to hear, and everything will sound really positive to you. Whereas if we do something arm’s length with a market research team, that is much more unbiased information.” — Accelerating Excellence, 2023

So the analyst’s job was broken into pieces and handed to specialists — what the interviewer called AKO’s “black belts”: forensic accountants, forensic linguists reading conference-call transcripts, behavioural psychologists assessing management, a data-science team, and an in-house market research operation. Tangen’s own list:

“In addition to interviewing, we brought in forensic accounting group, really important… We brought in forensic linguists — important, again, to understand conference call transcripts and meetings. Very deep market research. So a big team of telephone interviewers and then data scientists and analysts.” — Capital Allocators, EP.354

Note what this costs. Whale Rock runs ten people and 2,500 meetings; Valley Forge runs two people and seven stocks. AKO grew to seventy employees for a book of twenty-odd names. It is the most expensive research architecture in this series, and the bet is that removing the analyst’s ego from the evidence-gathering is worth more than it costs.

4. Process over outcome, because the feedback loop is three years long.

Thomassen’s version of the discipline:

“You make an investment decision. It could be 3 years until you know whether you’re right or wrong. And in the meantime, the market will tell you maybe every single day for a year that you are a complete idiot and that you are completely wrong… If you get 55% right and 45% wrong, you’re a complete genius — which means you’re getting 45% wrong, which is quite a lot of getting stuff wrong.” — Accelerating Excellence, 2023

Two mechanisms follow. The first is an accounting tripwire with no discretion attached: “there are no good explanations for why companies start to go aggressive on their accounting. So if the accounting team tells us that this is not looking so good, then routinely, we will sell it.” That rule was bought with tuition — an early position in a company that turned out to be cooking its books, where AKO “lost more than 90%,” and which he can still feel “more than 10 years later.”

The second is a database built from post-mortems, designed to catch what he calls smoke signals:

“A company will not go wrong overnight… If you go and do a postmortem of companies that have gone wrong, you will see in hindsight there are little signs, but individually, none of these signs are big enough for you to sit up and pay attention… So we’ve built a database to collect all the work we do on each individual stock. And then that way, we can capture these clusters of little things.” — Accelerating Excellence, 2023

The sell question, in his phrasing: “when is a potential problem just a cough and when is it actually cancer.”

5. “Lazy” is deliberate.

“We’re taking this really lazy approach in that we identify good companies and the best managers, and then we sort of rely on them to do all the hard work. And we just sit on the sidelines, talk to them, make sure nothing’s going wrong. I can’t teach Michael O’Leary anything about running an airline… I can’t tell the guys at L’Oréal anything about how to run a cosmetics business, but we think they are the best, and we think they’ll make the right decisions.” — Accelerating Excellence, 2023

The Dream Job

In March 2020, Norges Bank announced that the next chief executive of Norway’s sovereign wealth fund — then $1.1 trillion, now the largest single shareholder on earth — would be Nicolai Tangen. What followed was described inside the institution as the biggest crisis in the fund’s history.

The trigger was a party. In November 2019, Tangen had flown roughly 150 prominent Norwegians and international guests to Philadelphia for what he called a “dream seminar” — academic programming at Wharton, wrapped in two chartered aircraft (one of them the Crystal Skye, a Boeing 777 with a restaurant, butlers and beds, at a reported total flight cost of about $1.4 million), plus concerts by Sting, Gregory Porter and Sara Bareilles. His own framing of it: “You are all exceptional. You all want to extend your range and open your minds. But, like me, you may not always have sufficient time in your daily lives to do this, which is why I have had to whisk you away.”

Among the guests was Yngve Slyngstad — the sitting CEO of the fund Tangen would shortly be appointed to run, who flew home on Tangen’s plane and later admitted, in a public apology, “I really screwed up.”

Then came the audit of everything else: £67 million of Tangen’s personal money in Cayman- and Jersey-registered funds, at an institution that had made tax transparency a public campaign; a live dispute between AKO and HMRC over an incentive scheme; a supervisory council that submitted 26 questions and received an 83-page answer; a parliamentary hearing in August 2020; a finance committee that instructed the minister to intervene.

He could have walked away. Instead:

• He gave his 43% stake in AKO Capital to the AKO Foundation, the charity he had established in 2013 — which is why AKO Capital today describes that foundation as its largest owner.

• He closed the AKO Trust, put his remaining wealth in a blind trust, resigned from all AKO boards, moved back to Norway and became a Norwegian taxpayer.

• He took a salary of roughly $630,000 to run $1.1 trillion.

Stated plainly: one of Europe’s best hedge funds is now majority-owned by a charitable foundation. Every year AKO earns its fees, a large share flows to education and the arts rather than to a founder’s balance sheet. The AKO Foundation has distributed over £100 million.

He started on 1 September 2020. In March 2025 the board reappointed him for a second five-year term — Ida Wolden Bache: “Nicolai Tangen is the right person to head NBIM in the years ahead.” Tangen: “Creating wealth for the Norwegian people is the most rewarding job I have ever had.”

And the fund he left?

“I felt that the learning had flattened out a bit. I felt that it was time to pass it on to the next generation. We had spent a lot of time on preparing the rest of the team to take over. I probably spent 8 years preparing their transition… AKO had been very successful. I thought if I could leave that to the next generation in a successful way, that would be fantastic because that rarely happens in this industry.” — Capital Allocators, EP.354

Patrick Hargreaves — ex-Goldman, co-author of Quality Investing — became CEO. Gorm Thomassen, the co-founder from the launch, became CIO. Six years on, the succession has held.

What Does AKO Own Today?

AKO runs long-only and long/short strategies, with put-option hedging and a book that has always been substantially non-US. The 13F shows 22 positions worth $4.29 billion at 30 June 2026. The firm has been described at $20–22 billion in recent years. So this is a window onto perhaps a fifth of the firm.

The disclosed US book expanded rapidly into 2021, peaked above $10 billion, and then contracted in two stages. The latest leg is the sharpest: reported value fell from $7.9 billion in June 2025 to $4.29 billion in June 2026. Some of that is portfolio performance and security selection; some may be capital moving between vehicles.

Two structural numbers from Alpha-File frame the behavior: average turnover of 30.2%, an implied hold of about 3.3 years, and 73 distinct investments across the filing history since Q4 2016.

The first table shows weights, trading direction, holding periods and estimated P&L; the second puts current valuation, growth and balance-sheet characteristics beside each holding; the third shows the price performance that sat underneath the reported position changes. Together they make the distinction the 13F table alone cannot: whether a weight moved because AKO traded, because the stock moved, or because both happened at once.

What the book costs — and how fast it grows

AKO’s disclosed portfolio now trades at a 33.7× weighted-average LTM P/E. Interestingly, it is near a period low.

Growth is not very dramatic. Steady, and a relatively narrow band. The weighted rolling five-year revenue CAGR is 9.1%. The combination — low-double-digit portfolio growth at a low-30s earnings multiple — makes the quality premium visible.

The portfolio’s industry mix has changed more than its stated philosophy. Consumer exposure that once dominated the US book has steadily receded, while industrials, technology and financial infrastructure now carry most of the weight. While the industries have changed, as you’ll see, the portfolio irrespective of sector is comprised of recurring demand, pricing power, scarce assets and high switching costs.

Sunbelt Rentals

Sunbelt Rentals went from 3.19 million shares to 6.07 million — a 91% increase, taking it from 4.1% to 10.6% of the book and making it AKO’s second-largest disclosed holding. In a quarter when they were selling nearly everything else, this is where the money went.

This is not a new idea. Sunbelt Rentals is Ashtead Group, the UK-listed equipment rental compounder, which completed its move to a US primary listing and NYSE ticker in early 2026 and took the name of its American operating business. The story is one of local market scale economies and a long penetration rate as rental penetration in the US is still relatively nascent but with heavy demand coming from AI infrastructure spend, there is both an organic and inorganic growth opportunity.

A long-standing European holding became a US-listed one — and then AKO doubled it. The same thing that made Ferrovial appear in TCI‘s 13F in 2024 is happening here.

Alcon — 15.2%, the seven-year conviction

Alcon, the Swiss eye-care business spun out of Novartis in 2019, is the largest position at 15.2% ($652M, 9.68 million shares) and one of the few names AKO added to in the June quarter. Interestingly, Bill Ackman’s Pershing Square just initiated (it trades in both the the US and Switzerland with Pershing owning the Swiss shares. Held for seven years, it is the purest expression of the AKO screen: a global duopoly position in surgical equipment and contact lenses, a consumable razor-blade model, and demographics that only run one way.

It has also gone nowhere. Seven years in, the reconstruction puts it at −$3.3 million on a $652 million position — a 1.0× MOIC, an IRR of −0.1% — cost basis $71.7 against $67.3 in June. AKO’s largest, longest-held, most characteristic position has returned approximately nothing for seven years, and they keep adding.

The toll booths: Visa, Mastercard, Moody’s, ICE, FICO, Equifax (~21%)

This is where AKO overlaps almost exactly with the rest of this series. Visa at 7.9% has been held nine and a half years. Moody’s at 4.5% and FICO at 2.9% are the same nine-and-a-half-year vintage. ICE, Mastercard and Equifax fill out the set.

TCI owns the ratings duopoly. Valley Forge owns FICO, S&P Global, Moody’s, Mastercard and Visa and almost nothing else. Egerton owns Moody’s and Visa. AKO owns the lot. Four managers, four countries of origin, four completely different processes — one portfolio.

Perhaps this is strong evidence that quality investing converges on a genuinely small set of businesses, or it is the most crowded consensus in global equities dressed up as independent thought.

The trading tells you AKO is at least alert to it: ICE cut 46%, Equifax cut 67%, Moody’s 16%, FICO 19%, Visa 17% — while Mastercard was added 22%.

The industrial quality sleeve: GE Aerospace, CPKC, Carlisle, Air Products, Linde, Williams, Copart (~20%)

GE Aerospace at 6.7% is the engine-aftermarket annuity that TCI and Hibbert also own — cut 23% on the quarter. Canadian Pacific Kansas City at 4.9% is the same single-line North American railroad TCI backed. Air Products was added 137% and Copart 140%, while Linde — historically AKO’s second most profitable disclosed position — was cut in half.

The two smaller March initiations complete the sleeve: Carlisle, the building-products compounder, and Williams, the natural-gas infrastructure operator.

The technology sleeve, quietly dismantled: Amazon, Microsoft, Analog Devices, Cisco (~22.9%)

Every US technology position was cut in the June quarter: Amazon −34%, Analog Devices −40%, Cisco −35%, Microsoft −15%. Nothing was added.

Ferrari — 7.7%, the one that isn’t like the others

Ferrari has been held nine and a half years and sits at 7.7%.

What left, and what it says

Five exits in the June quarter: Booking Holdings (a six-year hold and, on the reconstruction, one of the book’s best trades), Salesforce (held about a year), Flutter Entertainment (about a year, and painful), Marsh & McLennan, and a residual RELX ADR line. Alongside them, Waters cut 75% and Stryker cut 92% — the healthcare sleeve outside Alcon effectively closed.

Put the whole quarter together and the shape is unmistakable: sell the periphery, keep the tollbooths, and put the proceeds into one equipment-rental compounder and more of the eye-care business.

The Ghosts

The cumulative P&L reconstruction makes that asymmetry in AKO’s performance visible: a handful of long-duration winners account for most of the green, while Flutter is the one loss large enough to alter the picture.

Note the shape of that top five. Two of them are nine-and-a-half-year holds compounding at 2.3–2.6×, and one — GE Aerospace — did 2.51× in under three years.

The losses are unusually small for a book this size, which is the real signature of the process: Flutter Entertainment −$399M (0.44×, one year), then a steep drop to Equifax −$108.1M (0.83×, five years), United Technologies −$67.7M (0.82×), Copart −$64.6M (0.86×) and Charter Communications −$60.3M (0.76×, eight months). Outside Flutter, AKO has essentially not had a disaster in a decade. That is what the forensic-accounting department is for.

The pattern across AKO’s exits is different from every other manager in this series. TCI sells when it sees around a corner. Hibbert sells when the reinvestment runway shortens. Whale Rock sells when the S-curve matures. AKO sells when the evidence-gathering machinery flags something — an accounting drift, a cluster of smoke signals, a management change — which is why the holding period is three years rather than ten, in a portfolio of businesses it would happily own for twenty.

What Remains Unknown

Almost everything that matters about the non-US book: European and UK holdings, the short book, the put-hedging position, gross and net exposure, and the split between the long-only and long/short vehicles. No public letters. No published performance figures since the 2020-era claim of “returns three times better than the market since launch” — which means the most important open question in this profile, whether AKO has kept compounding through the quality drawdown of 2022–2025.

I’ll be tracking every filing on Alpha-File.com, where you can pull this same view — position history, initiations and exits, fundamentals, cumulative P&L, and Tangen’s full interview transcripts — for AKO or any manager, free. If you enjoyed this, subscribe: the next manager profile in the series is already in the works.

All portfolio data and charts in this piece are sourced from Alpha-File. The same position-history, trading, composition, industry, P&L, and fundamentals exhibits are available for every tracked manager on Alpha-File, including 13F filers, US N-PORT mutual funds and ETFs, and supported European, Canadian, and other global funds.

The fourth bloodline

This series has traced the TCI tree (TCI, WindAcre, Theleme), the Tiger line (Lone Pine, SurgoCap), the star PM inside the institution (Hibbert at BlackRock), and the Omaha line (Valley Forge, Greenlea Lane). Whale Rock belongs to the fourth great American training ground, the one that produced Atreides: Fidelity.

Gavin Baker joined Fidelity out of college in 1999 and left in 2017, at which point he was running the $17 billion OTC fund Alex Sacerdote arrived as a summer intern in the same 1999 class, left after six years, and started Whale Rock Capital Management in Boston in 2006 — a full thirteen years before Baker went out on his own. The two now run two of the best-known technology hedge funds in America, from the same city, out of the same apprenticeship.

Whale Rock has compounded at roughly 44% a year over the past three years, gained 51% in 2024, ran +72.5% through the first half of 2026 — and then lost 21.7% in July alone. This is a fund that has twice been left for dead: down 43% in 2022, down 20% in the first quarter of 2025. It is one of the most volatile manager in this series.

The Investor: A Share of Apple

Sacerdote grew up in New York with the market in the house. His father, Peter Sacerdote, was a Goldman Sachs partner for forty-one years — head of corporate finance in the eighties, chairman of the private equity group in the nineties, chairman of the commitments committee for two decades.

“I think actually the first stock that I owned was Apple back when the PCs were coming out, and I owe it to my father… We had a few shares of Apple, and we were so excited when it split. Then in college, I remember buying Boeing on the theory that they were somewhat of a monopoly and air travel had a huge future.” — Capital Allocators, EP.436

The father helped helped Sacerdote form a weekly habit — Barron’s, every week, a habit which stuck:

“One of our classmates basically decided to invest in my firm early on because we were on a weekend at HBS, and he saw me reading with a flashlight Barron’s at 2 AM.” — Capital Allocators, EP.436

Smith Barney’s TMT group came first, then a year at an internet advertising startup called Interactive Imaginations — a DoubleClick competitor that pioneered the ad network, “a good, not great company,” and the source of the ad-tech literacy that would pay off spectacularly twenty-five years later. Then Harvard Business School, class of ‘99, and the internship that made him.

Fidelity handed the interns a sector each. Sacerdote got the one nobody wanted:

“And I said, oh, you know about the Internet. There’s this new sector called e-commerce, and our current Internet analyst thinks Amazon’s a fraud. Why don’t you cover these 8 companies? And it was Amazon, egghead.com, Preview Travel, On Sale, N2K, CD Now, which sold CDs online. I was like a kid in a candy store.” — Capital Allocators, EP.436

He spent the summer visiting every one of them, attended Amazon’s first investor day, had lunch with Jeff Bezos (”He still had that laugh”), and wrote a forty-page report whose conclusion is the seed of everything that follows: the leader grows bigger, faster, and wins. Buy Amazon, short everything else in the space.

“Half the people at Fidelity thought I was crazy because here was this book company that was gonna beat Barnes and Noble and made no sense. But we looked very carefully into the unit economics, the growth. At that point, even if the Internet did not grow, they only had 2,000,000 customers, and there were 90,000,000 Internet users at that point.” — Capital Allocators, EP.436

One person bought it: Will Danoff. Sacerdote got the job.

Six years at Fidelity followed, and the rotation? Internet, then software, then gaming/lodging/leisure/media, then the Select Leisure Fund, then the aerospace and defense fund, then two years as sector PM running large-cap and mid-cap tech. Sacerdote also got the Peter Lynch treatment — Lynch met analysts a few times a year and started an egg timer when they walked in:

“You’ve got to figure out how to do it quickly… You’ve gotta get your earnings estimates, your P/E, get to the crux of it very quickly. And I find that’s a hard skill for a lot of people where everybody’s pressed for time.” — Capital Allocators, EP.436

He left because Fidelity’s career track ran toward big diversified funds benchmarked to the S&P 500, and he had concluded the opposite: “In the technology sector, it’s important to be a specialist.” Whale Rock launched in 2006 — and the first phone call was to his father, who answered it with a counter-offer:

“I’ve been at Goldman for 41 years. How about I come and join you? I’ll be the gray hair, I’ll be the oversight, I’ll be the chairman. You do what you do — you build the firm, build the team, run the money. I’ll help raise some money.” — Invest Like the Best, EP.477

They worked together for six years, until Peter Sacerdote died in 2011. Asked by O’Shaughnessy for the kindest thing anyone has ever done for him, that is the story Sacerdote tells. “We never raised our voices… And if I could be half the person he is, I’d be completely winning.”

Philosophy & Process

1. The S-curve is a map into the future

Many manager hunt duration: a moat that can sustain longer than expected. Sacerdote is hunting an inflection: a window of predictable explosive change. The framework has three parts:

“All technology adoption starts very slowly. It can be held back for a variety of reasons: high price, complex products, lack of an ecosystem. At some point, these barriers are removed, and the technology moves on the S-curve from the early adopter phase into the majority phase. At that point a massive wave of demand kicks in, and you can see three to four years of incredible unit growth. Everybody says tech is so unpredictable, but if you understand the way S-curves work, it actually can be quite predictable during certain time periods.” — Graham & Doddsville, Issue 25, 2015

Jobs did it to the smartphone with a $200 price point, a touchscreen a grandmother could use, and a 3G network. Flat-panel TVs went from 2 million units to 50 million in four years once the 40-inch screen hit $1,500 and Monday Night Football arrived in HD. Smartphones went from 1% to 50% penetration in five years.

“One of the reasons Warren Buffett has not liked tech is he found it to be so unpredictable. But this S-curve framework gives you insight into how big the market can be, where it might be in 4 or 5 years. It’s like a map into the future. And that’s why you want to invest in tech.” — Capital Allocators, EP.436

The map also tells you when to leave. Around 30–40% penetration, exponential becomes linear and the sell-side finally catches up. That is the sell signal, and it is why he exited Apple in 2012 — US smartphone penetration hit 50%, Android had stopped fragmenting, and “the under-appreciated earnings power had become appreciated.”

2. Moats matter.

“Growth investors will occasionally find an attractive S-curve, but the important piece really is competitive advantage and operational abilities. Finding that competitive advantage, understanding it, appreciating it before other people, and developing a more in-depth understanding of its strength are really important to us.” — Graham & Doddsville, 2015

The evidence he keeps returning to is a graveyard:

“Just about every e-commerce company I evaluated in the Fidelity report in the late 1990s is gone, except for Amazon. In smartphones, Apple created half a trillion dollars in wealth but HTC, RIM, Nokia, Motorola, and LG destroyed value.” — Graham & Doddsville, 2015

Being right about the S-curve and wrong about the winner is the most expensive mistake in technology. Which is why the second screen is so specific: network effects on the internet, standards in software, brand and logistics scale in e-commerce, and critical intellectual property in semiconductors including Qualcomm’s royalty and ASML’s lithography. We’ll get to AI.

3. Market can’t think exponentially.

“When you get the S-curve and a rising margin, your earnings don’t grow linearly, they grow exponentially. And it’s very hard for the world to think exponentially because very few things in our day-to-day lives do this. So if you can get those two things right, very often you can buy amazing companies at extremely low multiples. And it might look expensive on the first or second year P/E, but 3, 4, 5 years out, it might look crazily cheap.” — Capital Allocators, EP.436

An example:

“We were buying Tesla at literally 4 times the earnings they did 3 years later. And similarly, we loaded up on Nvidia right after ChatGPT in January 2023. We were buying Nvidia at 4 times the earnings they’re gonna do this year.” — Capital Allocators, EP.436

The filings back both up. Alpha-File dates the Tesla initiation to December 2019 and marks the round trip at 4.7× MOIC, a 162% IRR and +$1.8 billion of P&L on $489.8 million invested — Whale Rock’s single most profitable position ever. Nvidia was initiated in December 2022, at an average cost of $100 against $174 today: +$1.2 billion.

4. S-curves are dynamic

Sacerdote gives the example of Cloud:

“We looked at in the traditional world, that was $600bn of spend, and AWS was going after all of it… And then we said it’s gonna be 50% deflationary. So we sized it at $300bn… And what happened? It actually wasn’t deflationary. It was equal. So now we’re saying, we thought it was $300bn... it’s actually $600bn. We have another 5, maybe even 10 years to go on this.” — Capital Allocators, EP.436

And the discipline cuts the other way too.

“One S-curve that we got wrong, but we’re able to get out, was the EV S-curve… The idea was that 60% of cars were gonna be electric. And that’s happened in China. But when we hit about 10% in The US, it hit a wall. Tesla and the EV S-curve was amazing, and they had a great competitive advantage, and they had underappreciated earnings power. But then that S-curve went from being a 10 out of 10 to a 0 out of 10.” — Capital Allocators, EP.436

5. 2,500 meetings a year

“We’ve got a team of 10 of us, and we’re doing 2,500 face-to-face meetings with management teams a year… We call it the Whale Rock Learning Machine… Munger and Buffett talk about compounding knowledge. We’ve been compounding that knowledge for 20 years.” — Capital Allocators, EP.436

Ten people, 2,500–3,000 meetings — 250 meetings per person per year, one a working day, each. The method is explicitly borrowed: “The system we use is right out of Common Stocks and Uncommon Profits, which was written by Philip Fisher in the 1950s. It’s the scuttlebutt approach.” The continuity of the team matters with this approach: “Andrew and Michael have been with me for 18 years.”*

6. On the biggest companies in the world

Every allocator “knows” there is no alpha in mega-cap tech. Sacerdote thinks that consensus is the opportunity:

“With a small cap, it just takes one person to figure out it’s good and move it up. But it takes 100 people — 100 diversified PMs — to realize Google’s not a loser, it’s a winner. And can we figure that out before 95% of those generalist PMs? And we’ve been able to do it.” — Invest Like the Best, EP.477

Two Episodes: One Positive, One Negative

At the end of September 2021, Whale Rock’s disclosed US book peaked at $15.2 billion across 69 positions — a fund that had returned +71% in 2020 (the long-only sleeve, in its first year, +86.2%) and was closed to new money. Here is what the top of the book looked like at the top of the market:

Nine of the top ten were software or consumer internet. It is, in retrospect, a photograph of an era. The fund fell 9% in 2021 and then 42.7% in 2022 — worse than the Nasdaq, and enough to take the disclosed book from $15.2 billion to $2.98 billion by December 2022. That number is not just performance; it is redemptions and gross-down together.

ChatGPT launched on 30 November 2022. Sacerdote’s description of what happened inside the firm is the most operationally revealing passage in any of the three interviews:

“It’s just like the Netscape moment. And then four of us really spent 110% of our time just sprinting to learn everything we could possibly learn, meeting with private companies, public companies, mega caps, chip companies in Taiwan, in the US, software companies, venture capitalists, you name it, to put together the pieces and anticipate that we might not have all the answers, but we knew the questions to ask and where the fulcrum issues would be.” — Capital Allocators, EP.436

The conclusion they reached was a bet on sequence: the infrastructure layer inflects first, and it is the only layer where you can already identify the winner.

“Our thesis was invest in the infrastructure layer first, because that’s always the first to inflect on the S-curve. And also, no matter who wins above, we know we’re gonna need tremendous amounts of AI infrastructure… At the foundational level and at the application level, especially with startups, it’s very challenging to figure out. The ecosystem’s not baked. It’s a lot like Internet 1 when you had 7 or 8 search engines, and then later Google came out and crushed them all.” — Capital Allocators, EP.436

Nvidia was initiated in December 2022 — the same quarter the book bottomed. By March 2023 it was 6.4% of the fund, second only to Meta at 8.1%, in a portfolio that still contained Workday, HubSpot and Splunk. The fund returned +31.8% in 2023, +51% in 2024, and reopened for the first time since 2021.

“Every 7 to 10 years, there’s a significant sell off in tech… But each sell off has been followed by a major innovation cycle… When you’re at the top, just realize you’re not as smart as you think you are. When you’re at the bottom, you’re not as dumb as you think you are.“ — Capital Allocators, EP.436

Fourteen years of the disclosed book by position. The software era shows up as the broad band of Zendesk, Netease, Alphabet, Netflix and Shopify running 2015–2021; the 2022 collapse to a handful of names is visible as the trough right before Nvidia enters.

What Does Whale Rock Own Today?

Whale Rock runs a long/short fund, a long-only fund, a hybrid that can be up to 80% privates, and now a mega-cap fund — roughly $19 billion in total as of mid-2026. The 13F shows $12.46 billion across 35 positions at 30 June 2026: It does not show the short book. Privates are about 10% of capital, and Sacerdote says 55% of that sits in four names. What follows is a window onto perhaps two-thirds of the book.

The headline number moved a long way in three months: $7.74B (Q1) → $12.46B (Q2), up 61%, on an unchanged 35 names.

Average turnover is 84.2%, an implied hold of about 1.2 years — against Valley Forge’s 15.4 years. And Whale Rock has held 341 distinct investments across its filing history: “usually we’ll have one name at 10, and then we’ll have a handful at 7 or 5, and the top 10 will be 60%. We like to have 4 to 6 different S-curves.”

The full position history — every current holding plus the June quarter’s six exits, with weights, cost basis, MOIC and IRR:

The fundamentals view puts valuation and growth underneath the same 35 names — P/E, EV/EBITDA, five-year revenue growth, ROIC:

And the performance view isolates price return, TWR and IRR per position:

Valuation & Growth

Whale Rock has never been a “buy it cheap” book, and the 2019 trough (median near 19×) coincides with the pre-ChatGPT trough in the markets.

Growth has always been health with Median growth in the 20%+. You can clearly see the excesses in 2020/2021 when COVID and one-time pull forward of digitization led to a massive pull forward in growth, one the market was projecting out indefinitely followed by a rude awakening in 2022/23.

Below, I’ll summarize the current book.

The hyperscalers: Alphabet, Amazon, Meta (~13.7%)

Alphabet is 6.9% ($856M), trimmed 10% in the June quarter after being the largest position in March. Amazon went the other way — added 35% to 4.7% ($583M) — and Meta was cut 11% to 2.1%, though it still carries +$751 million of lifetime P&L across three and a half years. Microsoft, the stub that survived the March cull, is now gone entirely.

The hardware components sleeve: SanDisk, TTM, Corning, MKS, Celestica, Coherent, Fabrinet, SiTime, MACOM, Viavi, Lam, Western Digital, Seagate (~52%)

Nearly half of Whale Rock’s disclosed book sits in companies that spent the last two decades as commodity suppliers: memory, contract manufacturing, printed circuit boards, power supplies, optical components, laminates.

“For the past 40 years, nothing has changed in the data center… And there really was almost no growth in hardware for years and years and years. And the whole industry basically commoditized… Now you go to AI. The workloads are growing 10× every year, and they’re pushing every single aspect of this hardware to the physical limits of what it can do. Not only are you creating tremendous unit growth, but we call it the decommoditization of the hardware industry.” — Invest Like the Best, EP.477

Celestica is the paradigm case and the second-best trade in the fund’s history: +$1.5 billion, 4.40× MOIC, a 207% IRR over 2.7 years, against a cost basis of $48 versus $365 in June.

“Celestica was a contract manufacturer. This has been a disaster industry since 1999… But they hung on. Celestica’s heritage was IBM supercomputing, and it kept all that talent and skill. And then we noticed they were the sole supplier of the Google TPU server three years ago. The stock was trading at eight times earnings.“ — Invest Like the Best, EP.477

What turned a cheap assembler into a franchise was physics: an AI server is liquid-cooled, runs vastly hotter, and costs two to three hundred thousand dollars instead of five. “If it breaks, the whole thing goes down. So you become a commodity-like supplier to like selling a critical part on a plane. You’ll never get swapped out.” Add 50–60% share of cloud Ethernet switching and it stops being a contract manufacturer at all.

SanDisk is now the largest position in the fund at 10.9% ($1.36B), and the second-most spectacular gain in Whale Rock’s history behind only Tesla: a reconstructed +$1.7 billion at 8.2× MOIC on a position roughly six months old, cost basis $237 against $2,274 at the end of June. The stock was up 489% year-to-date and 3,375% over twelve months at that mark. Memory was the purest commodity in technology; high-bandwidth memory requires stacking ten dies with 10× the I/O and years of co-design with Nvidia, and the shortage did the rest.

It is also the position that cost him a fifth of the fund a month later. Note what the filings show: Whale Rock cut the SanDisk share count 20% in the March quarter and another 25% in June — trimming into strength through the entire move, exactly as the framework prescribes, and still finishing June with 10.9% of the fund in a stock that would fall about 30% in July.

The rest of the sleeve is a tour through the physical stack: TTM Technologies added 20% to 6.4% ($791M, +$500M, 2.2×) — “a regular server, you need 10 layers. These AI servers, you need a 40 layer, and there’s very few PCB suppliers that can make this”; Corning at 4.6%, now +$473M at 3.9× in a year, on fiber — “I was reading about this Microsoft data center they just built; there’s enough fiber to circle the world four and a half times in that one thing”; MKS 4.5%; Coherent added 25% to 3.9%; Fabrinet added 22% to 3.4%; SiTime 3.4%; MACOM 3.0%; Viavi doubled to 2.6%; Lam Research 2.6%.

And the disk drives came back. Western Digital sits at 2.1% (+136%, 2.4×, the stock up 550% in a year) and Seagate arrived new at 1.2%. Advanced Energy and Tower Semiconductor, both initiated in March, were sold outright in June — two of the fastest round trips in the book.

Advanced Energy Industries — initiated in March, sold outright in June, one of the fastest round trips in the book:

Nvidia — 0.6%

Here is the thing that jumps out of the position history. Whale Rock’s most famous AI trade is now the twenty-seventh largest position in a 35-name book: $75.5 million, 0.6%, 377,204 shares. Nvidia peaked at 8.5% of the fund; it was 2.3% in March; it is 0.6% now. Broadcom followed it down — 2.9% in March, 0.2% in June — and ARM was exited outright.

The lifetime record on Nvidia is still +$1.2 billion at 1.81× over three and a half years but reading between the lines, Sacerdote thinks Nvidia’s S-curve is well past the inflection, its earnings power is thoroughly appreciated, and the underappreciated earnings power has migrated down the stack to companies nobody has modelled. Perhaps this is prescient or perhaps this si Sacerdote selling Apple too soon in 2012.

Where the money went instead? AMD is now the second-largest position at 7.0% ($877M, +$534M in under three months). And Astera Labs arrived at 6.1% ($757M) as a brand-new top-five holding — the connectivity business that sits between the GPUs, and the one name where Sacerdote and Gavin Baker’s paths cross most literally: Atreides was in Astera’s Series C.

The application and consumer tail: AppLovin, Carvana, JFrog, Klaviyo, MongoDB, Duolingo (~10%)

AppLovin is down to 1.9%, cut 34% in the June quarter — but it remains the fund’s best multiple ever at 5.53× MOIC, +$1.1 billion, a 332% IRR, and it is now a falling stock, off 48% year-to-date. The story he tells about it is his answer to whether AI replaces analysts:

“Michael and Sam really figured out the AppLovin story before anybody. They followed it when it was private. They know all the competitors, they know all the intricacies of terminology. Sam went to the Las Vegas app advertising conference, and we went to Cannes, and talked to scores and scores of people… I don’t see AI doing that.“ — Invest Like the Best, EP.477

Carvana (2.3%, +$493M), JFrog (3.3%, now +$232M at 1.7×), Klaviyo (0.7%), MongoDB (0.6%) and Duolingo (0.4%) are what remains of the old book, and the tail is thinning fast. Roblox was exited outright in June.

The software short — and the quarter it started to turn

The single most important Whale Rock position was invisible to a 13F. In the March 2026 quarter, the fund that was 40–50% software five years ago was net short enterprise application software:

“Their AI products weren’t very good. Nobody could charge for them. We basically sold almost all our software. Entering this year, we were net short.” — Invest Like the Best, EP.477

Four reasons, and they compound: incumbent AI features aren’t moving the needle and can’t be charged for; CIO budgets are being redirected to tokens because the ROI is faster; annual price increases are no longer sticking; and hiring freezes shrink seat counts. His scoring rule is a modified Rule of 40 — AI as a percentage of sales, times market share — and “most software companies score 1 or 2.” HubSpot and Okta were gone in the March quarter; MongoDB was cut.

Then, in the June quarter, he started buying software again. Snowflake came in new at 1.8% ($225M) and Twilio new at 1.4% ($170M), alongside the existing JFrog stake at 3.3%. Three months after telling a podcast the fund was net short the category, the disclosed long book has three enterprise-software names totalling 6.5%.

SpaceX and Cerebras

Two IPOs in the June filing: Space Exploration Technologies (SpaceX), $26.1 million, 152,517 shares, and Cerebras Systems, Class A, $11.2 million, 50,696 shares. Together they are 0.3% of the disclosed book, both flagged NEW in the position history.

The private book you can’t see: Anthropic

Anthropic, Stripe, Databricks, Canva and Revolut — the names Sacerdote actually describes as roughly 10% of capital — stay completely invisible in a 13F, because they’re still private. Whale Rock’s largest conviction, by his own account in June 2026, isn’t in the 13F at all.

“Anthropic has critical intellectual property. They’ve maintained high market share in code. They’ve built a strong brand for enterprise. They’ve reached escape velocity and scale.” — Invest Like the Best, EP.477

The framework maps cleanly: the S-curve is enterprise AI, which he argues is “less than 1% penetrated — we call this an L curve, just straight up”; the competitive advantage is coding IP plus an emerging API ecosystem modelled on AWS circa 2013; and the earnings power is a coding market he sizes at “20 million coders in the world spending $20 to $30k a year on tokens — a half-trillion-dollar market” from that alone. His team built a 90-page deck using Claude Code to analyze the coding market and pitched it to Anthropic to win an allocation. He was scheduled to pitch Anthropic publicly at Sohn Montréal in June 2026 and had to pull it: the company had filed to go public and the quiet period intervened.

The Ghosts

Whale Rock has held 341 investments. The cumulative P&L reconstruction on Alpha-File — covering thirteen years of filings to June 2026 — is a map of eras.

• Tesla (Dec 2019 – Jun 2022) — +$1.8B, 4.73×. Still the largest realised gain in the fund’s history, exited because the curve flattened, not because the company did. SanDisk’s reconstructed +$1.71B is now second, ahead of Celestica (+$1.5B, 4.40×) and AppLovin (+$1.1B, 5.53×) — see the caveat above on why SanDisk still doesn’t show up in Alpha-File’s own ranked winners list.

• The SaaS cohort — Shopify +$689M, MongoDB +$604M, Bill.com +$591M, Datadog +$502M, Coupa +$439M, CrowdStrike +$365M, Twilio +$293M, Okta +$292M. Enormous winners — and the category he declared himself short of six months ago. Twilio is now back in the book, which tells you how quickly the ledger can flip.

• KE Holdings (−$290M, 0.44×, held three months) — the largest single loss, a Chinese property platform bought into the regulatory crackdown. Fast in, fast out.

• DoorDash −$138M, Monday.com −$121M, Robinhood −$113M, Marvell −$109M, Roku −$94M, Wayfair −$87M — the 2021 cohort, marked where the music stopped.

The pattern across the ghosts is the inverse of Hibbert’s. His exits aren’t businesses that stopped compounding — they are businesses whose curve matured while the business stayed fine.

+$24.3B of gross gains against $1.9B of losses over the trailing ten years. Tesla, SanDisk, Celestica, Nvidia and AppLovin are the five green blocks that paid for everything else.

What Happens Next

First, will we see software exposuring continue to increase? Snowflake and Twilio arrived new in June, JFrog was retained, and the disclosed software sleeve is back to 6.5%. Are these longs against the beginning of a reversal?

Second, does the hardware trade survive? The June book is the last clean picture before the fall. SanDisk closed the quarter at 10.9% of the fund, up 489% year-to-date and 3,375% over twelve months, with a reconstructed 8.2× MOIC. The September filing is the first one where we find out whether he adds into the drawdown or whether his own rule quietly triggers.

A note on the data: All portfolio data in this piece is sourced from Alpha-File.

A Harvard-trained doctor who never practiced, a McKinsey consultant who quit, a biotech VC who decided venture math was for people comfortable losing — and, since July 2007, the purest expression of “buy monopolies” playbook running anywhere at scale: seven stocks, four of them held every single quarter for nine years of filings, a 6.5% turnover rate: meet Dev Kantesaria.

The bloodline

The series so far has traced the TCI tree, the Tiger lines, the institutions, and — with Greenlea Lane — the Graham-Buffett-Columbia line. Valley Forge is that same Omaha bloodline run as the opposite experiment in scale: Dev Kantesaria started in July 2007 with reportedly about $300,000 of his own money, and the firm now manages billions from Miami:

“I wish that there were easier ways to make money... But for what works, and is tax-efficient long term — there’s no playbook better than the Buffett-Munger playbook.”

The overlap with this series’ oldest profile is literal, not stylistic: Moody’s, S&P Global, and Visa — the spine of Hohn’s book at $70 billion — are the spine of Valley Forge’s at $3.4 billion. Two shops, two continents, one conclusion: own the monopolistic businesses that tax the world’s debt issuance and payments, and do nothing.

The Investor: Eight Years Old, Then the Long Way Around

“I’ve loved equities since I was eight years old,” Kantesaria told ValueWalk in 2019 — and then took possibly the longest detour in this series to get back to them. MIT undergraduate. Harvard Medical School — “I was top of my class... and fancied myself becoming a renowned surgeon” — before concluding that he loved the intellectual side of medicine and none of the practice. McKinsey for two years. Then nearly two decades in venture capital: principal at TL Ventures, co-founder and general partner of Devon Park Bioventures, four to five hundred million dollars deployed into biotech startups, hundreds of CEOs, more than fifteen boards.

The venture years produced the founding rejection that defines the fund:

“The thing about venture capital investing is that there is a portfolio strategy there, where you’re hoping for one or two home runs on a portfolio of say 10 to 12 names. That didn’t resonate with me... I wanted my batting average to be nine out of ten. I wanted almost every one of my investments to be a winner, and not to be subject to the luck factor.”

What VC did teach him: distilling regulatory risk, patent risk, and management risk “into a single pre-money valuation,” and the discipline of underwriting the next ten years rather than admiring the last ten. Valley Forge launched July 1, 2007 — thirteen months before Lehman. The tuition: −5.1% in the 2007 stub, −20.3% in 2008 (against −37.0% for the S&P), then the answer: +69.5% in 2009, +22.7% in 2011 when the index did 2.1%, and a cumulative 1,088.9% net through 2024 — 15.2% annualized against the S&P’s 10.2%, with a 1%/20% fee load in between before hitting another stumble: a reported -2.8% in 2025 and -23% YTD 2026. A short stumble or are the monopolies Valley Forge and TCI relied on to generate their returns permanently impaired?

Philosophy & Process

1. Pricing power is the hallmark — and the book is the proof.

“Pricing power is the hallmark of a great business. If you can raise your prices above the rate of inflation consistently, you have a phenomenal business model.”

While Tarasoff preaches untapped pricing power, almost an unwillingness to use pricing, Kantesaria targets businesses that can leverage their power to the maximum extent. On the rating agencies: “Effectively, there is no way to undercut Moody’s or S&P on price” — issuers would happily pay 30-50 basis points more, because the rating unlocks the debt sale. The FICO thesis (below) executed at 24% of the fund (which is biting them today) is similar.

2. The fishing pond has fifty fish, and predictability is the water.

Kantesaria’s universe, per his own description relayed across interviews: roughly fifty companies in the world — monopolies, duopolies, the occasional oligopoly — with quality pre-determined and price the only variable. (Hibbert, inside BlackRock, arrived at the same number: “clearly no more than 50 companies... ever.”.)

“We hate to lose money. It pains me to lose a single dollar... Predictability is not something that we sacrifice at Valley Forge. We’re happy to give up some growth in exchange for predictability.”

Anything that could plausibly drop 90% is “just not our fishing pond.” The criteria from the firm’s own materials read like a checklist for the same fifty names: capital-light, minimal R&D risk (”if you must spend $1 billion to find the next blockbuster drug, the ROI isn’t predictable”), operating leverage, flexible cost structures, earnings insulated from commodities and macro — companies that are “a small but extremely valuable part of the industry supply chain.”

3. Entries come from neglect more than panic.

The mechanics: new positions at a 30-60% discount to intrinsic value, and — the surprising split — only about a third of entries come from misunderstood sell-offs. Two-thirds come from what he calls “market neglect”: great businesses the market simply ignored — the rating agencies in 2009-10, Mastercard around 2011, FICO in 2018. He calls himself a “business historian”: go back to the founding, study every difficult period, so that when the storm comes you already know how the ship handles.

4. Valuation precision is a silly exercise; market timing is a worse one.

“Getting worked up about whether a company has a P/E of 24, 28, or 32 is far less important than making the right decisions about what companies you buy.”

A true compounding machine’s intrinsic-value growth “overcomes whether you’ve overpaid by 10 or 15 percent” — the exact asymmetry Chit Chat’s hosts summarized as disciplined on the entry, lenient while you own it. Asked in late 2024 whether Buffett’s cash pile argued for selling: “For us, market timing has been a poor way to generate returns over the long term.” Forward multiples he treats as actively misleading for his kind of holding — the sell-side only models FICO’s mortgage-volume recovery and price increases “when they happen.”

5. The renounced temptation: Big Tech, and the AI heresy.

Valley Forge owns none of the Mag 7, and Kantesaria is candid that this is his great error of omission — “I knew that Google Search was a powerful business model... That’s certainly a big omission for me” — rooted in a real objection: dual-class structures and “generally poor capital allocators.” But the forward-looking version of the stance is the most contrarian AI view in this series:

“It’s our view that AI becomes commoditized, and then it becomes difficult for these companies to monetize their AI offerings... R&D and capex are going up exponentially because these companies are fighting each other ruthlessly to stay ahead... I think it will have a bad ending for most people, because it’s very hard to predict the one or two companies that are the real winners.”

Put that next to Avala’s 50% AI book and Lone Pine’s trillion-dollar-capex framing and you have the series’ full spectrum in three managers. The nuance: he expects the Nvidias and Googles “to do well for the next few years” — his objection is the decade, not the quarter. And note the one exception that proves every rule in his framework: ASML, first disclosed in Q2 2023 and subsequently scaled — a literal monopoly, pricing power, a small-but-vital toll on the world’s most important supply chain. The only AI exposure he’d take is the one that fits the fifty-fish pond.

6. Errors of omission outrank errors of commission.

His favorite cautionary tale: the investor who passed on the Microsoft IPO at 35x earnings to buy a washing-machine maker at 12x — “What a tragic mistake that was.” And on over-learning from losses: people treat a burn the way Depression survivors treated banks, with blanket bans instead of nuanced lessons. The fund materials’ sell discipline is correspondingly narrow: sell at 100% of intrinsic value, on deteriorating long-term quality, on capital-allocation sins, on excessive leverage, or for a superior opportunity — nothing about price targets, macro, or momentum.

The Record, and What It Cost

The brutal 2007 start meant −5.1% in the 2007 stub, −20.3% in 2008 (against −37.0% for the S&P), then +69.5% in 2009, +22.7% in 2011 when the index did 2.1%, and a cumulative 1,088.9% net through 2024 — 15.2% annualized against the S&P’s 10.2% before hitting another stumble: a reported −2.8% in 2025 and −23% through Q1 2026. Compounded onto the year-end 2024 record, those declines reduce the cumulative gain to roughly 790%, or about 12.4% annualized since the July 2007 launch through March 2026.

Extending the S&P comparison to the same endpoint produces an estimated 10.2% annualized benchmark return, or roughly a 517% cumulative gain, using the index’s 17.9% total return in 2025 and −4.3% total return in Q1 2026. Valley Forge therefore still leads over the full period—about 12.4% versus 10.2% annualized, and 790% versus 517% cumulative—but much of the comfortable margin visible at year-end 2024 has narrowed.

The recent reversal is severe. From the start of 2025 through Q1 2026, Valley Forge lost approximately 25.2% compounded, while the S&P 500 gained roughly 12.8%, a relative shortfall of about 38 percentage points in fifteen months. The fund’s −23% first quarter was worse than its −20.3% loss in the full 2008 calendar year, although the periods are not directly comparable. After nearly two decades of beating the market through concentration and patience, Valley Forge is now enduring one of the largest performance struggles in its history—the same handful of monopolies that created the long-term record are also making the current drawdown unusually concentrated.

The portfolio record makes the philosophy visible. Valley Forge has never diversified its way out of conviction: the same four toll roads—Moody’s, S&P Global, Mastercard and Visa—formed most of the disclosed book from the first 13F, while FICO gradually became the dominant fifth position. The narrow bands at the top are brief auditions.

The disclosed US book grew from well under $100 million in the first filing to $3.4 billion in Q1 2026.

What Does Valley Forge Own Today?

The Alpha-File record: first 13F Q4 2016 (the fund ran below the threshold for its first nine years), 38 quarters, and — the number that defines the firm — 17 lifetime positions ever, with average turnover of 6.5%, an implied hold of 15.4 years. Avala turns its book in 1.4 years; Greenlea in 2.6; Valley Forge in a decade and a half. A true punch card investor with 3 holes left to punch.

The Q1 2026 book is seven names and $3.4 billion: FICO 24.3%, S&P Global 22.2%, Mastercard 22.0%, Moody’s 17.0%, ASML 6.7%, Visa 5.4%, and Intuit 2.4%. S&P Global, Mastercard, Moody’s, Visa and Intuit have appeared in every filing since Q4 2016. The reconstructed lifetime winners are FICO (+$774M, 3.2×), Moody’s (+$263M, 1.7×), Mastercard (+$254M), S&P Global (+$214M), and Visa (+$116M).

The corrected position history now carries both holdings continuously through the latest filing: FICO from Q2 2018 and ASML from Q2 2023. Neither is a Q1 2026 initiation. That matters for the economics as well as the labels: FICO’s reconstructed cost basis is roughly $366 versus a Q1 implied price of $1,068, while ASML’s is about $922 versus $1,321.

Q1 2026’s activity was unusually decisive for a 6.5%-turnover manager: Valley Forge exited the two smallest experiments, MSCI and Equifax, added to ASML, trimmed Visa and Intuit, and left FICO untouched.

The fundamentals tables put company-level context underneath the weights. They show a portfolio of high-margin, capital-light businesses, while the performance view makes clear how much of the reconstructed gain is concentrated in the same handful of long-held positions.

Valley Forge’s weighted trailing P/E fell from 59.5× in Q4 2025 to 33.7× in Q1 2026 as these “monopolies” fell in part due to fears of AI disruption. The median fell from 51.4× to 31.9×. With ASML now incorporated, both quarterly observations cover 100% of disclosed value—nine companies in Q4 and all seven current holdings in Q1. That makes the compression genuine rather than a coverage artifact and leaves the book at one of its lowest multiples since Valley Forge started reporting 13Fs.

The growth chart supplies the counterweight. Weighted rolling five-year revenue CAGR eased only modestly, from 10.9% to 10.8%, while the company median rose from 12.2% to 12.9%. The latest reading also covers 100% of disclosed value across all seven holdings. In other words, the quarter’s dramatic multiple compression was not accompanied by a comparable collapse in the portfolio’s backward-looking revenue growth. But the chart also clarifies what Valley Forge is—and is not—underwriting: this is not a hypergrowth portfolio. The return case depends on durable pricing power, margin expansion, capital-light reinvestment and buybacks converting roughly low-double-digit revenue growth into faster per-share value creation.

FICO — 24.3%, the thesis of the firm. The US credit-score standard: network effects, switching costs, brand, and regulatory entrenchment since Fannie and Freddie mandated FICO scores on mortgages in 1995. The 2018 entry was pure market neglect at ~30x EV/EBIT: the tell was that FICO hadn’t raised the score’s price in roughly 25 years. The harvest since: mortgage-score pricing from ~$1 toward $4.95, +100% in 2023, +40% more the year after, margins tripling. The position ran to roughly 40% of the fund at FICO’s ~80x peak — and then 2025 delivered the stress test: political pressure over the price increases and a ~42% drawdown. His response, per the latest filings: he added nothing, trimmed nothing material, and let the weight ride. Tarasoff’s Ambassadors Group warning (aggressive price harvesting “using up whatever finite amount of pricing power they have... and I’m worried that it’s ending”) is the bear case; Kantesaria’s mortgage math (score cost still trivial against a $1 million mortgage) is the bull case.

The rating-agency/index toll booth — S&P Global + Moody’s, 39.2% combined. The duopoly that prices the world’s debt, held since the first filing. “There is no way to undercut Moody’s or S&P on price.” TCI cross-holds both sides of this trade.

The payments rails — Mastercard 22.0% + Visa 5.4%. Fees that scale automatically with nominal spending—inflation protection as a business model. Note the quiet relative call: Visa was cut from 7.3% to 5.4% in Q1 2026 while Mastercard remained untouched at 22%.

ASML — 6.7%, the exception. The EUV monopoly, initiated in Q2 2023 and added to in Q1 2026 — his single concession to the AI build-out, made through the one name where the moat is physics and the customer has no alternative. The corrected chart now shows the position continuously from initiation rather than treating the latest increase as a new holding.

Intuit — 2.4%, the smallest survivor. Intuit has appeared in every Valley Forge 13F since the first filing, but the position has been steadily cut and fell another 15% in Q1 2026. Its shrinking weight says it no longer competes with the core toll roads for capital.

The recent auditions. MSCI and Equifax were both held for about a year and exited in Q1 2026 at small reconstructed losses. Their brevity matters in a portfolio whose core positions have lasted nine years: Valley Forge was willing to admit quickly that neither deserved a permanent seat.

The tuition ledger. Seventeen positions ever, and the losses tell you more than the wins. Amazon lost an estimated $141M at a 0.49× MOIC over 4.5 years—the one time the fund appears to have chased a great business model and sold at roughly half its reconstructed cost. Adobe lost an estimated $25M. Autodesk, Monster, Meta and Nike were much smaller outcomes. Aspen Technology, the software-industrial Valley Forge owned in two separately mapped stints, made money both times.

The cumulative P&L reconstruction shows how thoroughly a few winners paid for the tuition. FICO, Moody’s, Mastercard and S&P Global dominate the positive side, while Amazon accounts for most of the loss pool. This is not a fund return series: it is a quarterly-filing reconstruction that cannot see intra-quarter trades, cash, shorts or undisclosed foreign positions.

What Happens Next

FICO. A quarter of the fund, freshly repriced by politics and AI fears. If the mortgage cycle turns and volumes recover onto the raised prices, the operating-leverage math he described makes the position look prophetic; if Washington decides the score’s price is the product’s vulnerability or if AI replaces the need for credit scoring, Valley Forge becomes the series’ cleanest test of what happens when a pricing-power thesis meets its regulator. He is on record that the sell trigger is deteriorating long-term quality — watch whether the filings ever blink.

Seventeen positions in nineteen years is the slowest hand in this series, and the recent cadence — ASML scaled, MSCI and Equifax tried and released within a year — shows the filter working in real time: two candidates failed the audition; one monopoly earned more capital. The question his own framework poses: when the fifty-fish pond’s prices all reflect their quality, does discipline mean waiting years for the next market neglect, or does the definition of the pond quietly widen?

Check out alpha-file.com for detailed charts and analysis for every single fund.

Scottish Mortgage is Britain’s largest and best-known investment trust. At 31 March 2026 it had £15.4 billion of total assets, an ongoing charge of 0.33%, gearing of 11%, and 80.7% of assets in its top thirty holdings. Its largest holding, SpaceX, was 19.3%.

The managers are Tom Slater, who joined Baillie Gifford in 2000 and succeeded James Anderson as lead manager in April 2022, and Lawrence Burns, deputy since March 2021. Slater inherited the flagship twelve months after its peak and directly into the great growth-stock drawdown. Anderson himself took over Scottish Mortgage in April 2000, at almost exactly the top of the dot-com bubble.

Philosophy & Process

Scottish Mortgage is a listed company, not an open-ended mutual fund. Shareholders can sell their shares to somebody else, but cannot redeem capital from the trust. The managers therefore have a form of permanent capital (with a caveat). They can borrow modestly, hold private companies through long gestation periods, and let a position become large through appreciation.

The share price can still trade below net asset value, potentially creating pressure to do buybacks or from activists seeking to take control and arbitrage the spread. But the capital itself is permanent. That is why the trust became the purest public expression of Baillie Gifford’s Long Term Global Growth philosophy as the capital base is truly long-term.

1. Start with the shape of returns

Baillie Gifford’s study of global equities from 1984 to 2013 found that roughly 20% of companies created half the return, while only 5% of stocks rose at least fivefold over five years.

That changes the objective. A conventional process tries to be right often and avoid embarrassment. Scottish Mortgage tries not to miss the tiny group of companies capable of returning ten, twenty or fifty times the original investment.

James Anderson’s own conclusion was that his greatest mistakes were omissions and premature sales, not holdings that went to zero. Asked when investors should sell, he answered: “very, very few” times. If downside is capped at the amount invested and upside is not, research should be directed at what can go extraordinarily right.

Tom Slater, in the FY2026 Manager’s Review, states:

“Academic research has shown consistently that the majority of wealth creation in equity markets comes from a very small number of companies. Most stocks underperform cash over their lifetimes. The entire excess return of the equity market is generated by the outliers.”

“The implication is that if your portfolio does not own the outliers, or sells them too early, you will almost certainly underperform. Our approach is designed around this reality. We concentrate the portfolio in businesses we believe have the potential to be exceptional, we hold them for years rather than quarters, and we accept the volatility that comes with that conviction.”

2. Research the destination

The relevant questions are long-duration ones. How large can the market become? Does the advantage strengthen with scale? Can management keep reallocating capital as the original product matures? Can the culture survive success?

This is why the portfolio can look optically expensive. A near-term multiple is useful when the future resembles the present. It is less useful when a company’s business mix and competitive position may be unrecognizable in five years.

The same idea explains its hostility to automatic mean reversion. During technological change, yesterday’s average can describe an obsolete industry. The task is not to wait for a great company to become statistically average, but to decide whether the market still underestimates the length and scale of its growth.

Baillie Gifford is not indifferent to price. It is suspicious of false precision applied to nonlinear businesses.

3. People are part of the moat

Long horizons move management quality from a soft factor to a central variable. If the case depends on choices not yet made - new products, geographies, acquisitions, capacity and regulation - the forecast is partly a forecast of the people making them.

SpaceX is not merely a launch-cost curve; it is a judgement about a company that repeatedly converts engineering capabilities into new markets including datacenters and potentially orbital ones. MercadoLibre is not merely Latin American ecommerce; it is a management system that moved from marketplace to payments, logistics, credit and advertising.

4. Concentration is the output

Scottish Mortgage does not begin by deciding that one stock should be 20%. It starts with smaller positions and lets the return distribution decide which become large. Concentration is partly the residue of success, and trimming winners too early would destroy the skew the process is designed to capture.

The trust did not buy a 19.3% SpaceX position. It invested £151 million between 2018 and 2021 and then stopped. By March 2026 the stake was worth £2.98 billion, roughly 19.7 times invested capital. The position became a fifth of the trust because the thesis worked faster than the rest of the portfolio.

5. Accept a low batting average

An outlier portfolio should contain failures. The relevant scorecard is value created by the winners minus capital lost in failures. Tesla, Amazon, NVIDIA, MercadoLibre and SpaceX can pay for many Northvolts. The Swedish battery maker collapsed into bankruptcy after production delays, cost overruns and a failed effort to secure enough new financing.

That does not excuse poor diligence. It sets the economic standard for judging it. The emotional problem is that failures arrive early and visibly, while the great winner may need ten years.

6. Sell on opportunity cost

A holding can leave because the advantage weakened, execution broke, the market was smaller than expected, management changed, or another idea offered materially better upside.

Scottish Mortgage has one additional category: sales made to manage the trust itself. Its £3.1 billion buyback could only be financed from listed securities. Some trades therefore say less about Amazon or ASML than about the board’s attempt to close the discount. Portfolio construction and capital structure became entangled.

The record

Total returns to 31 March 2026, from the trust’s annual report:

Across the full manager era from April 2000, Scottish Mortgage’s share-price total return was approximately 1,775%, against roughly 578% for the FTSE All-World. We are starting to see a recovery from what was clearly a tough period. FY2026 was the best year since the boom: NAV rose 27.4% against the index’s 18.0%. The three-year record moved back above the index, 57.9% against 50.5%. Over ten years, Scottish Mortgage’s record remains strong as it compounded at 18.3% against 12.8% for the FTSE All-World. The ten-year result shows what a handful of outliers can do. The five-year result shows the price of owning the tail when markets reprice duration all at once.

*The chart converts the trust’s official NAV total-return index into cumulative profit or loss on £100 invested.

What does Scottish Mortgage own today?

Below is a broad look at the portfolio’s evolution over time.

To keep the portfolio section useful, the discussion below concentrates on positions around 1% of assets or larger.

The overarching theme is that, in order to achieve outlier returns, you need an extremely large TAM and in almost all cases, these TAMs are best-attacked through platforms. Baillie Gifford is in most cases investing into a horizontal layer rather than a vertical niche and trying to buy the winner in that layer.

The private book, summarized

At 31 March 2026 Scottish Mortgage held 53 private companies equal to 41.5% of total assets. Scottish Mortgages is the only real cross-over fund a retail investor can access. The top five positions represented 75.7% of private exposure and 31.4% of the trust.

Since 2012 the trust has deployed £4.9 billion into private companies. The outcomes are 40 listings, eight takeovers, one liquidation and 53 still held.

SpaceX

SpaceX contributed 14.9 percentage points of Scottish Mortgage’s 27.4% FY2026 NAV return - more than every other holding combined. Its value rose 178.6% to 19.3% of total assets.

The managers describe SpaceX as a “dual monopoly” in launch and global connectivity, with datacenters now the third (and potentially biggest) leg with Cursor/xAI now a credible fourth leg, all four of which are potentially huge TAMs.

SpaceX can be viewed as the ultimate expression of the archetype Baillie Gifford is looking for: platforms attacking a huge TAM with a management team that can take this into adjacent markets: none potentially bigger than orbital datacenters.

SpaceX was still private at the March year-end, but completed its NASDAQ IPO on 12 June 2026. Its first-day close lifted the position to approximately 25% of the portfolio on 15 June.

TSMC, ASML, and NVIDIA

Baillie Gifford positions in the AI semiconductor value chain is anchored to players dominating their part of the massive value chain.

TSMC (5.7% of assets), ASML (3.2%) and NVIDIA (3.2%) are all dominant: the foundry, the lithography monopoly and the accelerated-compute platform.

Lawrence Burns compressed the ASML case into one line on the trust’s own podcast (S4E4, with ASML CEO Christophe Fouquet, July 2026) and something similar can be said of TSMC and NVIDIA:

“Whatever happens… ASML wins. It’s an agnostic royalty on AI demand, and more broadly, compute demand.”​

AI Platforms

Scottish Mortgage’s AI portfolio is broader than the tech mega caps. New private holding Anthropic was £171m, 1.1% of assets at year-end; Databricks £226m, 1.4%; Cloudflare £248m, 1.6%; Snowflake £144m.

The common thread is infrastructure. Anthropic supplies intelligence. Databricks and Snowflake organize governed enterprise data. Cloudflare provides network and security. Rather than make a bet on a vertical or niche, the managers are trying to make bets on the horizontal platform or rails that agents will need.

This theme is where private company access is particularly beneficial. Privates does not mean small. By early 2026 Databricks had a $5.4 billion revenue run-rate, was growing above 65%, and generated positive free cash flow. Anthropic’s annualized revenue reached $30 billion by April 2026 (with press reports now claiming $100bn+ exiting 2026).

E-Commerce Platforms

Baillie Gifford’s pattern is a founder-led platform that uses one successful product to enter adjacent markets and it applies to e-commerce too.

MercadoLibre (4.0% of assets) moved from marketplace to logistics, payments, credit and advertising. Amazon (3.6%) started with e-commerce but expanded to cloud computing and a third-party marketplace. Shopify (2.0%) is turning merchant software into an AI-enabled operating system.

Stripe, still private at year-end, deserves explicit sizing: at £609 million it is 4.0% of assets — the trust’s fifth-largest holding. It processed $1.9 trillion in 2025 and is expanding from payments into stablecoins and agentic commerce, a business that is larger than public peers but is only accessible in the privates market.

Meta (£415m, 2.7%) and Spotify (£285m, 1.8%) round out the platforms with Shopify. Nu (£204m, 1.3%) is a smaller but revealing addition. Scottish Mortgage has kept adding to the position as Nu as profitably scaled across Brazil, Mexico and Colombia.

Asia & China

China was a deliberate retreat that has since become a selective return. Taken together, the Asia/China complex is roughly 13% of assets across seven-plus names.

ByteDance (4.7% — the third-largest holding in the trust) is the anchor, and still private at year-end. It generated quarterly revenue above $48 billion in Q2 2025 and was valued at a fraction of Meta despite a comparable revenue base — the widest public-private valuation gap in the portfolio.

CATL (1.6%) extends the AI compute thesis from compute to electrification. It was one of FY2026’s largest new positions. The team emphasized its manufacturing lead and years of sustained investment, not simply Chinese EV demand.

BYD (£204m, 1.3%) benefits from vertical integration and engineering scale to the point where traditional auto OEMs struggle to be in the same ballpark in terms of price competitiveness.

Sea (£221m, 1.4%) was increased, but the managers conceded those purchases were poorly timed as Temu and TikTok Shop subsidised aggressively in Southeast Asia.

PDD (£315m, 2.0%) is the survivor of the old China-internet book, and the two FY2026 additions — MiniMax (0.5%) and RedNote (0.4%) — point at AI models and newer consumer platforms.

Brands and transport

Ferrari (1.9%) and Moderna (2.0%) were each close to 2% of assets; Aurora (£146m, 1.0%) and Tempus (£158m, 1.0%) roughly 1% each.

Ferrari protects scarcity rather than maximizing volume. After a capital-markets day triggered its worst trading day as a listed company, management had chosen brand protection over the growth path investors expected. Scottish Mortgage called that the “right decision for the business.” It is a useful example of the trust siding with management’s appropriate long-term actions rather than push for higher volumes to beat near-term analyst estimates. Ferrari has returned roughly eleven times Scottish Mortgage’s investment over a decade.

Aurora launched commercial driverless freight between Dallas and Houston and passed 100,000 autonomous miles without a safety incident. Private Zipline (£304m, 2.0%) reached commercial scale, passed two million deliveries and expanded its US network. Neither is yet producing revenue commensurate with the eventual opportunity.

The discount

The managers run the investment portfolio but the board runs a “second portfolio”: the trust’s own capital.

In May 2023 the Trust’s discount to NAV reached 23%. In March 2024 the board pledged at least £1 billion of buybacks over two years, then went much further. By March 2026 it had repurchased £3.10 billion. The discount ended FY2026 at 9.5%; after year-end the shares moved to a premium and issuance resumed.

Buybacks below NAV create value for remaining shareholders, but they also force asset sales and alter portfolio weights. The board’s discount policy is therefore part of the investment process and may drive higher weights in less liquid privates during periods of particularly large discounts.

Privates ratcheted from 11.8% (2016) to ~28% (2023–25) — then jumped to 41.5% in FY2026, a move that is mostly the SpaceX/xAI re-mark. The discount panel shows why the board acted: a decade of premium/par ended in 2021, blew out to −19.6% at March 2023 (−23% intra-year), and the £3.1bn buyback walked it back to single digits. The deepest discount coincided with the private book’s peak opacity, and buying back stock could only be funded from the listed sleeve — which mechanically pushed the private percentage higher still.

The Ghosts

Baillie Gifford’s portfolio has two kinds of ghosts: failures that consumed capital and winners sold before the full outcome, the latter being far more painful.

Baillie Gifford looks to have entirely exited niche e-commerce platforms: Wayfair was the complete exit in FY2026; Zoom, HelloFresh and Zalando already gone, likely a shrewd move given the business has massive pull-forward from COVID.

Tesla is the opposite ghost. Scottish Mortgage first bought it in 2013 at roughly $6 a share, held through years when the position looked indefensible, realized extraordinary gains, and then reduced it as the company matured and dominated the trust.

Whatever one thinks of the exit timing, Tesla remains the canonical proof that tolerance for looking wrong can create the right asymmetry. (A residual £76m, 0.5% position remained at 31 March 2026 — the ghost is not fully gone.).

Scottish Mortgage’s Tesla cost basis was under £100m (first purchases 2013 at a split-adjusted ~$6). The position peaked around 11–13% of the trust (~£1.4bn) in late 2020 — roughly 28–30 million shares on today’s split-adjusted count. The FY2021 report discloses “we sold around 80% of our Tesla shares” in that single year, and James Anderson put the 2020 profit alone at $5.5 billion; cumulative sale proceeds across 2020–2025 work out to roughly £4.3–5.0 billion of realized gains (estimated from reported year-end values and share counts).

The counterfactual: the never-sold peak position would be worth about £7 billion at mid-2026 prices — nearly half of today’s trust. Net “left on the table” after crediting what was realized: on the order of £2–3 billion, before crediting what the proceeds were reinvested into (a meaningful share of which became the SpaceX, semiconductor and AI-infrastructure book — so the true opportunity cost is possibly negative).

China is a partial reversal rather than an exit. The trust had sold Alibaba and Tencent and deliberately reduced exposure. In FY2026 it added CATL, MiniMax and RedNote while retaining PDD, Meituan and ByteDance. The new exposure favours companies that are better positioned for AI and perhaps better position in the eyes of regulators and the CCP.

What happens next

What happens to SpaceX? SpaceX is now listed and, immediately after the IPO, roughly a quarter of the portfolio. However, SpaceX is also first innings on datacenters and potentially orbital datacenters which are far larger TAMs to what Baillie Gifford would have initially underwrote to. How large will they tolerate that position getting to?

Long-term, can Slater and Burns produce their own outliers? Anderson’s generation supplied Tesla, Amazon and the original private-company push. The current team is navigating a disruption for the ages, and looks to be building a more layered portfolio around semiconductors, AI infrastructure, autonomy and financial rails. How will the next generation fare?

A hundred and seventeen years ago Scottish Mortgage financed assets conventional lenders would not touch, with a conversion option if the future proved better than expected. The modern trust does the same thing in a different vocabulary. It supplies patient capital where the market’s horizon is short, accepts that many forecasts will fail, and owns the option on a few outcomes becoming much larger than anyone can model.

This one is different. Baillie Gifford is an Edinburgh partnership owned entirely by its working partners, with no external shareholders or listed parent. Its flagship is not a fund but a company: Scottish Mortgage Investment Trust, a London-listed vehicle whose largest position by a wide margin is SpaceX.

The firm and the trust have been intertwined for more than 116 years. This piece describes the house — where it came from, what it believes, and the crises that shaped it, especially the near-death experience of 1978. Part 2 will examine the current portfolio and what today’s Scottish Mortgage Investment Trust looks like.

A rubber planter walks into a solicitor’s office

In December 1908 a young Scots rubber planter named Alistair McGregor came home on leave from Malaya with a financing problem. Credit was scarce after the Panic of 1907, and planters could not borrow against their estates. The man who listened was his family solicitor: the 27-year-old T. J. Carlyle Gifford, junior partner in the Edinburgh legal firm Baillie & Gifford WS, founded that year with Colonel Augustus Baillie — Boer War veteran two decades his senior.

The Ford Model T had just launched. Every car needed tyres, every tyre needed rubber, and every rubber estate needed capital. The structure Baillie & Gifford WS floated was a mortgage secured on the estates, with one addition that turned a lending vehicle into an investment company: the right to convert the loan into shares of the estate. Downside secured by the mortgage; upside captured by the conversion option.

Convertible debentures were already an established tool of company finance by 1909. What was unusual was pointing that structure at an asset nobody else would finance — half-planted colonial rubber estates years away from their first harvest — through a dedicated, publicly floated Scottish trust.

The prospectus described the risk in language the firm could still use today: “This method safeguards the investor from the risk of loss of capital to which he is exposed by investment in the ordinary shares of rubber companies,” while conversion offered “an interest in the plantation rubber industry which will yield large returns, even if the price of rubber falls 50%.” Protect the downside, buy the option on the upside.

The Straits Mortgage and Trust Company floated in 1909 with subscribed capital of just over £50,000. It diversified almost immediately — the Cuban Telegraph Company, the Chilean Northern Railway, and in 1910 its first US investment, the Tacoma Land Investment Company of Washington State. In 1913 it was renamed The Scottish Mortgage and Trust Limited, the company now known as Scottish Mortgage Investment Trust. In 1927 the partners created Baillie Gifford & Co to manage the trusts separately from the legal practice. Two years later, the Wall Street Crash happened.

The crash was severe, but it was not a near-death experience for Scottish Mortgage. Reserves accumulated during the 1920s helped the trust weather the Depression, although it cut its dividend in 1933 — the only reduction in its history. Baillie Gifford actually expanded amid the failures around it, taking over the management of three trusts and opening its first London office.

One historical aside. In November 1939 the Bank of England sent Carlyle Gifford to New York to sell an estimated $4.4 billion of British-held dollar securities to fund the war effort. Before Lend-Lease, Britain paid for the war partly by liquidating its US share portfolio, and Carlyle Gifford was the man the Bank of England chose to value and negotiate that portfolio.

1978: the year the firm was worth nothing

The modern Baillie Gifford is incomprehensible without the decade that nearly killed it — and here Richard Burns’s centenary history, written from inside (he joined in 1973 and served as joint senior partner), is unsparing.

An investment trust is a listed company that owns a portfolio of shares. Investors buy the trust’s shares on the stock exchange, and the trust hires a management firm to run the portfolio for an annual fee. Baillie Gifford was that hired manager — and in the 1970s, its handful of trusts supplied roughly 90% of the firm’s income. The trap in the structure: the trust’s share price can drift below the value of the portfolio it owns. When the discount gets wide enough, a predator can buy the whole trust on the stock market, take control of the portfolio, and pocket the difference — firing the manager on the way out.

The 1965 Finance Act introduced capital gains tax and complicated the administration of investment trusts, contributing to a difficult environment for the sector. Discounts blew out — from under 10% in the late 1960s to over 30% by 1977. Meanwhile UK inflation passed 20% and the market fell over 70% in nominal terms from May 1972 to December 1974, roughly 80% after inflation.

With ~90% of the business in trusts whose market had fallen by more than two-thirds, the firm’s funds under management had shrunk with it: at the end of 1974 they were £179.5 million — in total. Performance, Burns concedes, was poor. Marketing did not exist. The firm was, in his word, inward-looking.

Then the discount invited a predator. In August 1977 the British Rail Pension Fund — flush with nationalized-industry cash — made a takeover bid for Edinburgh & Dundee Investment Trust. The pension offered to buy every share. Edinburgh & Dundee was Baillie Gifford’s second-largest client, whose annual fee — though nobody outside knew it — was bigger than the firm’s entire annual profit. Losing this one client would tip the whole firm loss-making. Baillie Gifford fought and lost and in the spring of 1978, the client was gone.

After Edinburgh & Dundee was lost, Baillie Gifford still managed three investment-trust clients. Its survival, however, depended overwhelmingly on the two most important: Scottish Mortgage and Monks, another London-listed trust the firm had managed since 1931. Both still traded at wide discounts. If a predator did to either of them what Railway Pension had just done to Edinburgh & Dundee, Baillie Gifford would not be viable. Had the senior partners been paying themselves market salaries, the firm would have been loss-making. They were not: they were cutting their own profit shares so that junior partners with young families could have a modest raise.

What saved them was Edinburgh. Not the city government, but the network of professionals that encompassed the city. The boards of all three remaining trusts voluntarily raised the below-market fees they paid Baillie Gifford, explicitly so that no investment staff would have to be let go.

Scottish Mortgage’s chairman at the time, Rob McGregor, was the grandson of the McGregor who had brought the firm its first client in 1908, and Carlyle Gifford’s godson.

The Widows Fund of the WS Society — the pension charity of the very solicitors’ guild the founders had belonged to — handed the firm its first new mandate in years, steered there by a senior partner of the successor law firm.

The Church of Scotland followed with an income fund, guided by a friendly investment-committee chairman who ran a neighbouring trust. None of these were open competitions. They were relationships — a professional community deciding, individually, that a 70-year-old local institution should not be allowed to die.

In May 1979 the firm produced its first promotional brochure in seventy years of existence — “a low-key monochrome five-page document.” Funds under management were a little over £300 million, a third of it Scottish Mortgage.

Everything Baillie Gifford now says about itself — the loyalty, the long view, the indifference to fashion, the horror of dependence on any single client — was forged in that year.

The rebuild, and the Gang of Four

The more important change was generational. The investors recruited between 1963 and 1973 were gradually given real responsibility, and a group that had joined a cautious, inward-looking firm began rebuilding it. A 1980 reorganisation helped by assigning clearer geographic ownership: Max Ward, “by far the most bullishly minded partner,” took the UK; William Carnegie the US; Robin Menzies the Far East.

Ward’s team started picking British stocks “believed to have the potential to make a lot of money for clients rather than simply to avoid losing much,” the UK record turned, and the firm could finally compete for domestic pension mandates.

The generation that executed the turnaround — recruited between 1963 and 1973 — led the firm through the following two decades. Burns’s own chapter title for the era from 1989 is “The Gang of Four”: Gavin Gemmell managing the business, Douglas McDougall the clients and the industry relationships, Max Ward the UK book, Richard Burns the pension funds.

Their machine worked. Between 1989 and 1994 alone the firm won 46 new accounts and £2.75 billion — Scottish and English local authorities, then in 1992 the £500 million Universities Superannuation Scheme, the second-largest pension fund in Britain. Assets went from £300 million in 1979 to £3.5 billion in 1989 to over £55 billion by the end of 2007, as the firm approached its centenary.

Two of that era’s decisions matter enormously for what came later.

The first was going to America early. In 1991 the firm signed a joint venture with Guardian Life of New York — and the young partner who made the follow-up trip to win it, making “a much more dynamic and enthusiastic impression” than the stuffy image the visitors had formed of the Edinburgh office, was James Anderson. Anderson had joined in 1983, made partner in 1987, and ran European investing; the American relationships he helped open led, in 2003, to the mandate that still anchors the US business — sub-advising part of Vanguard International Growth. By August 2007, Baillie Gifford managed approximately $7.1 billion, or 39% of the fund.

The second was what happened when the Gang of Four left. McDougall retired in 1999, Ward in 2000, Gemmell in 2001 — and in April 2000, at almost exactly the top of the dotcom bubble, Max Ward handed the management of Scottish Mortgage to James Anderson. The most bullish partner of the old generation passed the flagship to the most restless partner of the new one, at the worst conceivable moment, and it turned out to be the best succession decision in the firm’s history. Anderson tore the trust away from its home market — UK equities were 43% of Scottish Mortgage at the end of 2000 and 15% by 2007 — and pointed it at global growth. From his appointment through March 2022, Scottish Mortgage delivered a share-price total return of 1,483%, compared with 354% for the FTSE All-World Index.

2003: Long Term Global Growth

The most important thing for understanding modern Baillie Gifford is where Long Term Global Growth came from.

LTGG and Scottish Mortgage are separate portfolios. LTGG is an unconstrained global-equity strategy launched in 2003; Scottish Mortgage is a listed investment trust with its own independent board, mandate and capital structure. The direct link was Anderson: he helped build LTGG while managing Scottish Mortgage, and the trust became the most visible expression of the same high-conviction, long-duration global-growth philosophy.

By the early 2000s Anderson had become — in the words of the firm’s own history — “acutely frustrated by the way in which benchmark-based investing in the 1990s had led to a lack of creativity, flexibility and imagination in the firm’s investing and an apparent inability on the part of many investment staff... to take significant risks.”

His answer, built in 2003 with Mark Urquhart and Craig Collins: portfolios constructed purely bottom-up, ignoring country and index weight entirely, concentrated in 40–50 holdings, held for many years.

Scottish Mortgage, already under Anderson, became the philosophy’s most visible public expression: Amazon, Tesla from 2013 at around $6 a share on a split-adjusted basis, the Chinese platforms, and — from 2012 — private companies. Its closed-end structure, with permanent capital and no investor redemptions, made it easier to hold illiquid businesses without maintaining the same degree of daily liquidity as an open-ended fund. By 2008 LTGG was the most popular product the firm had ever launched, and its portfolio held no UK equities at all.

The claim

Underneath the century of narrative sits a single statistical claim — published by the firm and rerun a decade later.

In 2014 a Baillie Gifford manager studied stock returns from 1984 to 2013:

“When I did my original study, looking at data from 1984 to 2013, it revealed that about 20 per cent of companies create half of the return. Furthermore, it showed that only 5 per cent of stocks grew at least five-fold in any five-year period. There was a small number of big winners that drive most of the returns.” — *Back to Blue Sky and Base Rates: 10 Years On*, Scottish Mortgage, June 2025

Ten years on, they ran it again on a broader index over a different period. The structure held:

“Again, over the past decade, 20 per cent of the companies drove about half of the return. The top 5 per cent of companies returned at least four and a half fold. It remains true that, when seeking the best returns, only a few companies actually matter. That skew doesn’t bother us. In fact, it’s the source of our investment edge.” — *Back to Blue Sky and Base Rates: 10 Years On*

If a small minority of companies drives most long-run returns, excessive diversification can dilute the winners, conventional valuation screens can remove them too early, and premature selling can be more damaging than accepting a number of individual failures. Concentration, tolerance for apparently extreme multiples, decade-long holds, the willingness to look wrong for years — all of it follows. So do the drawdowns: if your returns come from the tail, you will periodically lose more than everyone else. Four corollaries the firm has put in writing:

Mean reversion is the wrong mental model. “In a world of exponential change, ‘the long run’ itself is shifting. The average of yesterday may bear little resemblance to the future.” (*The World Is About to Get ‘Weird’*, 2025.)

Apparent safety is a trap. “Safe havens can be the most exposed when technological change accelerates... Some of my worst investment mistakes have come from backing companies that looked steady, only to discover that steadiness was a mirage.” Most managers treat quality as a defence; Baillie Gifford treats it as an unpriced disruption risk. (*The World Is About to Get ‘Weird’*.)

Run the winners. “Stock ranking has been an important differentiator. We’ve ‘run our winners’.” (*Back to Blue Sky and Base Rates: 10 Years On*.)

Selling is the expensive mistake. “All my greatest mistakes have been selling, or not buying, companies — actually more, in my own personal case, selling truly great companies way too early, rather than the it’s-gone-to-nought, which is very painful but I think becomes too dominant.” Asked how one improves the selling decision, his answer was that the occasions when you should sell are “very, very few.” (James Anderson on the Money Maze Podcast.)

The partnership, plainly

One structural fact explains why the firm can behave this way: nobody outside owns any of it.

Baillie Gifford is a Scottish partnership with unlimited liability: every partner works in the business, owns a share of it, and stands behind the partnership’s obligations without a preset cap.

Like Goldman Sachs during its partnership era — the people running Baillie Gifford also own it and have their own capital, careers and reputations tied to its survival. Unlimited liability reinforces that alignment, but it does not by itself produce long-term investing. The more important advantage is that there is no outside shareholder base or quarterly earnings target pushing the firm to maximize near-term revenue.

When things go wrong, partners try to fix the firm rather than leave it. According to Burns, “the reaction has been to try to redeem the situation rather than it being a case of every man for himself.” As a private partnership, the firm did not have to issue public-company earnings announcements when large mandates were lost in the 1990s.

And the firm can refuse growth: it closed to new business during the 1990s boom, a decision that can be harder for a listed manager answerable to outside shareholders. The trade-off, which Burns concedes with equal honesty, is that there is no public shareholder or takeover market imposing discipline. The principal external checks come instead from clients, independent trust boards, regulators and counterparties; Burns’s example of the internal risk is the complacency about client concentration that set up 1978.

In Part 2, I’ll look at what Scottish Mortgage owns and looks like today.

Ted Weschler is now Berkshire’s sole dedicated investment manager after Todd Combs‘ departure. His IRA balance is the stuff of legends, but he’s given almost zero interviews, with all but one of his investor letters private (the final Q3 2011 letter survives). Moreover, since he walked into Berkshire Hathaway in early 2012, his ideas have been buried inside Berkshire’s 13F—no attribution, no way to separate Weschler’s positions from those of Combs or Buffett except by inference.

However, there are two things we can do. First, we can look at Weschler’s 13F filings pre-Berkshire when he ran a hedge fund called Peninsula Capital Advisors. We can also look at the holdings sold once Todd Combs left (December 2025, to run JPMorgan‘s new strategic investment group) and, based on Buffett’s comments, attempt to back-out Weschler’s portfolio.

Weschler ran approximately $2 billion out of an office above a bookstore, generated one of the best hedge fund records almost no one has heard of, and then an IRS data leak revealed that his retirement account had reached a quarter of a billion dollars. Of the leaked accounts, only Peter Thiel’s had a higher balance.

When asked which manager he respected in his only TV appearance, he cited David Tepper of Appaloosa—a surprising choice for someone widely assumed to be a “plain vanilla” value investor. As we’ll see, however, the two share important similarities in their investment approaches. What follows is a reconstruction of Weschler’s record from the fragments of information we do have.

The operator’s lens

Most of the managers we profiled so far came up through stints as analysts at a Tiger Cub or Fidelity. Weschler came up a third way: he was an operator first, a private equity buyer second, and a public-markets investor third. The result is a manager with no house style in the conventional sense. He has simultaneously held: compounders (DaVita in every available Peninsula 13F), bankrupt equity (W.R. Grace), special situations (the EchoStar/DISH split, four classes of Liberty tracking stock), cigar-butts (Wilsons The Leather Experts, Stanley Furniture), crisis-priced financials (BoA in Q1 2009, WSFS through the cycle), and airlines—traded opportunistically in 2007–09 (US Airways, AirTran, AMR, Delta, Northwest, Allegiant). The types of businesses and their capital structures varied significantly over time, reflecting Weschler’s ability to adapt to the opportunity set he saw.

Background

Weschler was born in Buffalo in 1962 and grew up moving around the Rust Belt — his father was an A&P executive who relocated the family every few years, and Weschler spent much of his childhood in Erie, Pennsylvania, where he attended Cathedral Prep.

A&P was itself one of the great American business rise-and-fall stories. A&P was the Walmart of its day: the world’s largest retailer by the late 1920s, the first to reach $1 billion in sales, nearly 16,000 stores at peak, built on John Hartford’s “Economy Store” formula — cut the price and margin on every good to drive volume — plus vertical integration (its own bakeries, canneries, coffee importing). A&P was so aggressive that 26 states passed anti-chain-store taxes and the federal government criminally convicted the octogenarian Hartford brothers in 1946, essentially for selling food too cheaply.

By the time Weschler was a boy, the machine was dying: the founders’ insular successors “had a poor understanding of how the world was changing” (Marc Levinson’s verdict in The Great A&P), stores sat in decaying urban cores “five to ten years behind the migratory patterns of its own clientele,” all led to a collapse from ~3,400 stores to just over 2,000 in 1975. That is the professional world of an A&P executive in the 1970s — constant relocations through a shrinking empire.

It’s hard not to see some link with the son’s later obsessions. Weschler has made a pretty penny in retail: Wilsons Leather, the Detlev Louis deal, a 30-year Furniture Today subscription, and, decades later, a personal 5.9% swing at Dillard’s. Depending on the timing of his sales, the approximately $45 million stake appears to have grown to roughly $260–300 million of gross value by the time the position was gone at year-end 2021.

He went to the Wharton School at UPenn, graduating in 1983. Details of his background are thin, but one detail from his NFM interview is that a friend handed him Buffett’s writings in Philadelphia in 1979, and while he did not enter public equities, the “tremendous clarity” stuck.

W.R. Grace: the apprenticeship

Weschler joined W.R. Grace & Co. straight out of Wharton as the M&A department’s first-ever undergraduate hire — “It was otherwise just MBAs… I had the business card that said, ‘Junior Financial Analyst’” — at a $22,000 salary:

“I didn’t make much money, but if you stayed until eight o’clock working, you got free taxi home and free dinner. And that made a big difference. So you know I was there till at least eight every night. And it was fun. I like working a lot more than studying.”

J. Peter Grace was, at the time, the longest-serving CEO of any major American public company — he took over the family firm in 1945 at age 32 and ran it for 48 years. He had inherited a Latin American shipping-and-guano trading house (founded 1865 by his grandfather, New York’s first Catholic mayor) and rebuilt it by force of acquisition into the fifth-largest U.S. chemical producer — buying more than 100 companies and selling more than 70 across chemicals, restaurants, retail (Herman’s, FAO Schwarz), cocoa, even co-founding Pan American-Grace Airways.

A formative event: after 100-hour weeks on the National Medical Care acquisition—announced in December 1984 when Weschler was 22 and completed in June 1985 after he turned 23—he was suddenly summoned to brief the CEO because every boss in his chain of command happened to be out of the building:

“I’m 23, I look like I’m 12. This is not what I signed up for.”

Grace opened with maximum contempt — “This is the biggest acquisition we’ve ever done, my grandfather founded this company, and you’re sending this kid up to brief me on this thing!“ — and then grilled him for ninety minutes. Weschler survived: “I’ve been doing 100 hour weeks on this deal. And I knew the deal well.” Grace’s parting shot, after the ordeal: “You know, there’s two people in this room that don’t have an MBA. You and me.”

The briefing earned him two years as personal aide to Peter Grace, attending every budget and operating review across the conglomerate’s wildly diverse divisions — a compressed apprenticeship in how dozens of different business models actually make money.

“Peter and I had a great relationship, but I viewed him more as a, almost a monarch, than a CEO. … He had to go through six doors to get to his office, and everything was maximum intimidation. … And then, Warren’s just like, I show up at headquarters, and he bounces out in the hallway and says, Great to meet you, Ted, come on in. Grab a seat on the couch.”

And there’s a crucial, little-known plot turn at the end of the Grace years: the October 1987 crash is the accident that created Quad-C. Just before the crash, the 25-year-old Weschler had accepted a job at a risk-arbitrage shop. Grace was upset but accepted it — until Black Monday hit, at which point he called Weschler and told him “you’re basically an idiot — he used much rougher language than that — for leaving,” and started sending senior executives to find a way to keep him.

The solution: Weschler became aide to Grace’s heir apparent (”probably heir apparent number four or five” — Peter Grace was 75 and not going anywhere), a man he’d already done leveraged buyouts with. That man was Terry Daniels, and two years later, over a dinner, the two of them hatched the firm:

“He was a gray-haired guy with gravitas and I was the maniac in the back room who knew the tax code and knew how to use a computer. And it worked.”

Quad-C

He left Grace at the end of 1989 to co-found Quad-C Management. Daniels was a heavyweight in his own right: at Grace he was vice chairman and a director with operating responsibility spanning the specialty-chemicals, health-care, retail/restaurant and natural-resources groups, negotiating over $4 billion of acquisitions and divestitures — and he remains Quad-C’s chairman today. Quad-C is an enduring firm that is on its 10th fund (~$1.7 billion in commitments).

It speaks to Weschler’s ability and the trust that he earned that Ted was brought on as a co-founder at age 27 and that he wore two hats at the firm:

“When we first set up Quad-C, it was this amalgamation of my partner’s idea that we wanted it to have control investments, where we actually bought companies. And my preference was I wanted to do investing in publicly traded securities... And so we created this hybrid where we said... we’ll ask folks that we know that have some money to put money in, and I would invest it in public securities while we were looking for control investments. And I kind of wore both hats.”

So for Quad-C’s first six years, Weschler ran a public-securities sleeve inside a private equity firm with permanent-feeling capital from people they knew, with a true private-ownership mindset.

“There’s a great quote of Warren’s about, We’d all be better investors if we had a punch card that every time we made an investment, we had to click it, and that punch card had only 20 punches available to you. So you really had to have this high bar to make an investment. Well, lo and behold, in the six years that we did this, I made exactly 20. Exactly. Which is really weird how that worked out. And 19 of the 20 worked out really well.”

And the one that missed?

Host: “Alright, I have to ask you this. You got 19 out of 20. What was the one you missed? I’m too curious, I’m sorry. I know you know it.” Weschler: “Yeah, I do. It’s funny, because we went by the name of TW Holdings at the time, and my initials are TW, so it’s like, just, of course, I should have stayed away from here. And it was a big restaurant operator at the time, it owned Denny’s and several other things. It was — and, you know, sometimes things don’t work out. But it did not work out.” Host: “It was the initials.” Weschler: “Yeah. Vanity took me.”

For the record, TW Holdings was the old Trans World non-airline empire — Denny’s, Hardee’s franchises, Quincy’s, El Pollo Loco, Canteen — taken private in a hostile $1.65B LBO in 1989, renamed Flagstar in 1993, and eventually crushed under $2.2 billion of debt into a 1997 Chapter 11 that wiped the old equity. A post-LBO, over-levered equity stub.

At the six-year mark Quad-C institutionalized — endowment-type LPs “tend to have checklists… they want you to be exactly this slot or that slot. They don’t want you to be a hybrid” — Weschler himself wrote the private placement memorandum that raised the institutional fund and, in doing so, wrote his own hybrid sleeve out of existence: “I realized it wasn’t exactly what I loved. The firm had grown a bit. I was doing more people management than investing in businesses.”

“I’m an analyst. I still want to have a business card that says analyst. … How many employment agreements can you negotiate? The learning curve had just flattened out.”

The Peninsula concept, in his own words, was to port the PE mindset back into the public market: “if I could take the private equity mindset where you’re buying the business and you expect to own it for five to ten years and do that with publicly traded securities — where I’d buy 3% or 4% of a business but I’d have a very long-term view — I thought that might be something that would work.” Against a market where “the average holding period for a share of stock on the New York Stock Exchange is less than a year,” that was an edge. He even has a name for the activist-adjacent, never-hostile version of it:

“Over time, I call it MAD investing — make a difference investing. If you can potentially change the outcome by being actively involved in some part of the process.”

In 2000 he struck out on his own and launched Peninsula Capital Advisors.

Philosophy & Process

Weschler has articulated his approach through five main sources: 1) a short March 2014 CNBC interview with Buffett and Combs—his only real television outing; 2) a longer-form NFM podcast interview and transcript; 3) a 2002 trade-press interview given mid-bankruptcy on W.R. Grace; 4) his final partnership letter; and 5) a little-known June 2016 interview with Germany’s *Handelsblatt* after Berkshire’s Detlev Louis deal. Beyond that, the public record is thin. Even so, it paints a surprisingly rich picture.

1. Variant perception is the edge.

“The only way you’re going to have success in the stock market is if you’ve got variant perception, something that’s different from the masses.”

“If you’re just reading the New York Times and the Wall Street Journal, there’s just no way you’re gonna beat other people.”

2. Therefore, read broadly.

He deliberately rotates which newspaper he reads (USA Today one day, the NYT or FT the next) to gain an “every man’s” perspective and keep his inputs from homogenizing.

“I read Furniture Today and Uranium Weekly. I’m not sure there’s a lot of people who subscribe to both of those, but you’re looking at one of them.”

“Trade journals, they’re really powerful and there’s a lot of investors that don’t subscribe to those.”

“I want to be able to look myself in the mirror and say that I’m reading enough weird stuff that nobody else is reading the same stuff that I am.”

He frames the whole exercise as compounding knowledge, not just capital:

“Investing is kind of a game of connecting the dots. … the longer you are in the business, as long as you are intellectually curious, your collection of data points of dots gets bigger and bigger.”

3. Value the business before you look at the price.

Weschler treats it explicitly as a game:

“And a game I always like to play is — do work on a company and say, without knowing what the price is in the stock market, study it and say, this is where I think it’ll be. And then you do the uncovering thing, you say, I think this should trade at $82 a share. And it’s like, Oooh, it’s at 40. This is interesting. Usually it means I missed something really bad. But it’s always been a game that I’ve played, and it keeps it interesting.”

Striking how precisely this rhymes with his former counterpart. Todd Combs, on the NFM podcast a year later, described the identical discipline:

“I don’t even look at the market cap of a name. So a game that I’ve always played with myself is like, look at the name, do your work, build up what you would buy that entire business for. … And Mastercard is a good example of that, I think it came public for three or three and a half billion, and I… purposely didn’t want to know… and did not know, or hear that at the time, and I valued it at like 30bn or something like that.”

Two managers who independently converged on blind-valuation-first as an anti-anchoring device.

4. The punch card — extreme concentration, earned by hours.

Bloomberg’s 2012 profile added the quantitative version of the discipline: he reportedly didn’t make an investment unless he had spent approximately 500 hours studying the idea.

“If I can come up with one decent investment idea a year, boy, that is great.”

“Why put a dollar in your 11th pick when you could put more in pick number 1, 2, or 3?”

5. Do nothing, on purpose.

Patience is treated as an active strategy, not a default.

“resisting the temptation to sell — doing nothing can run counter to a serious work ethic, but in the world of investing it can be a very effective strategy.”

“I have never been good at picking stocks over less than a one-year horizon — in fact, I have always guided towards measuring success or failure over a three-year period.”

“If you don’t buy something, the Wall Street Journal still shows up. … You don’t have that moral obligation.”

W.R. Grace — a full case study

If you want one position that is pure Weschler, it is W.R. Grace — yes, his first employer.

The setup. By 2000, Grace had shrunk from Peter Grace’s sprawling conglomerate into a focused specialty-chemicals company (catalysts, construction products) but its legacy Zonolite vermiculite mine in Libby, Montana had laced products with tremolite asbestos; the late-1999 press exposés detonated the liability, and as fellow defendants fell into Chapter 11 around it (Babcock & Wilcox, Owens Corning, Armstrong), Grace’s stock collapsed from ~$18 toward $1.50 on the assumption that the equity would be wiped.

The entry—before and through the petition. Weschler’s earliest filing, through Peninsula Investment Partners in Q4 2000, already shows 3,536,200 Grace shares at approximately $3.19. This partnership vehicle has a different CIK from the later Peninsula Capital Advisors filing series, but it was managed by Weschler and represents the same Grace investment campaign. He was buying into the asbestos panic months before the bankruptcy.

Grace filed Chapter 11 on April 2, 2001 facing 65,000+ lawsuits and over 129,000 claims. Three weeks later Peninsula disclosed 6.63 million shares — 10.1% of the company. Through May–July 2001, Weschler kept buying in the open market at $1.60–$2.00 per share, and by the July 2001 Schedule 13D held 10,546,100 shares (16.11% of the company) at a total cost of ~$26.8 million — an average of roughly $2.54 a share. Weschler had a ~$150M fund with ~18% in the equity of a company in asbestos Chapter 11.

When the U.S. Trustee appointed an Official Committee of Equity Security Holders in June 2001, Peninsula got a seat, and on July 18, 2001 Weschler was elected its Chairman, a table he would sit at for thirteen years.

The thesis—in his own 2002 words.

“It’s a great core specialty chemicals operation, relatively non-cyclical and led by a great management team. Last year they had earnings per share of $1.20. If you put a long-term market multiple on $1.20, which Grace would certainly have without the asbestos taint, that’s a $24 stock.”

The stock was at ~$3.35 when he said that — so his unimpaired-business estimate was ~7x the quote, and his stated target of $12–13 already assumed roughly 50% dilution to asbestos claimants. Dilution that, in the end, never came.

The precedent:

“It’s not unprecedented. The Grace situation is more analogous to the situation with Texaco years ago and Columbia Gas Systems.”

Both Texaco (1987) and Columbia Gas (1991) were solvent-company Chapter 11s where shareholders came out whole — that was the reference class he was underwriting from, not the asbestos wipeouts everyone else saw.

On why Grace’s capital structure was different:

“The big difference between Federal-Mogul and Grace is that Federal-Mogul has a ton of debt. So the argument is made that in Grace, the equity is more analogous to the position Carl Icahn has in Federal-Mogul.”

At under 2x EBITDA, the lenders were fully covered and effectively money-good, so they had no incentive to fight for the company; the real contest was asbestos claimants versus shareholders over a growing enterprise.

In Federal-Mogul, the debt stack was so heavy that the smart-money way to play it was to be the debt (where Icahn sat). At Grace, the equity itself was the fulcrum security. On his horizon and the risk:

“The nature of the fund is that anything I put in there I expect to hold for five plus years.”

“The risk in bankruptcy court is that there are a lot of wild cards that can come out of nowhere and hurt you.”

Todd Combs later described the same analytical move as finding the “fulcrum class” in a capital structure—the security that absorbs losses at the point where enterprise value runs out. It is a useful analogy for Grace, although Combs was discussing mortgage securitizations rather than Weschler’s investment.

What actually happened handily beat his valuation. The automatic stay meant Grace stopped bleeding litigation costs the day it filed. Behind that shield the business compounded: revenue went from $1.6 billion in 2000 to $3.2 billion in 2012, with 25 acquisitions completed during the bankruptcy. A bankruptcy lawyer’s verdict from the era: “Bankruptcy has been a great place to hide out. It has for all intents and purposes been business as usual.”

The endgame was a global settlement that moved current and future asbestos claims into dedicated trusts, funded by Grace and settlement contributions tied to businesses that had previously been separated from the company.

In 1996 Grace separated National Medical Care—the dialysis business and the very deal Weschler worked on in 1984—into Fresenius Medical Care, and in 1998 it merged its Cryovac packaging arm into Sealed Air. In each case, the operating business moved while the asbestos liabilities remained with the much smaller Grace.

When Grace then went bankrupt, asbestos claimants sued Fresenius and Sealed Air, arguing that the transactions were fraudulent transfers. Rather than risk having the deals unwound in court, both buyers settled: Sealed Air for approximately $1.13 billion and Fresenius for $115 million. Those payments became major funding sources for the trusts that allowed Grace’s shareholders to emerge intact. The deals Weschler helped build in the 1980s ended up helping pay for the bankruptcy he chaired in the 2000s.

When Grace emerged on February 3, 2014 after 12 years and 10 months — one of the longest Chapter 11s ever — the trusts held over $4 billion, and the old common shareholders retained their shares, undiluted.

Even the warrant never diluted anyone. Grace and the trust agreed to settle the warrant’s value in cash on a sliding scale — $375M if the stock averaged below $54.50, capped at $490M if above $66. Because the stock was above $90 at emergence, Grace simply wrote the trust a $490M check, retired the warrant, and never issued a share.

The payoff.

From Weschler’s approximately $2.54 average cost to Grace’s roughly $92 share price at emergence, shares held through 2014 appreciated about 36x. That 36x applies to shares retained beyond Peninsula’s wind-down; Peninsula distributed its Grace position in kind at approximately $33.30 in 2011, or roughly 13x Weschler’s average cost at that point. Bloomberg’s October 2012 profile made the related point that his settlement initiative “helped Grace’s stock climb 40-fold since the bankruptcy”—the move from approximately $1.50 on the petition date to roughly $60 in late 2012, rather than Peninsula’s own fund return.

When Peninsula wound down in December 2011, the Grace shares were distributed in kind — Weschler personally took 3,740,902 shares (5.1%) out the door, still holding them through at least the last disclosure in December 2012. Grace was ultimately taken private by Standard Industries in 2021 at $70 a share.

There were lots of elements that made this a great investment: variant perception (asbestos equity isn’t worthless), the operator’s lens (he understood the chemicals business), the seat at the table (chairing the equity committee for 13 years), and a time horizon nobody else would tolerate.

“That’s where this W.R. Grace bankruptcy came along, where it’s like, wow, I know this company well. … I’ve done a lot of work in bankruptcy for whatever reason, and I just thought it was something that I might be able to make a difference there by getting involved. And I bought about 15% of the business. And again, it took 13 years, but it ended up being a very successful restructuring that I was involved in.”

What Peninsula Owned

Peninsula Capital Advisors filed Form 13F from Q3 2003 until a final, empty filing for Q4 2011. Across the available record, its reported 13F long-equity book was never more than 19 names and usually held 9–14, with the top four or five typically representing 70–85% of reported long-equity value.

The chart below shows the evolution of Peninsula’s reported 13F long-equity book and the extreme contraction visible during the financial crisis. It should not be read as fund NAV or AUM. Fact-checked against the filings themselves, both 2008 totals are exact to the nearest thousand dollars: Q3 2008 sums to $915.181 million across 15 holdings and Q4 2008 to $389.143 million across 17. The 57.5% quarterly contraction is real, but it is not all investment losses.

Decomposed, roughly half the drop is the GE sale—approximately $255 million of stock converted to cash, which is invisible in a 13F because the filing shows only long U.S.-listed equities. Another approximately $80 million net came from other trimming: DaVita fell from 4.12 million to 2.86 million shares and DISH was more than halved. The mark-to-market decline on what he actually kept was 36%, worse than the S&P’s 22.5% decline that quarter, but driven heavily by W.R. Grace—down 60% on identical shares—and Valassis, down 85%. Excluding Grace, the retained book fell roughly in line with the market. The fund’s actual drawdown was therefore less severe than the 13F contraction suggests because part of the portfolio had been converted to cash, which helped fund the Q1 2009 purchases.

The table below shows how the portfolio has evolved and flowed through the years.

Finally, an estimated P&L by position.

Act I — the inception book (2003)

The first filing is thirteen names, $460M, and no “house style” — every archetype already present:

Note that he owned both DaVita and Fresenius—the two heirs of the dialysis industry he’d diligenced in 1984 and presented to Mr. Grace. The microscopic Cogent line, just $159,000 of a busted-telecom-bubble fiber operator, would later become a $51 million position in Peninsula’s reported 13F book. Public disclosures do not identify how much, if any, of Cogent was held inside Weschler’s IRA.

Per Ted Weschler’s last letter in 2011:

DaVita. The second-largest U.S. dialysis provider, offering a “life-sustaining therapy” with recession resistance, patient growth, international expansion, and value-accretive buybacks.

Cogent. An all-fiber internet backbone assembled from distressed assets for pennies on the dollar, competing on price—”it won’t be undersold”—with near-zero marginal cost and a long runway for growth.

WSFS. A 179-year-old Delaware bank with strong management, “deep local focus,” and room to take share after Wilmington Trust’s collapse.

Act II — the build-out (2004–2007)

The reported 13F long-equity book grew approximately 3.5x, from $460 million in Q3 2003 to roughly $1.6 billion by the end of 2006, on the back of a consistent core: DaVita—steadily the number-one or number-two position—the frozen Grace block, and a huge satellite/pay-TV complex. That complex included EchoStar, DIRECTV, first held from 2004–2006 at up to 12 million shares, Pegasus, and the FiberTower/First Avenue spectrum plays.

DIRECTV. A satellite platform with “high capital efficiency,” combining U.S. share gains with underpenetrated Latin American growth and aggressive repurchases.

Around it rotated the special situations: US Airways post-bankruptcy equity (4M shares, a top position 2006–07), XM Satellite Radio, News Corp, Sprint, and the Wilsons Leather stub (peaking at 15.5M shares). In 2007 he added a 10-million-share, ~$370M General Electric position.

Going into the storm, the Q4 2007 book ($1.6bn, 13 names) was: GE $371M, DaVita $338M, Grace $282M, EchoStar $226M, Wilmington Trust $99M, US Airways $59M, XM $43M, Cincinnati Bell $40M, News Corp $38M, Suncor $33M, and stubs of Wilsons, FiberTower, and Valassis: a mega-cap financial/industrial, a Delaware trust bank, and two airlines-adjacent names. Weschler did not see what was coming any more than anyone else did.

Valassis. A coupon and direct-mail business mispriced as a newspaper casualty: media fragmentation increased its ability to “deliver reach,” while free-cash-flow yield exceeded 18% and buybacks shrank the share base.

Cincinnati Bell. A “publicly traded leveraged buyout”: a stable but declining wireline operation funding a high-return data-center business, with tax assets and reinvestment adding rerating potential.

Act III — the crisis, trade by trade (2008–2009)

The site’s composition and value views, windowed to Q4 2006 → Q4 2011 — the crisis and the reload in one frame:

Q1 2008 — cutting the periphery. Out of XM Satellite Radio and News Corp entirely; Wilmington Trust slashed from 2.8M to 605k shares (a financial he did see through — Wilmington ultimately collapsed into an M&T rescue); and a first ~860k-share toehold in Valassis appears at the lows. The book: 11 names, $1.20B (from $1.57B at year-end — mostly price).

Q2 2008 — the last quiet quarter. Wilmington gone entirely; AirTran added (2.25M shares); Valassis held; DaVita trimmed from 5.74M to 5.48M shares. The book: 12 names, $1.07B. Two quarters into the crisis he had already exited the financial he distrusted (Wilmington) while GE was still fully intact at 10M shares.

Q3 2008 (Lehman quarter) — first blood into the panic. Still holding GE’s full 10M shares ($255M — painful), he added names as prices broke: Liberty Entertainment (1.5M sh), Cogent back up to 4M sh, Valassis to 4M sh, Cincinnati Bell 6.5M sh, plus airline scraps (Northwest, US Airways). 15 names, $915M.

Q4 2008—the capitulation-and-reload quarter. The recovered filing shows the reported 13F long-equity book at $389 million—down approximately 75% from its 2007 peak—and reveals his single biggest crisis decision: he sold the entire 10-million-share GE position into the wreckage, avoiding what was still to come as GE bottomed months later near $6, and cut US Airways from 1.3 million to 200,000 shares.

Where did the freed capital point? At maximum-beta reflation: he tripled down on the Liberty complex (Liberty Entertainment up to 3.46M sh, new stakes in Liberty Capital and Liberty Interactive — buying John Malone’s levered trackers at the exact moment forced sellers were dumping them), held every share of Grace, Cogent and (mostly) Valassis while trimming DaVita from 4.12M to 2.86M shares — and added AMR to AirTran (the Delta line that appears this quarter is not a new buy: his 300,000 Northwest shares converted into exactly 375,000 Delta shares in the October 2008 merger), assembling a four-airline basket at the bottom of the world. Seventeen positions, most of them option-like.

The filing-implied prices put numbers on the carnage: GE was marked at $25.50 at Q3 (10M shares, $255M) and closed the year at $16.20 — he sold somewhere in that window, dodging a further −36% quarter and the eventual ~$6.66 March 2009 bottom. Grace fell $15.12 → $5.97 on identical shares over the quarter (−60%), Valassis $8.66 → $1.32, the Liberty Entertainment tracker was picked up around $17.5 against $24.97 a quarter earlier.

Q1 2009 — the offense continues. At the actual market bottom he initiated Bank of America (2M shares), re-established GE at 3M shares (having sold 10M higher), took Cincinnati Bell to 10M shares, and rebuilt WSFS. The book: 19 names, $453M, positioned like a coiled spring.

The Q4 2008 sale was superb timing within a losing investment: he bought a levered financial-industrial at $33–37—its weight peaked near 31% of the reported 13F book in 2007—and the crisis proved the thesis wrong. What he deserves credit for is recognizing it before the worst of it.

Bank of America was the mirror image: first reported in Q1 2009 near the market bottom, held for roughly a year, and exited into the 2010 recovery—approximately 2.2x on AlphaFile’s filing-based estimate.

The spring uncoils. $453M (Q1) → $700M (Q2) → $1.05B (Q3) → $1.17B (Q4 2009), while he added DaVita back up to 5M shares and re-initiated DIRECTV (5M shares by Q4 2009) — the old 2004–06 favorite, repurchased in size for the second act. The Liberty Entertainment tracker he’d loaded at the bottom converted into DIRECTV shares in the November 2009 Malone restructuring: the structure trade and the conviction trade were the same trade.

Exiting the crisis. Through 2010–2011 he concentrated the winnings. Nineteen names became eleven, then nine. Bank of America, GE, DISH, the airline basket, TiVo, and the small stubs were all gone by early 2011, recycled into the core. The final act: DIRECTV grew to 10 million shares / $508 million by Q2 2011, the single largest position the fund ever reported, alongside Grace at $491 million and DaVita at $372 million. The reported 13F long-equity book reached $1.96 billion across nine names, with the top three representing approximately 70%.

The scoreboard on the crisis: peak to trough, the reported 13F long-equity book fell approximately 75%, with much of the mark-to-market damage concentrated in the frozen Grace and DaVita cores. From the $389 million trough, the reported book quintupled over the next 30 months. Decisions made in Q4 2008 and Q1 2009 drove a large portion of that recovery, while longer-held positions such as Grace and DaVita and the later DIRECTV build also contributed materially. AUM grew from approximately $150 million at inception to roughly $2 billion, while early investors’ cumulative return reached approximately 1,030%.