As we know, the entire LMND thesis hinges on IFP growing exponentially while Opex minus growth spend growing modestly. Here, however, we had Opex minus growth spend growing at a higher % than IFP!

Specifically, Opex minus growth spend grew 49% YoY (Cf IFP growing low 30% range), which is starkly counter to what Management and Shai (Co-CEO) had been representing, which is 5% or otherwise single digit growth.

Read more

Notwithstanding, I am a Palantir holder and so tuned into the Q2 earnings call.

For those of you who don’t know, Palantir smashed their Q2: They delivered 93% YoY revenue growth, 149% Yoy US commercial growth and increased their FY 2026 guidance. And Palantir shot up 35% in 2 days.

The best way to think about Palantir is not about fair values (as every value investor who has tried to value Palantir has been left for dead).

Instead, holding Palantir requires a view of the future which is as follows: LLMs will become commoditized and the value will be in the data. Palantir captures this data perfectly in their Ontology, which then allows clients to maximise the use of this data to generate business efficiencies. As LLMs become better and better and data more and more prized, Palantir’s ontology will become more and more important.

I personally think Palantir will be an indispensable company in the future - you either get on board or be left behind. For that reason, it will always have a place in my portfolio. But I’m also not telling you to buy it now as I acknowledge that there are more favorable R/R plays out there.

The reason I am writing this article is because I wanted to share some of the things that were said in Palantir’s earnings call that you DO NOT want to miss - these have significant implications for software.

Read more

On the surface, they had a great quarter, beating IFP and Revenue Guide - see summary below.

The Bullish

There were many other great things in the quarter.

IFP grew 32% YoY - well above the 30% YoY we need for our bull case to play out.

The LAE ratio (cost of handling claims that's part of the loss ratio) improved to 5%. This is genuinely incredible.

Car also achieved an LAE of 7% (compared to industry average of 8-12%).

Gross Loss Ratio continued to be great at 60%.

Management raised Revenue and Gross Earned Premium guidance:

• IFP: unchanged $1,632–1,639
  GEP: raised to $1,374–1,378 (from $1,369–1,373)
  Revenue: raised to $1,214–1,220 (from $1,197–1,203)
  Adj. EBITDA: unchanged ($51)–($47)

Car grew 59% YoY. This is the part I am most pleased about. In order for LMND to achieve the bull thesis I have, car needs to grow rapidly and be responsible for LMND’s 10B targetted IFP target in 2034. This is confirmation that car is growing rapidly and the thesis is in tact.

Overall, there were many things in this quarter which I was pleased about, and confirms that the LMND thesis intact.

One more thing needs to be said: LMND’s sales and marketing was $77.7M for the quarter. If they spent nothing on marketing, they would have had $34M in net income. Put differently, they would already be at breakeven if they only spent $34M on marketing in Q2. Two years ago, they spent around $30M a quarter on marketing. So it’s breakeven on 2024 growth spend levels. One interpretation is LMND likes its marketing efficiency and is deciding to spend aggressively now to reap the rewards later, which I think is likely the case.

The Bearish

Read more

Performance Overview

Here is my performance since 5 June 2026 (the date I started this Substack)

• Returns since 5 June 2026: +6.00%

• S&P 500 Returns since 5 June 2026: +1.00%

*Chart generated by Claude

As you can see, we significantly outperformed the S&P, which was really surprising considering how volatile the recent week has been (and how high beta the names I hold are).

Now to all the details.

Portfolio Overview

Live Holdings

If you’re interested to see (1) a snapshot of live portfolio (including weightings) and (2) a TLDR summary of each of my positions (including my price targets, entry and sell recommendations and what I am doing with each position), then continue reading!

__________________________________________________________________________

Welcome to Orange Grove Capital

I turned $80K into $1M in 4 years looking for asymmetric, high-conviction bets before they became consensus. I created this Substack to publish my research and thought process in the hopes that people benefit from my investing framework.

____________________________________________________________________________

Read more

Despite all the clamor, photonics stocks have been crushed lately ($SIVE down almost 60% and $AAOI down almost 50% from ATH). Memory stocks have equally been losing steam ($MU down almost 25% since it reported blowout Q3 earnings). I could go on and on.

I’m not throwing any punches, I’m just saying that it’s important not to get caught up in all the hype. Many of these stocks are priced to perfection and any blip in the roadmap or sentiment will lead to significant corrections. Can it go up 2 or 3x from here? Sure. But does it fulfil my criteria of “relatively low risk and disproportionately high returns”? Absolutely not. So that got me thinking: what traits are actually necessary for a business to fulfil this criteria ?

By the end of this article, you should understand the four traits which are necessary for a business to fulfil this criteria of providing “relatively low risk and disproportionately high returns”. Then, in that light, you will see for yourself why there is a glaring opportunity I need to grab with both hands.

1. AI should not challenge its moat.

The markets have shown you that sentiment can flip on a whim. Claude Opus was supposedly going to change the world six months ago and now LLMs are back to being considered a commodity (with open source now a very viable path forward). SaaS was also down in the dumps six months ago and starting with Data Dog, these are now starting to get bid. Palantir for example has risen 15% these past weeks without any significant news - the market was just being erratic because it simply did not know what the future held.

The truth is, no one knows how this AI race is going to play out. For a stock to satisfy the “low risk” criteria, AI should not be able to challenge its moat in any significant way (or even have the appearance of challenging its moat).

The areas where AI cannot challenge a moat are three-fold: (1) where an industry is heavily regulated; (2) where human intercourse and involvement cannot be done away with; and (3) where there is an otherwise physical component to the business. AI challenges everything else.

2. AI should be a perfect fit for its business

It goes without saying that AI makes many businesses more efficient. Even using ChatGPT to produce an infographic or written content can make businesses relatively more efficient. I’d say this alone increases efficiency by 20-30%.

But to satisfy the “disproportionately high returns” criteria, AI should fit the business like a glove. I’m not referring to saving efficiencies or improving operations here and there in a piecemeal manner. I mean the entire business should be fundamentally bolstered and in an ideal situation, replaced by AI.

3. It should be fundamentally undervalued based on current projections

When I say fundamentally undervalued, I mean it should be undervalued on its current and openly-declared trajectory. Any upside should not have to depend on any unexpected beats or management achieving exceptional results.

For example, if Applied Optoelectronics ($AAOI) achieves its target of $471/m in revenue by mid 2027, that implies a $5.5B run rate, which makes the stock stupidly cheap now at an implied 1.8x 2027 P/S. If $AAOI achieves these targets, I easily see a 2-3x from here. The issue is that management has failed to achieve its objectives several times before, and achieving these targets require a herculean effort with absolutely nothing going wrong. It therefore doesn’t fit this criteria for me.

Another example is Fluence Energy ($FLNC). It’s an altogether great business - they’ve contracted two new hyperscalers and have been growing very well. But in order for the true bull case to be achieved (from my perspective), they would need to beat FY 2027 EBITDA margins significantly. While I think they can beat, that is not the point. The point is that $FLNC would require a significant beat of analysists consensus / projections - it therefore doesn’t fit this criteria for me.

To satisfy the “relatively low risk” criteria, the stock should be fundamentally undervalued based on current projections (that is, projections already provided by management ann require nothing out of the ordinary).

4. There should be no questions about management

In order to satisfy the “relatively low risk” criteria, management should not have done anything questionable (even in the slightest).

For example, FuelCell Energy ($FCEL) is a great company. Yet, they recently came up with an offering despite having the balance sheet being strong enough for them to meet their expansion needs and get to EBITDA breakeven.

Equally, $IREN is promising but granting 9 million RSUs each to its to co-CEOS on a time-based (and not performance-based) schedule just reeks to me.

When I say “even in the slightest”, I mean it. If you’ve been following me for some time, you would know that I regard Hims and Hers’ ($HIMS) management very highly. Yet, they wouldn’t satisfy this criteria of having not done anything questionable. Launching a weight loss pill with the same active ingredients as Wegovy and consequently getting sued by Novo Nordisk ($NVO) is questionable. It would mean $HIMS is crossed off this list.

Why I need to own Lemonade Inc ($LMND)

When I really think about which company satisfies all of these four criterion - assuming any company does at all - all roads lead to Lemonade Inc. ($LMND).

1. AI does not challenge its moat

LMND is an insurance company and the insurance industry is heavily regulated. I don’t think there is a need to spell this out any further: Regulation necessitates human approval, oversight and interactions (that is the very point of regulation), which means $LMND’s regulatory moat will not be threatened by AI in the near future.

2. AI is a perfect fit for its business

Insurance involves a probability theory that is being monetized. It is just the ability to look at past patterns and convert them to future patterns. And AI is perfect for that because at its core, AI is just math: pattern recognition, probability, and prediction at scale.

Without going into the details (you can find my full deep dive below in you’re interested in understanding the business holistically), here are two examples of this.

First, look at how $LMND’s gross loss ratio (GLR) is cratering downwards. GLR is basically the percentage of premiums an insurer pays out in claims and a good AI model will help insurance companies price premiums properly such that they get the most amount of customers (and premiums) but also the lowest GLRs. A good industry GLR is 75% and $LMND has blitzed through this as you can see.

Second, look at how $LMND’s in-force premiums (IFP) has grown at roughly a 30% CAGR with a flat to modest growth in operating expenditure. This is what I meant by AI fitting like a glove - all these probabilistic models and insurance claim assessments can be run by LLMS now, heavily reducing the need for humans in the loop.

3. It is fundamentally undervalued based on current projections

Based on management’s projection alone - that is, without any guidance beat - $LMND will end 2030 with $720m in profits and end 2034 with $2.4B in profits. If you slap a 30x P/E multiple on both of those figures and discount it to today’s NPV at 12.5% discount rate, you get a NPV of $12.5B (based on 2030 projections) to $24B (based on 2034 projections).

That means management simply needs to achieve its projections for the business to be worth 2x to 4x its current price today.

And that ties us nicely into the next section, which is that management is excellent and has historically not missed any guidance - there is therefore little reason to believe they won’t execute and achieve its guidance.

4. Management has been excellent.

If you look at $LMND’s financials today and all of their past guidance, it has all been with notable precision.

$LMND has never missed guidance, and in fact they’re ahead of what they’ve historically told investors. The ability to predict their business has been nothing short of spectacular.

This gives me great confidence in management and until they do something which makes me think otherwise, I will continue to place my trust with them.

All roads for me lead to $LMND. I need to grab this opportunity with both hands.

If you’re interested in reading my $LMND deep dive, where I break down the business, its advantages, why it is perfect for an AI-era, valuation and personal targets, you can find it here: https://substack.com/@ogcapital25/p-200720938)

Remember, none of this is financial advice and you should do your own due diligence before you invest.

___________________________________________________________________________

⚡ If you enjoyed the post, please feel free to share with friends, drop a like and leave me a comment.

You can also reach me at:

X: Orange Grove Capital

Disclosure

These are opinions only of the individual author. The contents of this piece do not contain investment advice and the information provided is for educational purposes only and no discussions constitute an offer to sell or the solicitation of an offer to buy any securities of any company. All content is purely subjective and you should do your own due diligence.

Orange Grove Capital makes no representation, warranty or undertaking, express or implied, as to the accuracy, reliability, completeness or reasonableness of the information contained in the piece. Any assumptions, opinions and estimates expressed in the piece constitute judgments of the author as of the date thereof and are subject to change without notice. Any projections contained in the Information are based on a number of assumptions as to market conditions and there can be no guarantee that any projected outcomes will be achieved. Orange Grove Capital does not accept any liability for any direct, consequential or other loss arising from reliance on the contents of this presentation. Orange Grove Capital is not acting as your financial, legal, accounting, tax or other adviser or in any fiduciary capacity.

The Headline

PENG just released its Q3 results and delivered a monster double beat and guidance raise.

- Revenue: $479m, up 48% YoY (consensus $405m)
- EPS: $0.84 (consensus $0.54)
- Raised FY 2026 EPS guidance to $2.6 +- $0.05 (this is conservative IMO)

The Thesis

In my PENG deep dive here, I shared that PENG could be valued at roughly $8B MC if we apply DELL’s P/E multiples. In that analysis, I used the following assumptions:

“If you take the midpoint of Penguin’s guidance for 2026, then its revenues for FY 2026 can be assumed to be:

1. Advanced Computing: $519M (cf. projected to be down 25% to down 15% YoY)

2. Integrated Memory: $789M (cf. projected to be up 65% to 75% YoY)

3. Optimised LED $225M (cf. projected to be down 15% to down 5% YoY)

4. Total: $1.53B

In a bullish case, FY 2027 can see the following revenues in light of everything we’ve discussed:

1. Advanced Computing: 35% growth → $700M

2. Integrated Memory: 45% growth → 1.15B

3. Optimised LED: -10% growth. → $200M

4. Total: $2.05B”

Let’s review these numbers in light of PENG’s Q3 earnings call to validate or challenge the thesis.

Read more

Having received several requests from my paid subscribers, here’s a portfolio update for the month of June.

Performance

Here is my performance since 5 June 2026 (the date I started this Substack)

• Returns since 5 June 2026: +5.41%

• S&P 500 Returns since 5 June 2026: -1.16%%

*Chart generated by ChatGPT

As you can see, we significantly outperformed the S&P, which was really surprising considering how volatile the recent week has been (and how high beta the names I hold are). You can see just a few days ago we were beating the S&P 500 by a whole 10%. Notwithstanding, I am very pleased with these results and look forward to more favourable environments.

Now to all the details.

Portfolio Overview

Live Holdings

If you’re interested to see (1) a snapshot of live portfolio (including weightings) and (2) a TLDR summary of each of my positions (including my price targets, entry and sell recommendations and what I am doing with each position), then continue reading!

__________________________________________________________________________

Welcome to Orange Grove Capital

I turned $80K into $1M in 4 years looking for asymmetric, high-conviction bets before they became consensus. I created this Substack to publish my research and thought process in the hopes that people benefit from my investing framework.

If you’re interested in a detailed breakdown of that framework - including how I think about business quality, valuation, business sizing, risk management and what I look for in potential multi-baggers - you can download it here:

My full investing framework

____________________________________________________________________________

Read more

Investors already understand that AI is not just a GPU story. HBM, DRAM, NAND, SSDs, storage, and advanced packaging have all been pulled into the AI infrastructure boom.

But the recent strength surrounding memory names created the obvious pushback: The trade had already run too far, too fast. Memory margins looked to be peaking. Maybe investors were paying too much for earnings that would eventually mean-revert.

That was the debate heading into Micron’s earnings.

Then Micron reported.

And the company did not just beat expectations. It blew straight through them.

Fiscal Q3 2026 revenue came in at roughly $41.5 billion, versus estimates around $35.6 billion to $36.3 billion. Adjusted EPS came in at $25.11, versus expectations around $20.50 to $21.05. GAAP net income was $28.2 billion. Operating cash flow was $25.4 billion. Gross margin reached the mid-80s.

Then came the guide.

Micron guided fiscal Q4 revenue to roughly $50 billion, versus Wall Street expectations closer to $43 billion, with gross margin expected around 86%.

That is the kind of print that changes the conversation.

It means the trade just got validated at a much higher level of earnings power than many investors were prepared to underwrite.

The Old Memory Playbook Is Being Challenged

Historically, investors have been trained to treat memory as a cyclical business.

When pricing gets strong, margins expand. When margins expand, supply eventually comes in. Then pricing weakens, margins compress, and the cycle resets. That mental model has worked for a long time, which is why the market has always been reluctant to give memory companies the same kind of multiples as other semiconductor leaders.

But AI is proving that playbook wrong.

The AI buildout is not just compute-intensive. It is increasingly memory-intensive. Training large models, serving inference, handling long-context workloads, running agentic AI systems, and scaling enterprise AI infrastructure all require enormous amounts of memory and storage.

The market has definitely woken up to the idea that GPUs are not the only scarce asset in the AI stack. You also need HBM. You need DRAM. You need NAND. You need storage. You need advanced packaging. You need the entire memory ecosystem to scale with compute.

Micron’s quarter was the clearest proof point yet that this is not just a narrative.

The Shortage Has No Easy Fix

The most important part of the quarter may not even have been the numbers. It may have been the commentary.

Micron CEO Sanjay Mehrotra put it very clearly:

“With respect to supply, our customers are recognizing that supply shortages in memory and storage will take considerable time to improve. We currently do not have line of sight as to when memory supply will be able to catch up with increasing demand.”

That quote is the whole thesis.

This is not management saying demand is temporarily strong. This is management saying customers are recognizing that memory and storage shortages will take considerable time to improve.

Even more importantly, Micron is saying it does not currently have line of sight as to when supply catches up with demand.

That is why the market reacted so aggressively.

If this were just a one-quarter beat, yes, the stock would probably move. But this was bigger than that. This was a company telling investors that the supply-demand imbalance could remain tight for a long time, while also putting up numbers that looked almost unthinkable for a memory company not long ago.

But High Memory Margins Create Another Winner

There is another angle here that matters.

If memory margins are this high, and if supply remains tight for longer than expected, customers will not just accept the pain forever. They will look for ways to get more out of the memory they already have.

In the following sections I discuss which single company will benefit massively from Micron’s earnings and why.

+++

Welcome to Orange Grove Capital

I turned $80K into $1M in 4 years looking for asymmetric, high-conviction bets before they became consensus. I created this Substack to publish my research and thought process in the hope that others can benefit from my investing framework.

If you’re interested in a detailed breakdown of that framework - including how I think about business quality, valuation, business sizing, risk management and what I look for in potential multi-baggers - you can read it here:

My full investing framework

+++

Read more

Valuation

Hims & Hers Health ($HIMS) targets at least $6.5 billion in revenue and $1.3 billion in adjusted EBITDA by 2030. From its current financials, $HIMS needs to roughly achieve 22% revenue and 32% adjusted EBITDA CAGR to achieve this 2030 target.

As far as I know, $HIMS has not often missed its own internal guidance (not to be mistaken with wall street expectations). On the contrary, $HIMS has typically smashed projections - for example, it previously had a 2025 target of at least $1.2B in revenue and $100M in Adj. EBITDA but its FY 2025 revenue was ~$2.35B and Adj. EBITDA was $318M, which was a significant beat. All in all, I’d say $HIMS will likely achieve the 22% revenue and 32% adjusted EBITDA CAGR considering $HIMS historical performance.

If a company grows at this rate, it can generally expect a market cap / adjusted EBITDA* of 16-20x in the base case. That puts it at a market cap of $20B in 2030, from $6.29B now - an upside of ~3x in 4 years. Say you discount it further for regulatory risks, and maybe you get a ~2 to 2.5x in 4 years, not too shabby, but not exactly super. This is the “base” case out of the way.

*Note, I’d typically use EV / Adjusted EBITDA rather than P / Adjusted EBITDA, because EBITDA belongs to both debt and equity holders. But since HIMS has a relatively clean balance sheet, and standard valuation is not what I am focused on, market cap / Adj. EBITDA can be a quick-and-dirty proxy.

In the following sections I discuss (1) the peptide opportunity which will provide considerable upside, (2) other assets / advantages HIMS has, (3) risks the business faces, (4) the absurdly high short interest, and (5) my overall thoughts.

+++

Welcome to Orange Grove Capital

I turned $80K into $1M in 4 years looking for asymmetric, high-conviction bets before they became consensus. I created this Substack to publish my research and thought process in the hope that others can benefit from my investing framework.

If you’re interested in a detailed breakdown of that framework - including how I think about business quality, valuation, business sizing, risk management and what I look for in potential multi-baggers - you can read it here:

My full investing framework

+++

Read more

But I think that interpretation misses the more important signal.

This was not a Chair trying to tee up an aggressive tightening cycle…

Read more

Palantir at $17 (ATH $223) did not look obvious. Sivers Semiconductors at 8 SEK (ATH 104 SEK) did not look obvious. AMD at $80 (ATH $550) did not look obvious. ASTS at $20 (ATH $135) did not look obvious.

At my respective entry points, each company had a convincing bear case. Palantir was “just a consulting company.” AMD was “the second choice”. ASTS satellites were “never going to work”. Sivers had “no credible CPO pipeline”.

But that’s where the opportunity lives. By the time everyone agrees a company is high quality, the stock price usually already reflects it. That’s why the real opportunities tend to show up earlier - when the story is still messy, the numbers are still misunderstood, and the market has not fully connected the dots.

This article is not about the stocks after they worked. It is about what all of them looked like before the market cared.

1. Shifting Fundamentals

The best winners were improving under the surface before the stock fully responded.

The key is to realise that the market often prices companies based on what they used to be, not what they are becoming. When the narrative is stale but the business is changing, the stock can become mispriced. Those set ups are a dream for asymmetric investors.

You should not look for companies with no problems. Instead, you should look for companies where the problems are obvious, but the future (positive) trajectory of the business and upside is not yet clear to the average investor.

Just look at AMD. While the market still viewed it as the weaker number two to Intel, underneath the surface, the business was changing: Ryzen had made AMD competitive again in client CPUs, while EPYC was gaining real traction inside data centers. The market was not just missing that AMD had better CPUs — it was missing where those CPUs were starting to win, and how that could change AMD’s earnings power over time. As we’re seeing now,

$AMD

ran from $80 to $550, and I think it still has legs to run.

A current example where I think the market is missing a similar shift in fundamentals is Penguin Solutions (

$PENG

). While the market still seems to view Penguin through its old identity as a lower-growth memory and hardware business, the company is repositioning around AI factories, high-performance compute integration, memory architecture, and full-stack infrastructure services. If that transition continues, Penguin may not deserve to be valued like a legacy hardware company, but as an emerging AI infrastructure platform helping enterprises, neoclouds, and sovereign customers deploy complex compute environments at scale. I unpacked the full thesis on Penguin separately here:

2. Improving Operating Leverage

My best winners had the potential to grow revenue faster than expenses. And that matters: If a company can scale without costs rising at the same pace, earnings power can change quickly.

This is why I care about gross margins, customer acquisition efficiency, retention, product expansion, and fixed-cost leverage.

The market often focuses too much on whether a company is profitable today.

I care more about whether it can become meaningfully profitable at scale tomorrow.

There is a big difference between a company losing money because the model is broken and a company losing money because it is investing into a large opportunity.

One example of a company I think will benefit tremendously from rapidly improving operating leverage, and is poised to be meaningfully profitable at scale in the AI-era is Lemonade Inc. ($LMND). Just look at below chart - they are growing their In-Force Premiums at a 30% CAGR, but are reducing their employee count with thanks to AI efficiencies. They are projected to be full year Adj. EBITDA profitable in 2027 and from then one will likely be an amazing profit-making machine because of such efficiencies. I unpacked the full thesis on Lemonade separately here:

3. Real Optionality

Each of these companies also had more than one way to win.

This could be a core business improving, a new product gaining traction, a new market opening, a cycle turning, OpEx improving etc.

Optionality matters because the biggest winners rarely come from one neat assumption playing out perfectly. They come from more things going right than things going wrong. If this happens, you’ll have found yourself a multi-bagger.

The key is to find situations where the market is giving little credit to optionality that could become very valuable. That is when upside can surprise people.

Case in point: When Palantir was $17, everyone thought it was simply a “government contractor”. Yet, when you understood that the Ontology was bedrock of Palantir’s platform (and which they were using for government), you would have understood that there was no reason Palantir could not replicate its Ontology for commercial partners. If you truly understood that, you would have seen that it was entirely possible for commercial to be a huge market for Palantir and investors simply were not pricing this in. Indeed, commercial is massive for Palantir today, almost matching government revenues - further, US commercial is growing 133% YoY while US government is growing 84% YoY today. That “optionality” turned into a real bull case for Palantir.

4. Uncomfortable Setup

This may be the most important common trait.

The best ideas were not comfortable at first. They had obvious risks, loud bears, valuation concerns, bad headlines, and required a ton of patience.

But discomfort is usually the reason the opportunity exists. And more importantly, that discomfort can be tamed by doing deep due diligence on the stock and forming your own deep and grounded understanding.

If a stock has no controversy, no uncertainty, and no debate, there is usually no edge.

Remember, the goal is not to avoid uncertainty.

The goal is to understand which uncertainty is temporary and which uncertainty is fatal.

There is a big difference.

The Real Lesson

My biggest winners did not work because I knew the future.

They worked because the market was underestimating the size, duration, or quality of the opportunity.

That is what I am constantly looking for: a stale narrative, improving fundamentals, operating leverage, underappreciated optionality, and a setup that feels uncomfortable but not broken.

These combinations are rare.

But when they show up, they can be very powerful.

Orange Grove Capital is where I document this process in real time: the companies I am studying, the traits I am looking for, the risks I am underwriting, and the setups I think the market may be mispricing before consensus catches up.

Free subscribers get my investing frameworks, market thoughts, thesis previews, and occasional company write-ups. Paid subscribers get the deeper research: full company deep dives, valuation work, thesis updates, watchlist notes, and the process behind the names I am actively studying.

The goal is simple: find asymmetric public-market opportunities before consensus catches up.

Not because I had a perfect batting rate - I did not. It happened because the vast majority of my asymmetric bets carried my portfolio - companies where I believed the market was underestimating the magnitude, duration, or quality of the opportunity; and being proven correct.

Some of the names that shaped my process include Palantir, AMD, SIVE, LMND and ASTS.

Each situation was different. But the common thread was the same:

1. The market had a narrative.

2. I believed the underlying reality was better than that narrative suggested.

3. The underlying reality proved to be (significantly) better than the narrative suggested.

That is the core of my investing process.

I certainly do not try to trade every headline. I do not even not try to predict next quarter perfectly. And I certainly do not try to own the safest, cleanest, most obvious companies at any price. Instead, I try to find public companies where the market is wrong in a way that matters.

Usually, that means the market is underestimating one of five things:

1. Management capabilities

2. How profitable the company can become

3. How strong the competitive position (i.e. moat) really is

4. How much optionality exists

5. How quickly the narrative can change

The best opportunities rarely look obvious at first. And you do not want them to. If it were obvious, there simply would be no upside (the valuation would already reflect the great business).

The real opportunity tends to live in the gap between what a company is today and what it could become if a few important things go right. That gap is where upside lives and what I constantly chase.

What I Look For

I love companies with misunderstood narratives.

The market loves simple labels: “unprofitable growth stock,” “will get displaced by AI,” “too expensive.”

Sometimes those labels are right. Sometimes they are lazy. Sometimes they are just wrong. And that’s when I do the DD and pounce.

My Research Framework

When I study a company, I usually ask six questions:

1. What does the market currently believe?

2. Where might consensus be wrong?

3. Is this actually a good business?

4. What could the company earn three to five years from now?

5. What could force the market to reprice the stock?

6. What evidence would change my mind?

That last question matters most.

A good thesis should be flexible. If the facts change, the conclusion should change too.

I am not looking for certainty. I am looking for skew.

The goal is to find situations where the upside potential meaningfully outweighs the downside risk - and where the market is mispricing the path from what the company is today to what it can become. When you find that, lean heavy into it: it’s what helped me 10x my portfolio in 4 years.

+++

My Full Investing Framework

The article above is a high-level overview of my investing framework. In the full version, I break down the actual process I use to evaluate businesses, think about valuation, manage risk, size positions, and identify the traits that have shown up across some of my best investments.

If you want the deeper breakdown, you can read it here:

My full investing framework

+++

What I Am Building With Orange Grove Capital

Orange Grove Capital is where I document this process in public.

This is not financial advice. Nothing I write should be treated as a recommendation to buy or sell any security. Everyone has different goals, risk tolerance, liquidity needs, and time horizons.

My goal is to share how I think.

Free subscribers on Substack will get selected essays, investing frameworks, market thoughts, short thesis previews, and occasional company write-ups.

Paid subscribers on Substack get the deeper research product: full company deep dives, valuation work, bull and bear case breakdowns, thesis updates, watchlist notes, and more detailed thoughts on what I am studying, buying, avoiding, or rethinking.

The best investments often feel uncomfortable before they become obvious. By the time a company is clean, loved, de-risked, and widely understood, much of the upside may already be gone.

I want to spend my time earlier in the process - when the narrative is still messy, the fundamentals are beginning to inflect, and consensus may be too dismissive.

That is what Orange Grove Capital is about.

If this is something that interests you, do subscribe to the page and follow along the journey.

The reason I think this is worth sharing is simple: the discussion gives a very direct look into how Lemonade is positioning itself at a critical moment.

The broader insurance industry is seeing slower growth. Customer acquisition costs remain high. Incumbents are trying to layer AI onto legacy infrastructure. And yet Lemonade continues to argue that it is playing a fundamentally different game.

Whether one agrees with the stock, the valuation, or the execution risk, the strategic message from Schreiber was clear: Lemonade believes its technology stack gives it structural advantages in pricing, automation, customer acquisition, and cross-selling.

Before the full transcript, these were the key points that stood out to me.

Read more

After weeks of relentless upward pressure, the S&P 500 fell 2.6% and the Nasdaq dropped 4.5% on the back of a strong jobs report, marking the sharpest weekly setback in years. Small caps were hit as well. The reaction was predictable: every correction in a powerful bull market invites the same question.

Is this the beginning of a bear market?

The better answer is probably no. This looks less like a market top and more like the start of a necessary rotation.

That distinction matters. A bear market is a broad deterioration in earnings, credit, employment, liquidity, and investor confidence. A rotation is different. It is a transfer of leadership. Capital does not leave the market entirely; it moves away from the most crowded winners and into areas where the risk-reward has become more attractive.

That is what appears to be happening now.

The Easy AI Trade Is Over

The past several months were defined by a powerful catch-up trade in AI infrastructure. Semiconductors, memory, networking, power, optical components, and data-center-linked hardware all moved as investors converged on the same conclusion: the AI buildout is real, massive, and earlier than expected.

The problem is not that the AI story is fake. The problem is that everyone now knows the story.

That changes the odds.

A few months ago, the market was still debating whether AI capex would generate returns. Investors questioned whether the spending cycle was a bubble. Then the narrative changed. AI agents, inference demand, tokens-per-watt, Vera Rubin, Blackwell, optical networking, memory bottlenecks, and data center power constraints became mainstream institutional talking points.

The market moved from disbelief to discovery. Now it is moving from discovery to digestion.

That means the “buy anything with AI exposure” phase is probably over. The next phase will be more selective, more volatile, and more fundamental. Investors will have to distinguish between companies with durable earnings power and companies merely benefiting from a temporary bottleneck.

The AI buildout has not ended. The easy money phase has.

The Technical Damage Was Real, but Not Fatal

The Nasdaq and S&P both closed below their 20-day moving averages as a result. That matters because these indices had barely touched short-term support during the recent rally. Momentum was extended, positioning was crowded, and volatility had been suppressed.

A break of the 20-day does not automatically signal a bear market. It usually signals a change in tempo.

The weekly RSI readings had reached multi-year highs. In many major tops, momentum begins to diverge before the broader market breaks down. That is not yet obvious here. A more typical topping process would involve a rebound, new highs in price, and weaker internal momentum. We may still get that.

The bigger warning came from beta. High-volatility, high-beta stocks were hit hard, with one major beta factor experiencing one of its sharpest single-day declines in decades. That is a cleanse. It suggests that the most speculative parts of the market were forced to reset.

But cleanses are not the same thing as collapses.

On the same day the S&P was sharply lower, five sectors finished higher. Capital rotated into areas like healthcare and financials. That is not what indiscriminate liquidation looks like. It is what leadership change looks like.

Why This Does Not Look Like a Sustained Bear Market

The macro backdrop does not resemble the start of a durable bear market.

Earnings revisions are moving higher, not lower. Profit margins are near record highs and have not shown the kind of mean reversion that historically precedes recessions. Credit spreads remain tight. Jobless claims have not begun the kind of persistent rise that usually signals labor-market stress. Manufacturing indicators and capital spending remain supported by the AI infrastructure cycle.

A sustained bear market normally requires more than stretched valuations. It requires deterioration in the economic machine.

In 2000, the Nasdaq peaked before the broader economy cracked. But the broader bear market became much more serious once earnings rolled over, margins compressed, claims rose, and the real economy weakened. Something similar happened before other major downturns. Credit usually leads. Labor follows. Earnings confirm.

That is not the current setup.

Valuation is the missing risk variable. The market is not cheap, and some pockets are clearly priced for perfection. But valuation alone rarely causes a sustained bear market without a macro catalyst. It can cause corrections, rotations, and multiple compression. It does not usually create a durable downturn unless the fundamentals begin to validate the decline.

So the base case is not “everything is fine.” The base case is more nuanced: the market is vulnerable to rotation, but the evidence for a full bear market is not yet present.

The following sections discuss (1) what I think about the memory sector currently, (2) where the next bottleneck likely lies, (3) next areas to look at, (4) what I think you should start to consider owning.

Memory Is No Longer the Best Place to Take AI Risk

The most crowded part of the AI trade may now be memory.

The bull case is simple: AI models require enormous amounts of high-bandwidth memory, demand is exploding, supply is constrained, and pricing power could persist for years. On paper, the numbers look compelling.

The danger is extrapolation.

It is tempting to take token demand, project it forward, and translate it directly into memory demand. But that may be too linear for a nonlinear technology cycle.

Several things can go wrong for memory makers. Data center bottlenecks could delay deployment. Power constraints could slow buildouts. Safety concerns could affect the pace of frontier model development. Open-source and edge models could reduce dependence on centralized compute. Algorithmic improvements could reduce memory intensity. Recursive self-improvement could accelerate the discovery of architectures that need less memory to generate intelligence.

The more expensive memory becomes, the greater the incentive to engineer around it.

That is why memory still deserves a cyclical discount. It may remain fundamentally strong, but the risk-reward is no longer as asymmetric as it was when investors were skeptical. Once a bottleneck becomes universally accepted, the market starts paying for the solution before the solution is fully realized.

The better AI infrastructure opportunities may now sit elsewhere: optical networking, co-packaged optics, power architecture, batteries, grid support, and the application layer.

The Next Bottleneck Is Physical

AI may feel like software, but the next stage of the buildout is deeply physical.

The industry is trying to turn electricity into tokens. That requires chips, memory, networking, cooling, land, substations, transmission, batteries, and backup power. Each incremental gigawatt of AI data center capacity requires staggering amounts of semiconductor and infrastructure investment.

That creates a new kind of cycle.

The old economic cycle was built around labor and capital. Companies borrowed, built, hired, expanded, slowed, compressed margins, and laid people off. Households then transmitted the downturn through consumption, housing, autos, and credit.

AI changes that structure. Digital workers do not buy homes, lease cars, take vacations, or raise families. They consume compute, memory, electricity, and data. The bottlenecks shift from labor and credit to chips and energy.

That does not mean cycles disappear. It means they mutate.

An AI recession may not look like a traditional recession. It may show up as power constraints, delayed data centers, falling utilization assumptions, margin pressure in infrastructure suppliers, or a mismatch between capex and monetization.

This is the central tension: the capex cycle is moving as if demand is immediate, while the enterprise adoption cycle still has to pass through trust, integration, measurement, compliance, and return-on-investment hurdles.

The first AI phase was about securing capacity. The next phase will be about proving payback.

From Capacity Land Grab to ROI Discrimination

Another possibility, and one to keep in mind, is that the market might be moving from a capacity land-grab phase to an ROI discrimination phase.

That is a major shift.

During the land-grab phase, the winners are the suppliers of scarce inputs: GPUs, memory, networking, data center equipment, and power infrastructure. During the ROI phase, the winners are companies that can use AI to improve margins, accelerate discovery, reduce costs, create new products, or expand revenue without proportionally increasing labor.

This is why the application layer may become more important.

Healthcare to me is one compelling area. Drug discovery, protein design, clinical trial optimization, synthetic biology, diagnostics, and personalized medicine all sit at the intersection of proprietary data and AI acceleration. Unlike generic models trained on the public internet, specialized models trained on private scientific data may create more defensible value.

In that framework, a company with proprietary biological data, scaled revenue, high margins, deep R&D infrastructure, and the ability to fund internal AI factories becomes more than a pharmaceutical company. It becomes an applied AI platform.

That is the kind of shift investors should be watching.

The question is no longer simply: who sells the picks and shovels?

It is also: who uses the picks and shovels to create extraordinary returns.

Energy and Batteries Deserve More Attention

The AI buildout cannot happen without power.

That makes energy an increasingly important portfolio hedge. If oil prices remain elevated, geopolitical risks persist, or power demand continues to surprise, traditional energy companies may offer defensive value at a time when high-growth AI infrastructure names become more volatile.

Batteries also deserve more focus. Data centers require flexible power, backup systems, and grid support. As hyperscalers compete for energy access, battery storage becomes a critical piece of the infrastructure stack. It is not just a renewable energy story. It is a reliability story.

If data centers are the new factories, then energy storage is part of the factory floor.

The market is beginning to recognize that power is not a side issue. It is central to AI scaling. This is why I will be covering Fluence (FLNC) shortly.

What to Own Now

The AI trade is not over. It is becoming more complicated.

The market is likely to reward selectivity over blanket exposure. Areas that still look attractive include:

AI application winners with proprietary data and clear ROI.

Optical networking and co-packaged optics, where demand is early and competition may be less mature.

Power infrastructure, energy storage, and grid-related assets tied to the physical constraints of AI scaling.

Energy companies that provide a hedge against geopolitical and inflation risks.

Healthcare companies positioned to turn AI into measurable scientific and commercial acceleration.

Memory may still work over a longer horizon, but the risk-reward is less compelling after a parabolic move and broad consensus adoption. The same applies to any AI supplier where the market has already priced in years of flawless execution.

The next phase belongs to companies that can prove value, not just absorb capex.

The Bottom Line

This does not look like the start of a sustained bear market. It looks like a leadership rotation after an extraordinary run.

The macro backdrop remains too strong to justify a broad bearish conclusion. Earnings are rising. Margins are resilient. Credit is calm. Labor has not cracked. Capital spending remains supported by AI.

But within the market, the message is changing.

The AI infrastructure trade has matured. The most crowded parts of the market are now vulnerable to disappointment. Investors should expect more two-way action, more dispersion, and more scrutiny around return on invested capital.

The fireworks phase is over.

The buildout is not.

The next opportunity will not come from chasing the loudest part of the trade. It will come from identifying where AI moves from promise to productivity, from capacity to cash flow, and from infrastructure spending to real economic return.

That is where the next cycle begins.

But that is not the same thing as saying optical networking is delayed. It is not the same thing as saying pluggables are dead. In fact, I think the opposite may be true.

If full scale-up CPO takes longer to ramp, the near-term winners are likely the companies positioned in the bridge architectures that actually have to ship while the industry transitions. That means pluggables. That means NPO. That means 800G and 1.6T transceivers. And that is exactly where Applied Optoelectronics is focused.

So while the market heard “CPO delay” and sold everything optical, I see a buying opportunity in $AAOI.

The following section discuss: (1) why the market confused CPO with all optics, (2) why AAOI is more than a CPO play, (3) bridge architectures (and why they benefit) and more.

+++

Welcome to Orange Grove Capital

I turned $80K into $1M in 4 years looking for asymmetric, high-conviction bets before they became consensus. I created this Substack to publish my research and thought process in the hope that others can benefit from my investing framework.

If you’re interested in a detailed breakdown of that framework - including how I think about business quality, valuation, business sizing, risk management and what I look for in potential multi-baggers - you can read it here:

My full investing framework

+++

Read more

That is where Penguin Solutions ($PENG) sits.

For years, the company was easy to ignore. Under its former identity as SMART Global Holdings, it was often viewed as a lower-growth hardware and memory business with exposure to cyclical end markets. The market saw components, manufacturing, and legacy segments. What it missed was the slow construction of an infrastructure platform aimed at one of the most important bottlenecks in AI: deploying complex compute environments reliably, quickly, and at scale.

That transition is now becoming much harder to overlook.

Penguin Solutions is no longer best understood as a memory company or a conventional hardware integrator. It is becoming a full-stack AI factory enabler for enterprises, neoclouds, and sovereign customers that want advanced AI infrastructure but do not have the internal engineering depth of a hyperscaler.

The investment case is not simply that AI infrastructure demand is growing. Everyone knows that. The more interesting point is that Penguin has positioned itself at the intersection of three scarce capabilities: high-performance computing integration, memory architecture, and operational software.

From Legacy Hardware to AI Factory Platform

Penguin’s transformation did not happen overnight. The foundation was laid years before the company changed its name.

The first major step came in 2018, when SMART Global Holdings acquired Penguin Computing. That deal gave the company a foothold in high-performance computing, AI, machine learning, specialty compute, and large-scale cluster design. It was not just a brand acquisition. It brought engineering DNA that would later become central to the company’s AI infrastructure strategy.

The second important step came in 2022 with the acquisition of Stratus Technologies. Stratus added experience in high-availability computing, fault tolerance, and mission-critical infrastructure. That mattered because AI factories are not simple server rooms. They are dense, expensive, power-hungry systems where downtime, poor utilization, and slow deployment can destroy the economics of the project.

Then came the formal reset. In October 2024, SMART Global Holdings became Penguin Solutions and began trading under the ticker PENG. The rebrand was more than cosmetic. It marked the company’s attempt to define itself around end-to-end infrastructure rather than legacy product categories.

The market is still catching up to that change.

The Strategic Pivot: Avoiding the Hyperscaler Trap

One of the most important parts of the Penguin story is what the company is choosing not to chase.

Hyperscale customers can create huge revenue numbers, but they usually have the procurement power to compress margins. They often have their own engineering teams, their own reference architectures, and their own ability to optimize deployments. For a smaller integrator, that can mean large volumes with limited profitability.

Penguin’s pivot is different. The company is leaning into customers that need more help, not less.

That means enterprises, neoclouds, and sovereign AI buyers.

These customers want AI infrastructure they can control, but they often lack the internal capability to design, deploy, tune, and manage complex GPU clusters on their own. They need validated architectures, memory optimization, software orchestration, cooling expertise, and managed services. That creates a more attractive value proposition for Penguin because the company is not merely reselling hardware. It is packaging expertise.

This is why the “AI factory” framing matters. Customers are not just buying servers. They are buying time to production, uptime, utilization, energy efficiency, and a working infrastructure layer that can support real workloads.

In the following sections I talk about Penguin’s offerings, its key advantages, its technical and partnership advantages and how I am thinking about valuation.

+++

Welcome to Orange Grove Capital

I turned $80K into $1M in 4 years looking for asymmetric, high-conviction bets before they became consensus. I created this Substack to publish my research and thought process in the hope that others can benefit from my investing framework.

If you’re interested in a detailed breakdown of that framework - including how I think about business quality, valuation, business sizing, risk management and what I look for in potential multi-baggers - you can read it here:

My full investing framework

+++

Read more

Applied Optoelectronics Inc (AAOI) is a high-speed optical networking company that makes transceivers and related optical components used inside data centers, cable broadband networks, telecom systems, and fiber-to-the-home deployments.

Wait, what does all of that mean, you ask? Okay, in plain English, AAOI sells the optical (i.e. related to light / photons) hardware that helps data move quickly between servers, switches, and networks.

While the company has been around since 1997, the reason investors are paying attention now is that AI data centers are rapidly moving from 100G/400G links to 800G and eventually 1.6T optical modules, and AAOI is trying to position itself as one of the suppliers riding that upgrade cycle.

The Real Bull Case - Demand for AAOI’s Pluggables Looks Insane

The AAOI bull thesis is relatively simple: if hyperscale AI buildouts keep demanding faster optical links, and AAOI can actually ship meaningful volumes of 800G and 1.6T modules, the company’s revenue base could look very different over the next few years. Let me explain.

The demand signal for AAOI’s pluggable transceivers is almost absurd. Management is now talking about roughly $471M of monthly data center transceiver revenue by mid-2027, and importantly, they clarified this is not just theoretical capacity. It is based on orders in hand, minimum customer commitments, and equipment that the company expects to have fully qualified. Even more interesting, management said actual capacity is higher than the $471M/month figure, and actual customer demand is bigger than what they are committing to. Customers apparently wanted that output by April, but AAOI is guiding to June/July because equipment installation, qualification, and materials all take time. In other words, the limiting factor does not appear to be whether customers want 800G and 1.6T modules. The limiting factor is how fast AAOI can physically build, qualify, and ship them.

This is why the setup feels so asymmetric if execution holds. The monthly revenue target increased from $378M to $471M versus the prior earnings call, which is nearly $100M of additional monthly revenue being pulled into the model (can you imagine that?!). That is not normal incremental growth; that is a demand curve moving faster than the company can expand.

Further, the even bigger picture is that hyperscalers are racing to upgrade AI clusters from 400G to 800G and eventually 1.6T, while sovereign AI data centers and Tier 2 cloud players are also trying to secure capacity. Nvidia also tends to absorb the first wave of supply, which can create bottlenecks for everyone else, so it would not be surprising if customers are trying to lock up optical capacity far earlier than usual.

On top of that, FundAI has just reported that the number of NPOs used in Rubin Ultra has nearly doubled.

In this context, while everyone seems to be assuming ASP declines at scale, if 800G/1.6T transceivers become a real bottleneck next year, the opposite could happen: pricing holds better than expected, margins expand, and AAOI ends up with more demand than it can satisfy through mid-2027.

The following sections discuss (1) the risks to the thesis; (2) whether the other limbs of AAOI’s business present further upside; and (3) how to think about valuation.

+++

Welcome to Orange Grove Capital

I turned $80K into $1M in 4 years looking for asymmetric, high-conviction bets before they became consensus. I created this Substack to publish my research and thought process in the hope that others can benefit from my investing framework.

If you’re interested in a detailed breakdown of that framework - including how I think about business quality, valuation, business sizing, risk management and what I look for in potential multi-baggers - you can read it here:

My full investing framework

+++

Read more

At first glance, today’s market reaction looked strange.

The U.S. economy just printed a much stronger-than-expected jobs number. Employers added 172,000 jobs in May, roughly double what economists expected, and unemployment stayed at 4.3%.

Normally, strong jobs data should be good news.

But in this market, good news for the economy can quickly become bad news for risk assets.

Here’s the issue - the rally over the past few months has been built on two big assumptions:

1. AI spending would keep pushing mega-cap tech higher
2. The Fed would still have room to ease policy if the labor market weakened

Today’s jobs report challenged the second assumption.

A strong labor market makes it harder for the Fed to justify rate cuts, especially while inflation remains above target. Instead of pricing in easier policy, investors are now being forced to consider a more hawkish path: higher-for-longer rates, fewer cuts, and even the possibility of future hikes.

That matters because long-duration assets - i.e, AI stocks, high-growth tech, crypto and speculative names - are highly sensitive to interest rates.

When yields rise, valuations come under pressure.

That is exactly what happened.

The S&P 500 fell sharply, wiping out roughly $1.8 trillion in market value in a single session. The Nasdaq was hit even harder, with AI and semiconductor stocks leading the decline.

This also came after an unusually strong run.

The S&P 500 had just reached record highs earlier in the week, powered by AI enthusiasm. After a 9-week winning streak and a major move higher in only two months, the market was already priced for near-perfection.

Then several pressure points hit at once:

1. First, labor data came in too strong for investors hoping for rate cuts.

2. Second, job openings also surprised to the upside, reinforcing the view that the labor market may not be weakening as much as previously thought.

3. Third, Bitcoin and crypto continued to slide, with BTC now down more than 50% from its October 2025 high. That has damaged risk appetite across the board.

4. Fourth, reports that Meta may raise tens of billions of dollars to fund AI infrastructure added a new concern: will Big Tech increasingly rely on large equity raises to fund the AI buildout?

5. And finally, with major private-market deals and IPOs competing for capital (think SpaceX), investors may be selling existing winners to make room for new supply.

So the selloff was not really about one jobs report.

It was about positioning, stretched valuations, higher yields, AI funding concerns, and weakening crypto sentiment all hitting at the same time.

The key takeaway:

Corrections like this are uncomfortable, but they are also normal. After a powerful run, markets need to reset expectations.

With the capital currently flooding into AI / memory / semi names, SaaS names have (largely) been left for dead. PLTR being down almost 40% from its peak and META down 15% in the past 6 months is testament to this.

Yet, we’re starting to see signs of capital rotations back into quality software names. Data Dog’s ($DDOG) blowout Q1 earnings sent the stock rocketing 30%. Tom Lee says that market breadth will potentially return in June, with software names being a beneficiary.

That may be the case, or maybe not, but as prudent investors, you at ought to be paying attention to this (potential) shift. At the same time, you should not also be buying every single SaaS name under the sun.

It is my thesis that once the software names rebound (not if, but when), the software names that will benefit (and indeed, do well going forward) are the ones that aggressively use AI to scale and become more operationally efficient.

LMND is one such name. I’ll break it down very clearly for you.

1. What is $LMND: $LMND is a relatively young insurance company (founded April 2015) which focusses on employing AI strategies and tools to build their business. With a market cap of $4 billion, it is significantly smaller than insurance incumbents such as Progressive ($PGR), which has a value of $120 billion. This should give you a sense of the upside.

Parallels can be drawn with $PLTR - $PLTR was a company building for 20 years before LLMs came along. As Karp admits, LLMs completely changed their trajectory as LLMs essentially supercharged the application of their ontology. Similarly, $LMND is not simply chasing the hype. $LMND has been building since 2015 with an AI-first mindset and is wonderfully poised to take advantage of the improvements in AI and LLMs - Shai Wininger himself has said that without the right organizational infrastructure (which $LMND has and has been building since inception), you’ll unlock less than 1% of AI’s potential. The time is nigh for $LMND to fully capitalize on its decade long advantage and grow exponentially from here with the advance of AI.

In the following sections I talk about why LMND is well positioned to win in the AI era, and the price I think it can reasonably hit based on my models.

+++

Welcome to Orange Grove Capital

I turned $80K into $1M in 4 years looking for asymmetric, high-conviction bets before they became consensus. I created this Substack to publish my research and thought process in the hope that others can benefit from my investing framework.

If you’re interested in a detailed breakdown of that framework - including how I think about business quality, valuation, business sizing, risk management and what I look for in potential multi-baggers - you can read it here:

My full investing framework

+++

Read more