The human tendency to extrapolate current events into the future is one of the most consistent biases in markets, both today, in the past and, more than likely, the future (there I go, extrapolating again). Consistent, both in the sense that it appears with notable frequency and in the sense that it is one of the most repeatable hunting grounds for market outperformance. (h/t Bireme Capital, )

Forecasting - Then & Now

The human brain is a pattern-matching machine. Long before we debated Plato and Socrates, our brains were predominantly focused on survival. That survival forced our brains to process a lot of information and make decisions near instantaneously. The way the brain does this is through pattern-matching. Simple recurrences such as the sun setting and rising gave way to simple heuristics – the sun sets and rises in uniform ways, therefore we have set hours of daylight followed by nightfall. This worked flawlessly 99.9% of the time, though sometimes this was not the case. In 536 AD, for example, a mysterious fog embroiled Europe, blanketing the skies for years. Suffice to say, humanity was not prepared, and 536 AD has been dubbed the “Single Worst Year in Humanity” – yes, even worse than [insert any year lately you think has been particularly offensive].

Share

In the ~1,500 years since, our brains have changed relatively little and similarly in markets, we pattern-match onto trends. Trends give us a sense of structure, comfort and logic. When trends change, much like in 536 AD, our reactions can become violent. We then build new patterns, trends and extrapolations to fill in the blanks of the new reality we find ourselves in. This is why 6th century European peasants became fanatically God-fearing and megacap companies can see their market cap decline by over 20% when a single quarter misses after a long string of beats and raises. We tend to assume first and ask questions later. Recent examples that come to mind are Meta in 2022 (-76% peak-to-trough downturn!) and Google in 2025 (a tamer -23% drawdown).

The Value of Future Cashflows

I’m not immune to this either. One exercise I’ve found helpful is to look at the drawdown and ask myself objectively: does the drawdown truly reflect the change in the value of the company’s future cashflows? Put differently, when Meta declined by 76%, was it realistic that the future value of all its cashflows had declined by 76% as well? With the benefit of hindsight, we now know that it had not. This isn’t to say that large drawdowns occur without reason: Blackberry peaked in 2007, as the iPhone was released. It proceeded to decline by -95%. In that case, the future value of its cashflows indeed collapsed by a similar amount as its industry went through a complete transformation. Many companies have underlying issues that build up, and for whatever reason, the market realizes it in one moment and discounts the cashflows appropriately, despite the issues not materializing overnight but over a longer period of time.

This exercise in attempting to mitigate our need to extrapolate helps for several reasons: it forces you to stop catastrophizing and start thinking longer term. It also forces you to think relatively. More competitive pressures or product recalls may decrease the value of future cash flows by a few % points; the real nuance and value of proper analysis is understanding whether the drop in stock price is within a reasonable range of that change. The delta between the reasonable and unreasonable is where alpha lies.

Yes, DCFs are rife with problems, and I personally don’t rely on them. But in one way or another, every security’s stock price is a bet on the value of its future cashflows. All multiples are simply shorthand DCFs, whether we want to admit it or not. Also, I haven’t brought up the flip side – when companies rise many hundreds of % over a short period of time – is that an accurate reflection of the change in value of their future cashflows? In my experience, this is a harder question to answer and, honestly, a much nicer problem to have (when you’re long).

Netflix & Panic - The Extrapolation Example

One of the best examples of that extrapolation bias at work was Netflix. PhiloInvestor has a much better overview of the situation that has stuck with me ever since I originally read it years ago. I won’t do it enough justice, so I highly recommend giving it a read.

The summary of it is that going into 2022, Netflix posted its first decline in subscribers and the markets were unhappy: The Company subsequently declined 70% peak-to-trough.

There is more to it, but that’s the headline the average investor looked into when they slammed the sell button into oblivion. Now, let’s look at the situation through the lens of future cash flows. How many subscribers did Netflix lose? 200,000. Sounds scary, but Netflix at that point had 226.6 MILLION subscribers. Their subscriber base declined by 0.09%. Not 9%, 0.09%. Should that realistically translate into 70% less future cashflows? Probably not, without further validation. The market clearly didn’t care and slapped a 19x PE multiple at the trough, while at the same time sporting 40% gross margins and 20% operating margins at 10% sales growth despite losing subscribers.

The part that really struck a chord with me in PhiloInvestor’s analysis, however, was the element of mean reversion. Netflix was a Corona darling as quarantine supercharged their subscriber growth rates. Pre-pandemic, between 2017–2019, Netflix was enjoying 30% annual growth rates in subscribers. Then, in 2020, Netflix almost doubled its prior average growth rate. Investors salivated at the beginning of a new trend – the flywheel was spinning! That’s what makes Netflix the quintessential extrapolation example, as investors dealt with two incorrect patterns. First, that the pandemic growth was a ‘new normal,’ and then, when it mean-reverted to its old trendline, investors projected a new trend of stagnation or even terminal decline, rather than mean reversion. In the end, Netflix kept growing and found new levers to drive returns.

A Speculator Takes The Opposite Trade

Why else this example was particularly interesting to me was that at the time, while working at a long-short hedge fund, I was also engaged in profiting from post-pandemic extrapolations. However, in my case, I was shorting companies whose investors still believed that pandemic-era growth was the new normal. The industry in question was the large digital publishers such as Future plc, Dotdash Meredith (IAC), and Ziff-Davis. Their business models revolved around driving large amounts of traffic to web domains they owned, where they would monetize those eyeballs through ads and, increasingly, affiliate links. The predominant driver of these earnings was their tech sites, because the affiliate revenues were so profitable – the sticker price of a smartphone and PC made the commission revenues for tech websites much more attractive than lifestyle and beauty.

This was a great business to be in during the pandemic, as people were locked at home, were given free money, and were itching to upgrade their home tech. Meanwhile, management bragged about their sophisticated advertising engines, their elite SEO, and the health of the consumer who couldn’t and wouldn’t stop buying. Of course, in the end, these all came crashing down as the trend mean-reverted post-pandemic. I could get into how these companies then tried to hide their declining growth through acquisitions, but that’s another story for another time.

The true difference between them and Netflix was that, in a steady-state, after mean reverting to trend, Netflix was a high-quality business, and digital publishers were not. Digital publishers are extremely pro-cyclical and at the mercy of consumer spending habits. If there is fear of slowdown, advertisers are the first to rein in their spend, leading to a double punch of revenue declines in both ad and affiliate income. As their website traffic collapsed, so did their revenue streams, and the companies looked to acquire their way into maintaining growth, which also made them beholden to interest rates, and indeed, their own share prices, to be able to keep financing growth.

The pandemic turned a crappy business into a good one. Compare this to Netflix, where churn is low, brand is strong, and the pandemic bolstered their subscriber base with minimal post-pandemic churn at incremental margins on a new subscriber of ~85% - ’nuff said. The pandemic boosted a good business, and it was able to keep the majority of those gains

Its Human Nature - Forecasting What’s Next

Where can we see new examples of the extrapolation bias at work? One potential that comes to mind is AppLovin. After 7 quarters of acceleration, APP “only” grew its sales 53% year-on-year. A big miss versus consensus and only a 6% sequential growth rate have led to a near 30% slump post-earnings. Is AppLovin’s business 30% less valuable? Has the value of their future cashflows declined by 30%? With the benefit of hindsight, I’ll be able to tell you exactly; until then, we wait and see.

Subscribe now

There was a much hyped pre-order date in May, buoyed by rumors of retailers already receiving copies of the game that came and went, the market moved on.

Subscribe now

What I found was a divergence in the value of the asset and the credit the market was giving TTWO. On the one hand you had the successive disappointments in release delays, then the lingering effects of the AI Trade seeping into the gaming industry and sucking the oxygen out of the sector (See Sony, Nintendo, Konami etc). On the other hand, you have an asset whose last iteration did a mind-boggling number in terms of sales and spawned a completely new online game revenue stream that was by the developers’ own admittance, not expected.

We’re going to run it back again and now Rockstar will have learned exactly how to monetize their long-tail online product too.

Yet despite all of this, the stock is more or less flat on the year. Taking out the technical reasons (AI overcrowding, hatred of anything software), what is the market missing? Certainty. Namely the below:

1. The price point GTA 6 will sell at - Consensus was closer to $70, while it was clear to me that it would price at $80. Today Rockstar officially confirmed this, with special additions at $100+. This means blended ASPs will be $80+

2. Pre-order success - Many extrapolations will be made on the back of how successful GTA6 pre-orders do. But up until this point even the pre-order date wasn’t confirmed. We will finally started getting proper data now.

3. Will it release on time - The market still has some residual doubt about Rockstar sticking to their timeline. I also don’t buy this as TTWO’s CEO has started re-iterating the release date on investor calls and conferences. Confirming pre-order dates and prices further strengthens the idea that “its really happening”

I’m posting my original internal analysis below. I haven’t updated it at all (besides any sensitive points related to our portfolio), mainly out of laziness.

The main point here is that the cone of uncertainty has narrowed - we now know base prices, analysts can now model the range of potential sales performance in a much easier way. For those worried that $80 will lead to demand destruction, putting aside the fact that GTA Online is STILL active, keep in mind that $60 in 2013 (GTA 5’s price and release year) equates about $85 today with the impact of inflation. Adjusted for inflation. this isn’t much of a hike at all.

Thesis Overview

• TTWO has underperformed through the year along with the rest of gaming due to AI ‘fears’, despite having IP that AI can’t replicate. GTA may be the most valuable entertainment franchise in the world (GTA V holds multiple Guinness World Records for 1 day, 1 week sales records)

• GTA 6 is scheduled for release on November 19, 2026, after 18 months of delays and is the big catalyst for the stock, my research revolved around this one product, despite TTWO owning many franchises

  • The release is meant to fall on FY Q3 2027 – my work revolves around estimates of that quarter’s sales. How successful the launch is should set the cadence for TTWO’s FY27 results

• The market still isn’t 100% convinced there won’t be further delays, as pre-orders haven’t opened yet

Consensus Views

• My base assumption is that GTA 6 is released as planned. My numbers for GTA sales in Q3 2027 are 7% above consensus - Appendix 1

  • There is a large delta between consensus GTA 6 Q3 sales and total TTWO Bookings as only two Sell Side houses have numbers for GTA 6 Net Sales specifically whereas most Sell Side houses haven’t updated Total Bookings to reflect GTA 6

• Due to the perceived uncertainty around this massive catalyst, Sell Side hasn’t been updating their numbers. Q3 27 sales estimates haven’t moved all year – Appendix 2

  

• There is a timeline of catalysts pre-release that should lead to positive price action as the cone of uncertainty narrows

  • Confirmation of price point for the game

  • Pre-order data to confirm demand levels

  • Proper confirmation from distributers that the game is scheduled to be released on November 19

  • TTWO’s marketing campaign on the run up to release

• TTWO’s depressed multiples combined with growing certainty around its release and likely consensus upgrades as visibility clears should lead to solid price appreciation – Appendix 3

GTA 6 Sales Forecast – FYQ3 27

• My forecast is based on Installed Console Base x Expected Penetration Rate of Consoles x ASP – Appendix 4

  

• Installed Console Base (PS5 + Xbox): 143.6m units

  • Sony reports cumulative hardware sales, assuming 2026 will grow 0% yoy

    • PS5 units have declined yoy since peaking in 2023, I believe GTA’s release will likely spur a wave of console buying, especially as it is around Christmas, meaning at least flat should be realistic/conservative

  • Microsoft no longer reports Xbox unit sales but VGChartz has a monthly tracker for console hardware sales – MSFT numbers based on that – Source

    • Sense check: Microsoft has consistently sold less Xboxs than Sony’s Playstation and admitted that they had lost the console war

    • Also assume GTA’s release will lead to a wave of console buying, and that yoy sales will be flat for 2026

  • TTWO has not confirmed if there will be a simultaneous release on PC. Historically this has not been the case but I assume the PC market adds an additional 40.5m buyers that aren’t included in the calculation

    • Based on TTWO’s historic revenue split between Consoles and PC (22% to PCs)

    • Together would equate to a buyer base of 184.1m

• Expected 2 Week Penetration Rate: 17%

  • Based on historical data – GTA 5’s 2 Week Penetration Rate was 18% - Appendix 5

  

  • Once pre-orders for the game open, there should be more clarity on what this number could be

• Implied FY Q3 27 Units sold: 46.5m by December 31, 2026

  • Two Week Units Sold (Nov 19 -> Dec 03): 24.4m

  • Implied Rest of December Units Sold: 22.1m

    • As holidays ramp, expect pace of sales to stay strong and for a similar number of sales to occur over the remaining 28 days in December

      • (24.4/31)*28

  • Morgan Stanley, BMO, BofA at 45m

• ASP: $80

  • TTWO has not announced pricing but consensus is still assuming $60-$70, this is where the majority of my beat comes from

  • Consensus has not updated pricing despite TTWO CEO stating that they will price GTA 6 ‘according to the value it will give players’

  • Microsoft announcing that they will start pricing Xbox games at $80 - Source

    • There is no reason for Sony not to match pricing

  • An Italian store accidentally uploading a pre-order page on 18/05/26 at EUR 70 (Approx. $80) - Source

  • My ASP is conservative as it doesn’t take into account bundles/special editions which will sell for >$80 and raise blended ASP

• Gross CY 2026 GTA 6 sales: $3,716m

  • 46.5 units x $80 ASP

• Net GTA VI Sales: $2,676m

  • After netting off platform fees, assumed fee of 28% (From Consensus)

  • Console platform fee is 30%, but blockbuster releases usually negotiate discounted rates

  • I am 7% over consensus – Appendix 1

    • To match consensus we would need to see at least $80 ASP and 16% 2 Week Penetration or $70 ASP and 19% 2 Week Penetration – Appendix 6

• My GTA 6 forecast gets me to $4,075m in Total Bookings for Q3 27 (GTA 6’s release) vs Consensus $3,396m – Appendix 1

  • As stated in the Thesis Overview, this consensus # is outdated and ripe for upgrades

  • Compared to BMO, which are bullish and recently upgraded (18/5/26): $4,100m

  • Ex-GTA, my bookings numbers are based on consensus averages

• Important to note, I haven’t factored GTA Online at all. We don’t know how TTWO will bundle online offerings and majority of revenues from GTA Online should occur in the future, post-launch

  • But will likely add a material mid-term tailwind

After a long hiatus, finally back to it.

Write Ups

1. - AI Won’t Nuke SaaS: Why the White-Collar Apocalypse Is Just Tech Bros’ Dreaming Nonsense

2. - Charts of the Week: SaaSpocalypse Interrupted

3. - The hidden monopoly recycling Europe’s toxic waste (I like European, I like industrials)

4. - Reverse Engineering Buffett: The 7× EBIT Yardstick

Articles

1. Reuters - Apple’s China smartphone sales jump 23% to start 2026, bucking industry trend

2. X - CTO of Hubspot on AI

3. ETF.com - Roundhill Launches Memory ETF to Capture AI’s Hottest Trade (Lets put it this way, thematic ETFs don’t launch at the beginning of a cycle…)

4. Bloomberg - Microsoft hit ‘audacious’ Copilot goals after analyst input (I think if successful, this marks a business model shift away from ‘freemium’ AI offerings)

Podcasts

1. - Truth Seeking and Crossover Investing at Atreides (Haven’t taken this many notes from a podcast in ages)

   Thanks for reading The Tortoisehare! Subscribe for free to receive new posts and support my work.

The SaaS-apocalypse

Enterprise software has had a rough few months. Since October 2025, the iShares Software ETF (IGV) has fallen roughly 34%, wiping out most of the gains from the prior two years. That includes non-SaaS names like Oracle and Palantir. Digging deeper yields an even uglier picture.

The selloff isn’t random. It reflects a repricing of risk as investors question whether the SaaS model that defined the past decade remains structurally intact.

The pressure isn’t limited to equities. Private credit, now the largest lender to SaaS, is starting to feel it. Outstanding loans to SaaS firms grew from $8 billion in 2015 to over $500 billion by end-2025, representing 19% of all direct loans (BIS). Roughly 25% of BDC portfolios are in software (Morgan Stanley). When redemption requests spiked at vehicles like BCRED and Ares Strategic Income Fund this quarter, software exposure was central. That dampens new lending and tightens conditions across the sector.

Saying “the market is scared” is imprecise. The market is repricing terminal growth. The question is: why?

What the Market Is Pricing In

Three distinct fears are driving the selloff. Separating them matters because they have different implications for individual companies and help identify what to own—and what to avoid.

Fear #1: AI-native competitors
Vertical AI startups can build purpose-specific tools that compete with modules of enterprise suites at a fraction of the cost, without legacy code. For SMBs, switching costs collapse when alternatives are ~60% cheaper and deploy in weeks.

Fear #2: In-house development (“vibe coding”)
As internal tooling improves, enterprises may replace off-the-shelf software with bespoke solutions built by smaller teams. That implies a shrinking addressable market.

Fear #3: Death of the seat model
SaaS revenue architecture—ARR, NRR, expansion—is built on seats. If AI reduces required headcount, fewer seats means less revenue, even if the underlying business improves. The long-standing tailwind of seat expansion and legacy-to-SaaS conversion is fading.

None of these fears are irrational. The market is applying them indiscriminately. That’s where the opportunity lies.

A Simple Mental Model

Before company-level work, you need a filter.

The Rule of 40 (Revenue Growth % + FCF Margin %) has become the dominant valuation anchor. Empirically, the relationship is clear: a large study by McKinsey & Company shows top-quartile SaaS companies (by Rule of 40) trade at ~3× the EV/revenue multiples of the bottom quartile.

On FCF: Studies have shown that adjusting for SBC has no empirical benefit. Data shows flat or negative alpha. With 80%+ gross margins and stable, growing revenue, SBC is often a marginal cost over time. There are outliers, but I’ve used reported FCF.

Plotting 20 large-cap enterprise SaaS names supports this. Higher Rule of 40 correlates with higher multiples. Companies below 40% cluster in the bottom quartile of EV/NTM revenue. My universe trades at median ~6x EV/NTM revenue, with median Rule of 40 at ~45%. Company maturity determines the mix: mature names trade growth for margin.

Plotting Rule of 40 vs EV/NTM revenue gives a starting filter. The 40% threshold is the horizontal line; median EV/NTM revenue is the vertical. I focus on companies above 40% but below median multiples - candidates for rerating when sentiment stabilizes.

I also look at companies below 40% with a credible path to cross it. That creates a built-in valuation tailwind.

This leaves ~10 names. That’s the benefit of quantitative screening: quickly eliminating what not to spend time on.

Why a Basket Isn’t Enough

Buying cheap SaaS as a basket is tempting but risky. Dispersion is high for a reason. Some companies will navigate the AI transition; others face terminal reratings as growth flattens and retooling windows close. The legacy tailwind, migrating customers to SaaS is ending.

The difference between cheap and a value trap is quality. To assess that, I mapped market fears to company characteristics and reduced them to a set of filters. Though maybe calling them survivability factors may be more relevant.

The Filters

1. SMB Exposure: Prefer Below 25%

SMBs are the most likely to defect to AI-native alternatives. They have shorter procurement cycles, lower switching costs, less organizational inertia, and critically —the strongest incentive to cut costs wherever possible. A SaaS company with heavy SMB exposure is structurally more vulnerable to the competitive dynamics described above. Screen for companies where enterprise (typically defined as 1,000+ employees) accounts for the majority of revenue.

2. Revenue Acceleration (or a Credible Path to It)

The biggest piece of the SaaS growth story for the past decade was the ability to upsell an existing customer base: land and expand, consistent pricing increases above churn. That lever has matured. What you want now is evidence that growth is reaccelerating, ideally because of AI-driven product expansion rather than in spite of the AI environment. Companies showing a positive inflection in sales are signaling that they are bucking the current zeitgeist of SaaS is in terminal decline. Bonus points if the acceleration is directly attributable to new AI product lines.

3. Demonstrable AI Revenue Contribution

The market fears that AI will take share from SaaS incumbents. It’s largely ignoring the fact that SaaS incumbents can build their own AI products and have significant benefits in terms of distribution on the back of their large installed base. Theoretically, this means large enterprise SaaS can acquire new AI-product customers at low marginal cost compared to native AI start ups.

The companies that have moved fastest to monetize AI and not just tack on a “copilot” but generate real, attributable ARR from AI products, are likely the most important category within this framework. These companies can articulate that AI is a tailwind for them, and that they aren’t growing “in spite” of AI. Sort by AI revenue as a percentage of total, and prioritize companies where that contribution is already material and growing fast enough to become a tailwind for overall revenue acceleration.

4. System of Record Over System of Action

This is the filter I’ve seen written up most. A system of action, as I defined it, is a tool that executes a task. This is relatively easy to replicate. A point solution that automates a workflow can be rebuilt by an AI-native competitor in months. A system of record is different. These are the databases of proprietary, context-rich enterprise data built up over years, sometimes decades. Industry-specific workflows work better on these platforms because they’re grounded in real operational data. An AI-native competitor trying to replicate a 20-year CRM database isn’t just rewriting code, they’re asking a customer to abandon their infrastructure, essentially, to start over without it. The moat is the data, and it compounds over time.

5. Consumption-Based Pricing

One of the central fears in the current selloff is the death of the seat model. If AI agents replace workers, demand for per-seat licenses declines. The companies that have either shifted to consumption-based pricing or are actively building it into their AI product lines are the ones best positioned to actually benefit from the scenario the market fears. An agent that handles 10,000 customer service interactions doesn’t need a “seat” in the traditional sense but it does generate usage-based revenue with no incremental headcount cost. Look for companies where consumption pricing is already generating material revenue, not just announced.

6. Cross-Organizational Footprint

SaaS applications are stickiest when they touch multiple teams across an enterprise. A tool used only by the engineering team is easy to swap - engineers are exactly the people who have the knowledge and willingness to experiment with alternatives. A platform that runs through sales, finance, customer service, HR, and IT is a different animal. Migrating it risks genuine business disruption, which is something the C-suite will prioritize over any potential productivity gain from a new tool. The more teams that touch the software, the higher the switching cost, full stop.

7. Insider Buying and Net Buyback (Not Buyback Theater)

Two capital allocation signals that matter more than usual in this environment. Every SaaS CEO is saying publicly that they can’t be disrupted. Very few are backing that up with their own capital. Insider buying in a down market is a meaningful signal.

On buybacks: for years, software companies have used “buyback programs” that in practice did little more than offset SBC dilution. Shares outstanding barely moved. What you want now is companies that are actually net-reducing share count. With prices down 20–30% from peaks, accretive buybacks are finally possible the companies executing them or accelerating them are genuinely are being good stewards of capital, especially as most of these companies spit out gobs of free cashflow.

Walking Through the Filters: Salesforce (CRM)

Let’s run Salesforce through the framework.

SMB exposure: ~15%. Salesforce is overwhelmingly enterprise-focused. The SMB defection risk is relatively contained.

Revenue acceleration: Three-year average revenue growth sits at around 9.8%, with NTM consensus estimates pointing to roughly 11% for FY27. Not spectacular, but the direction is right, and the acceleration has a credible driver in Agentforce.

AI revenue: This is where the story gets interesting. Salesforce’s Agentforce and Data Cloud combined reached $2.9 billion in ARR in FY26, up 114% year-over-year. Agentforce alone surpassed $800 million in ARR in Q3 FY26, up 330% YoY. The company has closed over 9,500 paid Agentforce deals. At the current trajectory, AI is on course to become a genuine tailwind for overall revenue acceleration.

System of record: CRM is the canonical system of record. It sits at the intersection of every customer relationship a company has - sales pipeline, customer history, service tickets, marketing interactions, renewal data. That’s not a workflow tool you replace with an AI-native product. It’s an enterprise’s institutional memory, and it compounds with every interaction.

Cross-organizational footprint: Salesforce doesn’t just touch sales teams. It runs through customer service, marketing, finance, and IT. Over 50% of Q1 FY26 top 100 deals included six or more Salesforce clouds. The organizational surface area makes displacement extremely costly.

Consumption pricing: Salesforce is directly addressing the seat-model risk. Agentforce pricing is tied to “Agentic Work Units” charged based on the complexity and number of actions taken by an AI agent. The company is also offering Agentic Enterprise License Agreements (AELAs) as a replacement for traditional seat-based contracts. This is the right structural response to the fear the market is pricing in.

Insider buying: None to report.

Buybacks vs. SBC: Salesforce authorized a $50 billion buyback representing roughly 25% of market cap, against expected SBC of around $4.4 billion annually. In FY26, buybacks came to $12.7 billion against $3.5 billion in SBC - a genuine net reduction in shares outstanding. This is not buyback theater. The concern the market is raising is financing: Salesforce recently completed an $8 billion acquisition of Informatica, and investors are watching how it funds the buyback alongside integration costs. Worth monitoring, but the FY26 execution itself is a pass.

Overall: Salesforce passes six of seven filters cleanly, with insider buying being the one miss. The AI contribution metrics in particular are ahead of what most investors credited the company with twelve months ago.

How This Plays Out

When I started writing this a week or two again, it was clear that the path of least resistance in markets right now is long semiconductors, short software. AI capex keeps flowing - Microsoft, Google, and Amazon combined are spending well over $200 billion on infrastructure this year and semis are the direct beneficiary. Software sits on the other side of that trade until it can demonstrate that it’s a beneficiary of AI rather than a casualty. As the Iran war has gone into potentially longer with much longer knock-on effects, the tech sector has gone into risk-off mode, which makes what I write next even more prescient:

That changes when one of two things happens. On the semi side: the memory cycle turns, energy constraints bite harder, or demand weakens as data center buildout slows. On the software side: enough companies start reporting AI revenues that are material, growing fast, and not margin-destructive.

The margin question is the open variable. The consensus is that AI product margins will be lower than traditional SaaS margins, at least initially, inference costs are real, and the economics aren’t fully visible yet. The counter-argument is that inference costs have been dropping at roughly 10x per year, which means a company that takes a margin dip today could exit the transition with a meaningfully better unit economics profile as costs normalize. There’s also the fact that no SaaS company is revealing exactly how their AI products are priced internally. The cone of outcomes is wide and when that’s true, markets tend to punish the lower-multiple names until the picture clears up.

One external signal worth watching: Anthropic has been selectively partnering with SaaS incumbents rather than purely competing with them. That, in itself, is a signal about where the foundational model providers see the AI enterprise stack heading. It’s not winner-take-all at the application layer — and the companies with the data moats, the cross-organizational footprints, and the credible AI revenue stories are the ones most likely to be on the right side of that.

Thanks For Reading!

If you’ve found my work valuable, here are a few ways to support it, any help is greatly appreciated!

• Subscribe to my Substack - just sign up below

• Share feedback - let me know what I missed or could improve

• Spread the word - pass this write-up along to friends or colleagues

• Send me ideas - if I like them, I might write them up (win/win)

  Subscribe now

This content is for informational and educational purposes only. It is not investment advice or a recommendation to buy or sell any security. The companies or ideas discussed do not represent my employer’s views. Do your own research and consider your own financial situation before making any investment decisions.

Write Ups

1. - Cairo Mezz plc (A Greek special situation & “Weird Shit” Idea)

2. - How to increase your surface area for luck (You should just do things)

3. - Global Spirits: Q2 Earnings Update (Great analysis of an industry that has gotten on my radar)

Articles

1. Packer & Co - December 2025 Fund Letter (There are few funds with 25+ yr track records that still beat the market. Those are worth paying attention to)

2. Hosking Partners - PGM update (I had read this piece after building a position in PGM metals, which tied it all together nicely)

3. Horizon Kinetics - Q4 2024 Fund Letter (Why such an old one? On successful investing in world-changing technologies: Seek out a business that in some way is necessary to—or enables—the technology developers to do whatever they will do, which is winner-indifferent, and which is a long-term fee collector.)

Write Ups

• - Gas Fired Intelligence

• - Are commodities a hedge fund’s best friend

• - My Updated Step-by-Step Guide to Outperformance

• - Adobe ($ADBE) - Show me the numbers!

Articles

• - The outlook for IPOs in 2026: oversized and over there

• Goldman Sachs - 2026 Outlook: Tech Tonic — a broadening bull market

• Blue Tower AM - Q4 2025 Fund Letter

• Korea Ministry of Economy - Tax Support Measures to Revitalize the Domestic Capital Market and Mitigate Structural Supply-Demand Imbalances in the FX Market

Podcasts

• Monetary Matters - ’s 26 Trades for 2026

• Invest Like The Best - Ricky Sandler - Building an Investment Compass

I generally don’t go for ‘pure play’ tech stocks but here we have a company with a fortress balance sheet, gushing free-cash flow and is set to benefit from AI adoption. Despite this being a $130bn market cap company with a brand everyone is familiar with - no one talks about them.

When most investors think about AI winners, they picture foundation model companies or flashy startups. They’re missing Adobe, a company quietly building one of the most defensible positions in enterprise AI while trading at decade-low valuations.

Adobe’s next decade of value creation will likely mirror its last: converting adoption into monetization through bundled value and workflow lock-in. The Company is uniquely positioned to capture enterprise AI spend by solving the critical IP and legal risks that prevent broader adoption, while leveraging its unmatched distribution network to monetize AI at minimal customer acquisition cost.

Finally, I think Adobe is positioning itself to be a heavyweight in the nascent GEO (Generative AI Engine Optimization) space, an area I think is going to become increasingly important to DTC, ecommerce and digital media businesses.

This is a Company which has an incredibly stable recurring revenue stream at high, stable margins. Assuming no multiple expansion, revenue growth and buybacks alone should get you to a 17% IRR. Considering, Adobe is trading at a decade low, multiple expansion isn’t out of the question.

Special shoutout to , whose Netflix post which highlighted distribution has stuck with me ever since reading it (a rarity!) and who recently spoke about the enterprise AI ‘losers’ who may be gearing for a big reversal.

Business Overview

Adobe’s value creation over the past decade came from three core strategies: converting perpetual licenses to subscriptions, expanding average revenue per user through bundled value, and driving exceptionally low churn through workflow lock-in. The company now operates across three primary segments.

Creative Cloud remains the major business driver. Adobe commands an estimated 32 million Photoshop users and maintains a dominant position in creative software. Adobe itself estimates it controls 20% of the Creative software TAM, concentration you don’t see every day. The Company recently moved its all-in AI video editor to public beta, with early reviews putting it ahead of OpenAI and Google, who offer image and audio generation but not comprehensive AI-native video editing packages.

Document Cloud serves 650 million monthly active users for Acrobat. With 3 trillion PDFs in existence globally and over 400 billion documents opened in Acrobat annually, this represents one of the few products comparable to Microsoft Office in terms consumer penetration.

Experience Cloud serves 12,000 enterprise customers, including 87% of the Fortune 100 and 74% of the Fortune 500. This segment is probably the least-known to the average person on the street but here too, Adobe shows dominance: Gartner’s Magic Quadrant lists Adobe’s Experience Platform as a Leader in digital experience platforms. Experience Cloud focuses on marketing campaign management and data analytics. The segment also provides optionality through GEO exposure via LLM Optimizer and the recent SEMRush acquisition.

Adobe’s total addressable market stood at $205 billion in 2024 and is projected to reach $293 billion by 2027, representing a 14% CAGR. While growing slower than the overall market, Adobe already commands significant market share: approximately 8% of the Document TAM, 4% of Content Management TAM, and 20% of Creative TAM (the primary business driver).

Investment Thesis

Enterprise AI Adoption’s Pain Point: Security and Legal Risk

A 2025 study revealed that most enterprises admit they are not prepared for AI security risks, citing it as a major concern. OECD surveys show significant gaps between AI adoption intent and actual implementation due to security and privavy concerns. High-profile legal actions are shaping the competitive landscape. Midjourney, which has taken a laissez faire approach to image generation and IP has been sued by Anderson, Disney, Universal, and Warner Bros for copyright infringement.

This creates a fundamental barrier to enterprise AI adoption. Companies want the productivity gains but cannot stomach the legal exposure.

Adobe’s Triple-Layered Protection

Adobe addresses this pain point through three defensive layers that competitors cannot easily replicate.

Layer 1: Blanket Indemnification

Adobe, alongside Microsoft and Google, offers blanket indemnification for any output enterprise users produce. This protection creates a barrier to entry that only large-scale, highly profitable companies can afford to provide (such as Microsoft and Google). Smaller AI startups do not have the financials to offer similar incentives.

Layer 2: Licensed Training Data

Adobe trains exclusively on licensed data from Adobe Stock, strengthening its position with privacy-conscious customers. Adobe Stock contains hundreds of millions of images, videos, vectors, and 3D assets accumulated over decades. Contributors are paid, providing clear contractual rights for derivative model training.

The cost to replicate this moat is likely too large for any company to try: building a contributor marketplace, establishing reputation, paying thousands or millions of users per download, and providing indemnification. This moat took Adobe decades to build.

Layer 3: Firefly Foundry (Custom Enterprise Models)

Firefly Foundry trains custom models exclusively on enterprise customer data, addressing niche use cases where off-the-shelf large language models fall short. This guarantees privacy and data protection while eliminating in-house model building operating expenses.

More importantly, it creates massive switching costs. As one past user explained: “The Firefly Foundry strategy is clever because it sidesteps the commoditization risk that’s hitting most gen-AI tools. When enterprises train on their own brand assets and guidelines, they’re basically locking themselves deeper into the Adobe ecosystem. I saw this up close when a marketing team I worked with spent six months building custom models and now switching costs are enormous.”

The strategy also creates significant upsell opportunities. As per management: one media client expanded from $10 million to approximately $17 million in annual recurring revenue by combining traditional creative products with Firefly Services and Foundry.

Competitive Dynamics

While AI image generators may serve as ideation tools (Midjourney and Canva for concept development), enterprises need legally defensible, commercially viable final assets. Most professional designers may use other tools to quickly scale up ideas, the asset itself will still be done in Photoshop. Adobe is positioning Firefly as the safe choice for production-ready content, while competitors remain confined to the inspiration phase.

According to Adobe’s blog in October 2025, 86% of creatives use generative AI in workflows, yet 45% say they will never trust generative AI to produce final assets. Human post-production and editing remain critical.

Adobe’s brand trust reinforces this positioning. The company ranks #2 on Newsweek’s 2025 America’s Most Trusted Businesses and #20 on Interbrand’s Best Global Brands with a $41 billion brand valuation, ahead of Hermès, IBM, and Netflix.

Adobe’s Distribution Network: The Ultimate Competitive Advantage

If product is king (OpenAI), then distribution is god (h/t ). Adobe possesses one of the deepest distribution networks in enterprise software.

Consumer and Professional Reach

Acrobat serves 650 million monthly active users across free and paid tiers. It’s the only product comparable to Microsoft Office (approximately 1 billion users) in consumer penetration. Free AI features shipped into Acrobat serve as an on-ramp for premium AI features, creating another monetization vector for free users.

Creative Cloud serves 37 million users. Adobe Experience serves 12,000 enterprise customers, representing 87% of the Fortune 100 and 74% of the Fortune 500. These are deep distribution networks running across multiple markets and client-types.

Microsoft Copilot provides a case study: it became one of the most used chatbots primarily due to Microsoft’s distribution network, despite modest conversion rates.

Zero Marginal Customer Acquisition Cost for AI Products

Adobe can ship AI products pre-bundled or deploy remote updates to its entire installed base at zero marginal customer acquisition cost. There is no need to build sales channels from scratch as the sales funnels and relationships are all pre-existing. In the Creative segment, most users are already paid subscribers, which are significantly more receptive to AI feature upsells.

The proof point: Adobe’s AI generative fill feature in Photoshop achieved 10x faster adoption than average Creative features, according to the company’s 2023 Investor Day Deck.

Financially Underwriting AI Investment

Inference training requires upfront costs. Monetization only comes later once a critical mass of subscriber users has been acquired. Upstarts need to fund inference training through venture capital or debt markets. Adobe’s $25 billion subscription annual recurring revenue base provides highly predictable cash flow to underwrite inference costs with financial discipline without risking dilution or turning their balance sheet upside down (Cough, cough, Oracle).

The result: gross margins haven’t changed post-Firefly roll out. This margin management came about from zero incremental sales cost and training costs spread across millions of paying users.

The Walled Garden Thesis

AI models will eventually converge in quality and commoditize. Value will accrue to walled gardens integrated into pre-existing ecosystems and workflows. Think Microsoft Office versus alternative productivity tools, or Salesforce versus standalone CRMs.

Native AI within enterprise ecosystems will retain users through seamlessness and workflow integration. General AIs will face pricing compression as they get commoditized. Workflow-embedded AIs should retain pricing power through upsell opportunities in pre-existing bundles. Privacy and regulatory benefits provide additional defensible pricing power as true value-add in Adobe’s case.

As one Photoshop Essentials review noted: “If you have an Adobe Creative Cloud subscription and need AI images for your project, the obvious choice is Adobe Firefly. A big advantage is that it’s included with Creative Cloud at no additional cost. Your other option is a third-party AI image generator, with Midjourney currently the most popular, but it requires its own monthly subscription.”

Long Term Market Conditions Favor Established Players

While equity and debt markets currently fund upstarts with distant revenue expectations, sentiment shifts can reverse this dynamic. Companies like Adobe with established distribution and stable subscription bases will be able to continue developing AI products while upstarts face funding pullbacks.

Third-Party Model Integration

Rather than competing with OpenAI and Midjourney, Adobe integrates these models directly into its products. This transforms potential competitors into assets that further lock users into Adobe’s ecosystem.

Adobe also recently announced a ChatGPT partnership allows ChatGPT users to access Photoshop, Express, and Acrobat within the ChatGPT interface. This is another new top of funnel, making sure photoshop users keep using photoshop even if they want to also use OpenAI products.

Image Generation Volume Data

As of September 2025, 15 billion images had been created from AI generators globally. Seven billion came from Firefly (Source). Even excluding third-party models integrated in Firefly, Adobe’s proprietary model generated 1 billion images versus 964 million for Midjourney and 900 million for DALL-E (Source).

That’s the power of distribution, before researching Adobe - I hadn’t heard of Firefly yet it is one of the more proliferent models out there.

AI-Influenced Revenue Tailwinds

Adobe is capturing AI revenue growth across multiple vectors simultaneously.

A. Upsell to Existing Enterprise Base

Adobe has stated that nearly 50% of Enterprise Term License Agreements upgraded to AI-enabled Acrobat Studio. Assuming enterprise Acrobat Pro pricing of $15 per seat per month versus Acrobat Studio’s $29.99 list price, represents a 100% price increase. Applied to 50% contract penetration, this generates a 50% revenue increase for the Acrobat product line. Adobe hasn’t broken down exactly which products this applies to, but it shows the pricing increase opportunity.

B. New Customer Acquisition

First-time Adobe subscribers originating from Firefly grew 30% in Q2 2025 and 20% in Q3 2025. While conversion rate data has not been disclosed, the trajectory indicates meaningful funnel development.

C. Net-New AI-First Revenue Streams

Management began reporting AI-first ARR (covering Firefly, Acrobat AI Assistant, and GenStudio) in Q1 2025 at $125 million. By Q3 2025, this figure surpassed the full-year FY25 target of $250 million, implying 133% full-year growth from Q1 to Q4.

Extrapolating this run rate, pure AI ARR reaches $1 billion by Year 2 (approximately $1.36 billion), representing 5% of total ARR assuming it keeps growing at 11%. That means AI revenues will already become a sizable chunk of revenues for Adobe within 2 years, and will likely only become more so.

Notably, management ceased reporting this metric after Q3 2025, likely due to the difficulty of separating “AI-first” from “AI-influenced” revenue as products became more integrated.

D. AI-Influenced ARR

AI-influenced ARR grew from $3.5 billion in FY2024 to approximately $8.4 billion exiting FY2025. Management stated this represents approximately one-third of the “book of business.” Assuming this ratio applies to total ARR of $25 billion yields us $8.4 billion AI-influenced, representing 140% year-over-year growth. That’s substantially higher than overall ARR growth.

Optionality: The GEO Market and SEMRush Acquisition

GEO refers to Generative AI Engine Optimization, or engineers business's digital asses to get favored by chatbots like ChatGPT and OpenAI. This is a brand new industry growing alongside the more established Search Engine Optimization (SEO).

Market Opportunity

Global SEO spend ranged from $65.87 billion to $89.10 billion in 2024 depending on reports.

Bain research shows that 80% of search users rely on AI summaries 40% of the time. Napkin math implies that 32% of searches now occur through AI interfaces. Applied to the SEO market, this represents a $28 billion annual GEO opportunity that is growing rapidly.

As Bain noted: “Brands must evolve or risk losing visibility into their customer journey, and control over their brand positioning, in a world where traditional clicks are disappearing.”

McKinsey research suggests that up to 50% of brand traffic is at risk due to GEO, potentially impacting $750 billion of consumer spend.

We can safely assume that AI search will continue growing in important, we can make a safe assumption that GEO will become just as important to businesses as SEO, with the spend reaching comparable amounts. This is especially true as AI search retail conversion rates start matching traditional search:

Strategic Integration: Controlling the Full Marketing Funnel

The SEMRush acquisition allows Adobe to control three critical stages:

1. Content Generation through Creative Cloud

2. Distribution and Analytics through Experience Cloud

3. Optimization for AI Visibility through SEMRush and Adobe’s LLM Optimizer

Adobe recently announced an LLM Optimizer tool within Experience Cloud to tailor marketing assets to LLM crawlers. SEMRush, which has been an early pioneer of GEO analytics (along with competitor SimilarWeb) builds on Adobe’s LLM Optimizer strategy for AI-era brand visibility.

The acquisition provides optionality in a rapidly expanding market that Adobe is uniquely positioned to serve through its existing customer relationships and integrated product suite. I also believe that GEO still isn’t getting as much recognition as other areas of consumer AI. While people are trying to guess what the future of search looks like, its clear that AI will be a key part of it and so will GEO.

Comparables and Valuation

Adobe’s closest competitors among large legacy enterprise software providers are SAP, Oracle, and Salesforce. As we see above, Adobe has among the highest revenue growth, highest gross margin, and highest return on invested capital in this cohort. These key performance indicators have remained stable over three and five-year horizons.

Adobe also sports significantly better metrics than the S&P 500 average, yet trades well below it.

Despite the Company’s fundamental strengths, Adobe trades at 58% of its peers’ multiple.

Unlike other legacy enterprise competitors which primarily rely on AI for pricing uplift on existing products, Adobe has a tangible path to increasing paying user count by expanding freemium offerings at low marginal cost with brand-new AI-first products (Firefly).

Valuation Analysis

Adobe is trading at its lowest multiple in a decade and 56% below its median PE from 2017-2024.

Running a reverse DCF using company-guided operating margins and tax rate, a terminal growth rate of 3%, and a WACC of 9.58% (from Aswath Damodaran’s software industry estimate), the market is pricing in approximately 4% CAGR revenue growth over the next five years. This is despite the company guiding to 10% growth next year and all of its end-markets growing faster than 4%.

The fundamental question here is: Do you believe Adobe can clear a 4% revenue growth hurdle?

What multiple should ADBE command? Those historic PE ratios are associated with a period when Adobe revenue growth was higher and the company was enjoying the tailwind of converting its business to a subscription basis. A lower PE ratio is justified given slower growth. The question is how low the PE ratio should be.

I don’t think Adobe should command its historical median PE of 48.2 due to the slowdown in revenue growth. However, Adobe is a much higher quality company than the S&P average (31.3 PE) and the average of its peers (36.6 PE). Growth may be slowing, but Adobe has a strong route to ARPU expansion through generative tokens, freemium conversion, upsells, new pricing on AI products enabled by its distribution network, and disciplined margin control.

I hate doing valuation work because any one assumption changing will completely change the output. So lets start with a basic scenario, assuming no multiple expansion (remember, Adobe is trading at decade lows!).

Base Case: 17% Total Return

Assuming steady-state multiples applied to Adobe’s 2026 guidance plus management’s guided buyback yield generates approximately 17% total return. This is all based on guidance, where management has historically been conservative, and announced buybacks, which management has also shown to be willing to be aggressive in.

Not bad for base case.

Now lets get optimistic and see what happens if ADBE reverts back to its peer group average (which in itself has contracted in the last couple of years).

Upside Case: 85% Total Return

Do I think this is what will happen? The truth is probably somewhere in between. Remember, the current price implies 4% revenue growth. If the market sees that ADBE is able to sustainably grow above that, multiples should expand.

Where will it land? Hard to say. Adobe could be one major product release away from catapulting itself to the forefront of the market’s mind. Perhaps next year the market starts focusing on the GEO opportunity and hones in on Adobe as one of the few ways to gain direct exposure to it.

How sure can we be that management matches or exceeds their revenue target?

From the Q4 2025 earnings call: “Q4 was a really strong quarter and frankly, starting to be this inflection in terms of, you know, as we see the leading indicators, what is happening across the leading indicators. You know, which gives us a lot of confidence. And that’s why, you know, when you look at the total Adobe AR growth target, you know, which translates to approximately 2.6 billion, that’s the highest beginning of the year guide for total net new AR.”

Adobe management is usually formal and conservative in their outlooks. To use the word inflection is pretty heavy language.

Risks and Mitigants

Enterprise Spending Slowdown

Adobe noted enterprise spending slowdown due to macro uncertainty. However, enterprise spending is cyclical and I would say already priced in. ADBE has been slumping continually since management made this comment.

GenAI Models Replace Photoshop

Current GenAI companies focus on image generation, not comprehensive editing. Adobe has integrated AI tools into their editing workflows. More importantly, enterprise customers remain wary of legal repercussions from using AI-generated images. Adobe’s indemnification and fully licensed dataset solves this pain point while competitors remain confined to the inspiration phase.

Vibe Coding Replaces Legacy Software

One explanation for SaaS multiple contraction is that AI-powered application builders could enable enterprises to create bespoke solutions. However, vibe coding currently suffers from quality issues with no built-in updates or customer service when systems break.

Enterprise reliability requirements make the cost/risk versus savings calculation unfavorable. As one developer noted at Where’s Your Ed At: “They frequently get weird and make strange errors, generally getting worse the further from ‘typical’ the work is. Because of the high error rate, I still have to read and verify every line, which is frequently much slower than writing them myself.”

Potential Risk: Freemium and the Top of Funnel

While Adobe’s freemium strategy shows promise, cheaper and free options like Canva represent genuine competitive pressure at the top of the funnel.

Competitive Context

• Canva: approximately 220 million MAUs in 2025

• Adobe (Firefly plus Express combined): 70 million MAUs by end of 2025

• Midjourney: approximately 20 million registered members

Adobe’s Growth Metrics

Adobe’s freemium offerings are growing rapidly:

• Firefly MAUs grew 30% quarter-over-quarter in Q2-Q3 2025

• Paid Firefly subscriptions grew approximately 100% quarter-over-quarter in Q2 2025

• Acrobat Web serves as an entry point for all PDF interaction needs and is growing 30% year-over-year

• Student access (traditional customer pipeline) is growing 70% year-over-year

Product Parity Progress

Adobe’s Firefly initially lagged rivals but has rapidly closed the quality gap according to TechRadar’s comparison. Additionally, their partnership strategy transforms the competitive dynamic. Adobe now houses OpenAI, Nano Banana, and Midjourney models within its ecosystem. Users access leading models without leaving Adobe’s environment.

The freemium model offers Firefly free with limited monthly Generative Credits, with upgrades to paid plans unlocking full access and serving as a gateway to professional offerings.

Conversion Funnel Indicators

Adobe now combines Express and Acrobat reporting, complicating direct comparisons. However, Express Organization wins growth has strongly outpaced MAU growth, implying successful conversion of trials to enterprise seats with improved monetization.

The risk remains that Canva’s significantly larger user base and aggressive feature development could capture would-be Adobe customers before they develop professional needs. However, Adobe’s strategy focuses on controlling the professional and enterprise segments where willingness to pay is highest, while using freemium offerings primarily as customer acquisition and brand awareness tools.

Microsoft Copilot provides a cautionary case study: as of August 2025, Microsoft had approximately 8 million active licensed users of Microsoft 365 Copilot. That’s a 1.81% conversion rate across 440 million Microsoft 365 subscribers, according to Where’s Your Ed At. However, Copilot became the second-most popular AI search engine, nearly tied with Google Gemini according to FirstPageSage. That is the power of distribution.

Several factors differentiate Adobe’s situation. Adobe operates a highly profitable non-AI-first business with over $20 billion revenue at 80% gross margins. AI features serve primarily as retention and ARPU expansion tools, not sole growth drivers. Adobe’s integration strategy (bundling AI into existing workflows) differs from the standalone product approach.

Conclusion

The market is pricing in 4% revenue growth over the next five years. The question is simple: can Adobe grow faster than 4%?

Given Adobe’s massive distribution and balance sheet, the structural tailwinds generative AI is providing the creative software segment , and multiple growth vectors withing Adobe’s product portfolio, the answer appears to be yes.

Thanks For Reading!

If you’ve found my work valuable, here are a few ways to support it, any help is greatly appreciated!

• Subscribe to my Substack - just sign up below

• Share feedback - let me know what I missed or could improve

• Spread the word - pass this write-up along to friends or colleagues

• Send me ideas - if I like them, I might write them up (win/win)

• Pledge - If I start seeing a good amount of pledges, I may even do this full time!

Subscribe now

Disclaimer: This article is for informational and educational purposes only. It is not investment advice or a recommendation to buy or sell any security. Please do your own research before making investment decisions.

Write Ups

1. - The Best of the Legendary Value Investors Club

2. - Just Call Them ()

3. - Trainline PLC-TRN.L

Articles

1. Chicago Fed - What Similarities between Siblings Tell Us about Inequality in the U.S.

2. Undervalued Shares () - THE CRAZIEST (UNKNOWN) EUROPEAN BANKING STOCK (I can’t/won’t stop reading about ways to play Russian exposure)

3. Gavekal Research () - US Exceptionalism Versus Chinese Uninvestibility (Part I)

4. Tradium - Hidden champion germanium: defence industry sees supply gap

Podcasts

1. Didn’t listen to anything particularly interesting today - happy to get recommendations!

1. - There Is a Risk More Terrifying Than an AI Bubble

2. - Understanding ROIC on Low Growth Businesses

3. - Oil Is Incredibly Cheap

4. - Market Wizards: 16 Conversations That Will Change How You Buy, Hold, and Sell

Articles

1. - Which genius from history would have been the best investor?

2. Kopernik Global - London Calling

3. Where’s Your Ed At - The Case Against Generative AI

Podcasts

1. The Value Perspective - Conversations with David Iben

2. The Value Perspective - Conversations with Django Davidson

I know the containership market well. It’s an industry I covered in my first post on Substack.

I was short in late 2024 until Liberation Day made everything go wacky. Since then, my views of industry deterioration have come to pass, and announcements in the last month have guaranteed supply-demand equilibrium, lower rates and lower stock prices ahead.

When I was short coming into 2025, I modelled industry utilization rates at 98%-100%. Like any commodity, the cure for high prices is high prices and the market couldn’t get much tighter.

Containership supply grew 10% in 2024. In any other year, shipping rates would have collapsed. So why wasn’t capacity utilization dropping? The Houthis and the Chinese.

The Houthi stranglehold on the Red Sea was eating up 6% of supply, as ships were forced to re-route around the Cape of Good Hope to reach the West from the East, a far longer and more expensive journey. On top of that, China went all in on exports to stave off depression from their deflating real estate bubble, leading to their strongest export year on record with 6% growth from an already massive base.

That’s how despite supply growing 10%, utilization barely budged and was nearing 100%.

When I went short in late 2024, Israel had just forced Hezbollah into a ceasefire and was re-directing her air force to directly target the Houthis. The IDF was progressing through Gaza, encircling the main Gaza City complex. The US had also begun bombing Houthi targets. I believed these factors would lead to the Houthis giving in by H2 2025, as the Gaza war was entering its last stage.

I was wrong.

The Houthis kept attacking ships while absorbing losses, including their prime minister. Despite this, shipping rates began slipping through Q1 2025 as gravity pulled the market lower. More supply was coming online, maybe not at the 2024 rate, but still at an expected 7% through 2025 - far more than any regular year. This time, Chinese exports couldn’t maintain their momentum and capacity utilization started dropping. The fundamentals for the short trade were lining up.

Then Liberation Day happened.

In one announcement, shipping stocks had their bottom fall out. Tariffs were an optional tail event I had considered in my thesis. Remember, this was before Trump had taken office and no one was taking his tariff threats seriously. I figured he could install tariffs, but nowhere near what he was threatening. At the time, I modelled that a 60% tariff on China would lower shipping demand by 1.6%, basic napkin math based on the last trade war.

For the next week, shipping stocks had 5%-10% up/down days as the market reacted to every rumour, report and political comment. I realized fundamentals had taken a backseat and I didn’t know when they would return, so I sold out of my position. I made ~35% in 5 months.

I believe fundamentals must take the front seat again and now is a good time to short.

Subscribe now

The Houthis recently announced they will stop attacking Red Sea vessels as the Gaza ceasefire takes hold. Remember, this one factor is eating up 6% of shipping supply. Despite this announcement, shipping companies remained coy. This makes sense as announcing a return to the Red Sea would mean lower rates for them.

On November 6, 2025, Maersk announced their Q3 2025 earnings. This came after the Houthis announcement, but Maersk struck a cautious tone:

“We’re also figuring out what is the posture of the Houthis specifically to see if we can start to have a safe passage”

Then, two days later, on November 8 the Rubicon was crossed. CMA CGM, a private listed conglomerate and the third largest containership fleet, announced they had begun transiting through the Suez Canal.

The announcement was coordinated with a big media blitz by the Suez Canal Authority. The incentives are clear: The Suez Canal Authority just finished a big renewal project and is desperate to attract traffic for much needed revenues. CMA CGM, riding on the Authority’s media coattails, effectively announced to all customers in the containership market that they can ship goods for a lower price than competitors.

Behind all the romanticism of shipping, it is a commodity business like any other, with pricing set by the marginal producer. They all offer the exact same product. If one shipper can offer the East-West route at a fraction of the price, all customers would flock to them.

The Beginning of the End for Container Shippers

When a major pulls a move like this, all the other players must follow to stay competitive or risk losing all their customers. ZIM announced their results afterwards, striking a more open note to entering the Suez:

“A return to Suez is increasingly likely”

They also made clear they see rate softness continuing into 2026.

The final domino was Maersk, which after watching the majors make commitments to begin sailing through the Suez Canal, offered an updated view on November 28:

There is no going back to the artificially high rates of 2024.

The major reason containerships did well from 2024-2025 is now gone. Rerouting through the Cape of Good Hope was eating 6% of capacity. Demand growth cannot match that, as China begins a policy of capacity normalization, China tariffs stay higher for longer, and Europe lurches from crisis to crisis.

Supply is set to grow 5% yoy in 2025, with an orderbook standing at 30% of the outstanding fleet. Translation: It will take years to digest all the ships coming on-market. In 2024, demand grew ~4% and all the majors in their earnings releases expect softer demand through 2026. Simple arithmetic dictates that supply is outgrowing demand and that capacity utilization must fall, and with it, rates.

Then you have re-routing normalization. I believe this will take 1-2 quarters, as ships must complete their current voyages before going through the Red Sea again. But once they do, approximately 6% of capacity will be released. There is now no reason for rates not to return to their pre-2024 equilibrium price.

I’m modelling industry capacity utilization dropping from 96% in FY25 to 90% in FY26, assuming re-routing gradually ends through 1H 2026. This 6% may not seem like much, but in a commodity business, that amount of slack in capacity will have disproportionate impacts on pricing.

From Thesis to Execution

Shorting however, isn’t easy. Shorting an industry with only three real majors is even more difficult. Your three liquid options are Maersk, ZIM and Hapag Lloyd.

Maersk has always been valued at a premium to peers due to their robust contract structure and diversified terminals and logistics business. Looking at their price history from end of 2024, they held up surprisingly well, especially vs the other two names. I still believe they will decline in value, as their contracts have to reset, with management pointing to February as a major date of negotiations. The Company will likely remain resilient until it becomes clear their contracts are re-rating lower.

ZIM has been my favorite way to short the industry as they are far more concentrated. However, ZIM today is in the midst of a bidding war for itself. I would avoid shorting them as the increasingly high bids make blow-up risk real. That leaves German Hapag Lloyd, which has also bid on ZIM.

The fundamental picture for containerships has clearly gotten worse. Companies are openly stating that the good times are over, yet in the last month all three companies are up 8-15%, perhaps due to ZIM’s buyout prospects, which has attracted a lot of industry attention.

I think now offers a strong point of risk:reward for a short until rates normalize to pre-Oct 7 levels. I am also happy to add shorts to my portfolio, as I believe markets are wobbling.

A highly correlated and even more concentrated market is Car Carriers, which frequently use the same trade lanes and have been impacted similarly by COGH re-routing. The industry is facing the same supply overhangs, if not worse than containerships. Be aware that the companies here are even smaller than container shippers, and there is no public rate index for car carriers, making pricing more opaque.

Thanks For Reading!

If you’ve found my work valuable, here are a few ways to support it, any help is greatly appreciated!

• Subscribe to my Substack - just sign up below

• Share feedback - let me know what I missed or could improve

• Spread the word - pass this write-up along to friends or colleagues

• Send me ideas - if I like them, I might write them up (win/win)

This content is for informational and educational purposes only. It is not investment advice or a recommendation to buy or sell any security. You should do your own research and consider your own financial situation before making any investment decisions.

1. - Energy Barons of the Basin

2. - Liquidation play for a 20% IRR

3. - Inside Michael Mauboussin’s ROIC Framework

4. - Everyone is Gambling and No One is Happy

Articles

1. Myrmikan Research - This Time It’s the Same

2. Howard Marks - Is It a Bubble?

3. Bloomberg - AI’s Future May Be Written in Railroads’ Past

Podcasts

1. - What Travis Kalanick Taught Bradley Tusk, & Why He Closed His VC Fund

2. Cheeky Pint - Dan Sundheim of D1 Capital on the Art of Public Market Investing

1. - How GOLD Bull Market Corrections Look

2. - (Macro) The Chemical Sector Bottom And The Path To A Recovery

3. - The Turkey’s Done. The Dollar’s Cooked. And This Could Make You Rich Forever.

4. - The Anti-Conglomerate

5. Your Truly - ANALYSIS: Peloton (PTON)

Articles

1. Howard Marks - Cockroaches in the Coal Mine

2. The Economist - Vineyards are disappearing in France

3. American Bullion - Central Bank Gold Buying in 2025

4. Reuters - Flutter says UK gambling duties to hit EBITDA by $320 million in 2026

Podcasts

1. The HC Commodities Podcast - Wine With Nick Brewer

2. Invest Like The Best - Building An Entertainment Empire (Barry Diller)

All quarters and years follow the Company’s calendar. Q2 2026 corresponds to Calendar Year Q4 2025.

Peloton needs no introduction. It’s THE symbol of the Covid bubble: the stationary bike with an iPad attached and a textbook case of why CEOs shouldn’t extrapolate trends indefinitely.

Now it’s forgotten. Down 95% from its 2021 peak, people universally mock my interest: “I would never buy that, it’ll go to zero”. That dismissal creates opportunity.

People are overlooking a Company with churn rates better than Spotify and Netflix, despite charging 1.4x-2.6x more monthly, with a fanbase that idolizes the brand and content (spend 5 minutes on the Peloton subreddit page).

New management has right-sized operations, enabling growing profits on declining revenue. Their debt has been refinanced and the Company now has the balance sheet to handle it, with a clear path to reduction.

What pulled me in: Peloton announced its first price increase since 2022. Given subscriber loyalty, this move underwrites downside risk and may catalyze revenue growth, the elusive final piece required for revaluation.

Business Overview

I break Peloton’s history into 3 distinct phases, each with different management teams and goals.

Phase 1: Brand-Building

The original CEO spared no expense getting the word out that Peloton isn’t just a product but a lifestyle. It worked to a degree. A Peloton became a symbol of affluence, an image that remains with the upper-middle class and beyond. This phase overstayed its welcome. Costs ballooned and management went on a spending spree building manufacturing capacity that wasn’t needed. The massive excess inventory and demand collapse nearly killed the Company.

Phase 2: Turnaround (FY2023-FY2025)

Barry McCarthy was brought in to stop the bleeding and rebuild the business model around profitability instead of member growth. The results speak for themselves. Gross margins went from 33% in FY23 to 51% in FY25. Operating expenses as a percentage of revenue declined from 76% to 52%.

That’s 42% total margin improvement in two years. Management consistently hit cost savings goals ahead of schedule. Even now, Peloton is targeting another $100m in annual savings, half achieved within a quarter of announcement. Their FY25 operating expenses were $800m lower than FY23, a 40% reduction.

Margins improved from product mix as more revenue came from high-margin subscriptions and premium device sales. Peloton launched premium versions of their products plus a treadmill and rowing machine, both priced higher than the bike. The Company’s separate margin disclosures confirm customers are buying these higher-priced products, despite Peloton developing a second-hand marketplace to expand the customer pool.

On operating expenses, Peloton closed retail locations and converted them into micro-retail spaces within malls. Company commentary indicates these generate roughly the same revenue at a fraction of the cost. Peloton now has 10 micro-concepts. Sales and marketing saw major cuts as Peloton shifted from member maximization to cost reduction, heavily scrutinizing advertising channels.

Phase 3: Growth

This phase focuses on revenue growth from a healthier base. It could be marked by current CEO Peter Stern’s arrival in FY2025, but the Company is still transitioning. The Company is guiding to -2% year-over-year revenue growth, despite Q1 26 revenues declining -6% year-over-year. This implies positive revenue inflection in coming quarters.

This growth stems from a healthier operational foundation and controlled cost structure, which has driven consistent improvement in Peloton’s lifetime value to customer acquisition cost ratio. A ratio over 2x is considered healthy, and the Company has stated this as their goal. I am forecasting that ratio to reach 1.92 by end of FY26. Previously, this growth was driven by declining acquisition costs. In FY26, I am modelling growth in lifetime value as well, which should be the catalyst to reach that ratio.

Key players: Stern and new CMO Megan Inbres (hired Q4 2025). In Q3 2025, Stern stated his CMO would focus on top-line growth, a shift in tone and strategy. Megan is a strong pick. She worked with Peter at Apple running in-house marketing and communications, and previously headed marketing for Netflix Originals. This is the perfect intersection: Peloton needs someone with subscription-based consumer background who knows how to leverage their growing content library.

Peter Stern worked at Apple as Head of Services Marketing and at Ford as President of Integrated Services, giving him unique exposure to consumer services and hardware. What struck me more: he’s been a Peloton user for 8 years. He intimately understands what Peloton does right and what needs fixing. The Company updated compensation policy to reduce salaries in exchange for higher bonuses via stock tied to operational targets, one disclosed metric being free cash flow generation which is better to see than binary stock price targets.

All of this makes the holiday Q2 26 pivotal for Peloton. The foundation is set; now we need to see if the team can execute. The Company seems to be preparing with their first complete lineup update, higher prices, and signals to increase Sales and Marketing for the first time since cost cutting began.

Market

The luxury at-home fitness equipment market isn’t a massive opportunity. The upside: it’s more likely to be winner-take-all rather than commoditized. Peloton has the brand recognition over competitors.

In Company commentary, they defined their target US market as households with income above $75k who spend on fitness: 117m households. Further qualifying to those also willing to spend on fitness subscriptions and home equipment narrows the target to 17m households. This gives you a rough idea of the maximum size of their business. Lets say you multiply 17m * $44 (subscription cost) = $748m/ month or ~$9bn annually. This excludes device sales.

Peloton has 3.3m subscribers, or roughly 20% of their target TAM. This is a good starting point, but don’t forget that this is a company that has been losing subscribers since the Corona peak.

Catalysts

Price Hike

Peloton’s price hike sparked my interest. Initially, I assumed this was a last-ditch effort to salvage the business, maybe an attractive short. Upon inspection, the Company is in much better shape than assumed and has the pricing power to push this through with minimal blowback.

I modelled a sensitivity table for their subscription business with the planned price increase versus various quarter on quarter churn scenarios to understand potential gains under different churn rates.

Note: I’m assuming a weighted average price increase of 14.4% across connected fitness and paid app offerings.

In steady-state, Peloton’s subscription revenues would rise 6% despite a -7.5% quarterly drop in users. For perspective: since FY23, Peloton’s Connected Fitness subscription churn has averaged 1.5%, and 1.6% last quarter (Or roughly 4.8% quarterly churn). I expect elevated churn, but the question is: how high? At 12.5% churn, the price rise is neutral short-term. In the mid-term, every marginal user acquired will have higher Lifetime Value.

Bottom line: Will Peloton’s churn triple from the price hike? Unlikely.

Peloton’s churn is the envy of the consumer industry:

Despite Peloton’s subscription being far more expensive, customers love the product:

This shows in the socials data: Reddit members in the Peloton subreddit grew 15% year-over-year. Beyond their suite of content for fitness programs, these die hard fans are Peloton’s moat.

Finally, examine Company guidance for Q2 2026 (Holiday season): Management has insight into member response to price hikes and guided to 8% quarterly decline in subscriptions at the midpoint. That would leave the Company with ~6% more revenue purely on the price hike. Additionally, in the last two years, management has been conservative, looking to under-promise and over-deliver. I don’t see this changing.

The price hike gives investors a margin of safety:

This is key to my investment case: price increases flow directly to the bottom line.

Taking the company’s own adj. EBITDA guidance and Peloton’s current EV/Adj. EBITDA multiple, operating leverage amplifies the pricing effect. Even with 7.5% quarterly user decline, at current multiples, Peloton should be worth 12% more. This completely excludes any device sales.

Of course, this is all theoretical, but it’s important to quantify the benefit the Company can earn by raising prices and at what point the price hike harms the company.

Sales Growth

Company guidance for the first time implies an inflection to positive revenue growth through quarters 2-4. How realistic is this?

As noted, the price increase alone should push QoQ revenues by +3%, assuming nothing else changes and 10% quarterly churn. Peloton has updated its entire product lineup for the holidays, using the holiday season as their springboard. This includes their treadmill device, important as the treadmill market is roughly 3x the stationary bike market.

These products are at higher price points and consumers have been shifting to Peloton’s more expensive products, not cheaper ones. Peloton appears to understand the importance as the Company stated Sales and Marketing spending would increase in Q2 to ensure a successful holiday campaign. I’ve estimated sequential improvements in the Company’s LTV:CAC, with LTV set to get a boost from the price increase, Peloton should have the room to increase S&M and have that translate to incremental net revenue.

Peloton has long experimented with refurbished returned units with success. At one point these made up 30% of new device sales, per the Company. Recently they developed and released an entire second-hand marketplace they control. They saw demand for used models, which expands potential market, and leaned in by controlling the entire process. Their platform shows device history, includes Peloton’s installation and delivery service, and provides access to their highly-rated customer service.

The Company also instituted a one-time activation fee, increasing margins on the traditionally lower-margin used segment. This marketplace is being trialled in key markets but should contribute meaningfully in coming quarters.

These factors create a real possibility of positive year-over-year revenue inflection and according to my FY26 forecast, the Company’s first year of operating profit. If achieved, I believe the company will undergo significant revaluation as the market values it as a sustainable subscription-based DTC brand.

Looking beyond FY26, there’s one more significant lever the Company hasn’t pulled: international foreign language expansion. Peloton’s non-US markets are all English-speaking, besides Germany which is entirely managed by third-party distributors. Peloton, like Netflix, already has a vast content library. It just needs dubbing into foreign languages (Spanish, for example) to multiply potential market. With current AI tools, this dubbing could be done at high quality and minimal cost. This is the logical next step once the Company proves healthy, sustainable sales growth domestically.

Bottom Line: Q2 2026 is a defining quarter. Historically, Q2/Holiday season represents ~33% of device sales. This sets the tone for Peloton’s high-margin subscription business for the rest of the year. A weak Q2 showing would likely force retreat on guidance, while a strong showing will lead to a high probability of a strong annual beat. This is probably their best shot in years to outperform.

Comparables/Competition

Peloton has no direct comparable. Technogym is closest. While they produce high-end home gym equipment with a new, small subscription business, they focus more on Europe and have a sizable commercial equipment business. Despite lower sales and a smaller subscription business, it trades at a higher multiple to gross profit than Peloton, no doubt due to Peloton’s declining revenues.

The other comparison, as I’ve done above, is looking at other DTC brands. Some examples of the stark difference, keeping in mind that Peloton has lower churn and higher subscription prices:

One argument for the discrepancy is the finite lifetime of Peloton’s core cohort. Due to targeting affluent customers willing to buy an expensive machine and maintain a subscription, their customers likely skew older. This impacts terminal LTV calculations as Peloton’s customer lifetime runway is shorter versus a Spotify or Netflix. A Netflix customer starting in their 20s may subscribe for 40-50 years, well into their 70s. A Peloton customer joining in their late 30s may only use the product for 20-30 years before physical limitations cause them to leave.

This is valid, and PTON shouldn’t reach Netflix/Spotify multiples because of it. But how much more valuable is Peloton’s LTV today? I estimate Netflix’s LTV (LTV calculated as Monthly Subscription Price * Subscription Gross Margin% / Monthly Churn%) at $423 per customer, 79% lower than PTON. This makes sense: Peloton’s ARPU is multiples higher, and both gross margins and churn are superior.

The question: how much should Peloton’s 3.3m paying members be valued at given the average $50 monthly price and very low churn?

Valuation

I don’t like valuing businesses; I like investing behind change.

Valuation is nice for seeing a range of outcomes, but it’s all speculation. I want to invest where business fundamentals will improve over the next 18 months, or at least the range of potential outcomes has narrowed to have positive expected value.

That said, let’s get perspective. I’ve made the case that the price hike should give a 12% valuation uplift if quarterly churn stays under 7.5% on an increase to ARPU alone. That’s your downside covered.

What about upside? Let’s run a sum-of-the-parts valuation between gym equipment manufacturers and consumer subscription brands:

Note: despite over 80% of their gross profits coming from subscriptions, their multiple is more in line with gym equipment manufacturers than consumer brands.

Its easy to point out that Peloton’s multiple is low compared to peers. Today, there is good reason for that – declining revenues, subscribers declining, sub-optimal LTV:CAC. I believe all three of those factors have the highest probability in years to reverse, leading to a healthy re-rating.

Risks

All eyes will be on Q2 26 which is the holiday season to see if Peloton can turn around their sales growth. If we see that hardware sales declines again year on year, I believe investors will have to take a lot more caution with Peloton.

Without a strong Q2 showing, which historically drives ~33% of hardware sales, I find it hard to believe the Company will hit their FY revenue goals. Beyond that it would raise serious questions as to whether there is real sustainable demand for their products, considering they have been losing customers for 3+ years now. Its as of yet unclear if the majority of those users where Corona subscribers who would have never stuck around or if there is a legitimate issue with the product.

On the other hand, if they manage to generate positive yoy sales growth, I think the stars will have aligned for a classic re-rating story, as those activations would likely filter into Q3, driving subscriber growth in H2 2026.

Conclusion

At the end of the day, here’s a business left for dead that has quietly rebuilt its financial structure to generate high levels of free cash flow from each marginal user. Now it needs to shift into growth mode and prove to the market that it can generate revenue growth.

Peloton has a better shot at that now than it’s had in years. Q2 2026 will be defining. With strong performance, I predict multiple quarters of beats. I’m in it for the turnaround story and the re-ratings vs considering this a ‘wonderful’ business. This is still a luxury niche and should be treated as such.

That said, perhaps they can parlay momentum into successful international expansion, or (less likely) begin focusing on shareholder returns with a stable, highly valuable subscriber base that spits out free cashflow at high margins.

Thanks For Reading!

If you’ve found my work valuable, here are a few ways to support it, any help is greatly appreciated!

• Subscribe to my Substack - just sign up below

• Share feedback - let me know what I missed or could improve

• Spread the word - pass this write-up along to friends or colleagues

• Send me ideas - if I like them, I might write them up (win/win)

Subscribe now

Disclaimer: This article is for informational and educational purposes only. It is not investment advice or a recommendation to buy or sell any security. Please do your own research before making investment decisions.

1. - Peak Gold

2. - I put half my portfolio in one stock

3. - YWR: China Trip Highlights

4. - Peloton Earnings Preview: Record Subscriber Loss in Sight as Subscription Headwinds Mount

Articles

1. Financial Times - Houthis signal halt to Red Sea shipping attacks

2. Energy Connects - Gas Turbine Shortage Is Forcing Companies Back to Coal

3. Hosking Partners - Q3 2025 - Quarterly Report Commentary

Podcasts

1. Founders - Losing My Virginity: How I Survived, Had Fun, and Made a Fortune Doing Business My Way (Richard Branson)

2. My First Million - Howard Marks: 79 Years of Investing Wisdom in 55 Minutes

   1. The best performing funds over the long-term are those that are consistently in the second quartile and never in the fourth quartile

      1. Less losers, rather than more winners

   2. HY Debt as an alternative to equities when valuations are stretched

Books

1. Alan S. Binder - After the Music Stopped: The Financial Crisis, the Response, and the Work Ahead

   1. At 500pgs, expect this to stay on the list for some time

The company has built a strong position in unattended payments but the growth engine that got them here is showing signs of fatigue. My base case points to Nayax falling short of its own guidance, with a sum-of-the-parts valuation implying roughly fair value.

At its core, Nayax’s business depends on selling point-of-sale (POS) terminals, which in turn feed a recurring payments processing business. That model has worked well for years, but POS sales, the foundation of future recurring growth, are now slowing. Management’s acquisition strategy has yet to generate enough inorganic revenue to offset that. Meanwhile, the EV charging vertical, which management promotes as the next major growth driver due to its much larger average order value, looks at least two years away from becoming a meaningful contributor.

Competition in the vending market is also heating up, pressuring Nayax’s pricing power and customer acquisition costs. When combined with a stock that’s already priced for perfection, trading above sector averages, the setup doesn’t offer a compelling risk-reward in my eyes.

Valuation

My valuation uses a sum-of-the-parts (SOTP) approach, applying a segment-weighted average EV/Sales multiple of 4.1x to forecast FY25 EBITDA. Each Nayax business segment is benchmarked against an average of three pure-play peers. After deducting net debt and dividing by shares outstanding, this yields a target price of $43.52 per share - about where the stock trades today.

That target implies an EV/NTM EBITDA multiple of 17.7x, well above the recent acquisition average of 14.8x across comparable payments companies. Combined with stretched valuations across the Israeli equity market (currently showing their highest P/E ratios since 2007), Nayax’s multiple looks difficult to defend.

Business Breakdown

Nayax’s flywheel starts with POS terminals. Each new terminal installed generates future recurring processing revenue, creating a compounding effect. The problem is that incremental device sales are now representing a smaller share of the total installed base. As this mix shifts, the flywheel begins to slow.

Above, you can see that Nayax was incrementally adding ~10% to its installed POS base in 2022, and that has fallen to ~3% this year.

POS Terminals

Tariffs are another headwind. Nayax manufactures in the Philippines, which has been hit with a 19% tariff. Assuming U.S. POS sales remain around 38% of group sales, the tariff adds roughly $1.2 million in quarterly cost, cutting POS gross margins by about four percentage points to 31%.

Competition is most intense here. Average selling prices (ASPs) have fallen from $94.80 in 2022 to $72.80 in 2024 — a 23% decline. Interviews and industry commentary suggest that as Verifone and Incongenti enter the unattended space, price pressure has only intensified. Even on vending community forums, Nayax salespeople are reportedly offering deep discounts to win customers.

Customer acquisition costs tell the same story. A rough calculation using SG&A per new customer shows CAC up roughly 66% between 2022 and 2025. Speaking to the Nayax team, I was told that they are lowering price to take market share. What if they’re lowering price in an attempt to defend their market share?

Processing

Nayax’s processing take rate is 2.7%, roughly half of Cantaloupe’s 5.1%. That’s surprising, since unattended transactions typically command higher rates due to increased fraud and tampering risk. The explanation lies in geography: Nayax’s expansion into lower-interchange regions like the EU and Latin America drags the average down. In those markets, caps on interchange fees are government-mandated, whereas the U.S. (their highest-rate market), makes up a smaller share of new growth.

Management also confirmed that Brazil’s average transaction value is significantly lower than the U.S., which further limits fee income as Nayax is taking a bold bet on Brazil.

Market Overview

Vending

The company’s focus remains squarely on SMBs, which account for about 26% of total transactions. According to industry calls, Nayax has chosen not to pursue enterprise customers aggressively, as they struggle to compete on price and integration depth.

EV Charging

The EV charging vertical is where management sees the next leg of growth. It’s easy to see why: transaction values are much higher ($7 versus roughly $2 in vending), and POS terminals are “stickier” since switching requires licensed technicians. EV infrastructure growth is also strong - U.S. charging stations rose 24% year over year in 2023, and European stations grew 38% in 2024.

But timing matters, and I believe Nayax’s EV ramp will take longer than they imply. Management has acknowledged that EV charging contributes a negligible share of current revenue. Their optimism centers around regulatory trends mandating card payments on public chargers but the current state of adoption tells a different story.

The U.S. government’s own data shows that only about 8,000 of 90,000 public charging stations currently accept card payments. Instead, most operate via phone apps that must be charged with funds ahead of time. This was pretty shocking to me. Even with strong growth, it will likely take two to three years before card-enabled terminals reach critical mass.

In Europe, the situation is somewhat better, but still far from universal. User reports on Reddit and charging apps confirm limited adoption of direct credit card readers across key markets.

Nayax’s partnership with Autel, a Chinese OEM, illustrates the uncertainty. The deal calls for 100,000 units, but many of these terminals are destined for residential use, meaning the Nayax POS may never be activated. Most consumer chatter online relates to home chargers, not commercial ones.

Adding to that, the U.S. House Select Committee on China has requested a probe into Autel’s U.S. operations, which could put the entire partnership at risk.

Another partnership with BTC Power looks promising at first glance. The company claims to own 25,000 stations processing 12 million transactions annually. Assuming an average transaction value of $7, that equates to roughly $84 million in transaction volume. Applying Nayax’s 2.73% take rate would yield about $2.3 million in processing revenue — just 1.4% of my FY25 estimate — and that’s assuming Nayax terminals are installed across the entire fleet, which seems optimistic.

Management has publicly stated a goal of generating 15% of total revenue from the EV vertical by 2028. Based on their current trajectory, that would require EV transactions to represent 27% of total payment volume, up from essentially zero today. Given the regulatory, logistical, and partnership challenges outlined above, I don’t believe they’ll reach that target in time.

Final Thoughts

Nayax is a well-run company operating in an attractive niche, but the narrative of frictionless growth is showing cracks. The POS business is maturing, competition is biting, and the next growth engine, EV charging, is still too far out to matter. While management deserves credit for building a global footprint and strong technology stack, the stock’s current valuation leaves little room for disappointment or excitement.

If the EV vertical begins to scale by 2027, Nayax could re-emerge as a compelling growth name. But until then, I see better opportunities elsewhere.

Thanks For Reading!

If you’ve found my work valuable, here are a few ways to support it, any help is greatly appreciated!

• Subscribe to my Substack - just sign up below

• Share feedback - let me know what I missed or could improve

• Spread the word - pass this write-up along to friends or colleagues

• Send me ideas - if I like them, I might write them up (win/win)

Disclaimer: This article is for informational and educational purposes only. It is not investment advice or a recommendation to buy or sell any security. Please do your own research before making investment decisions.

Other applications are less visible but just as impactful: optimizing ambulance routing, automating hospital bed allocation, and reducing paperwork bottlenecks. Together, these trends fall under what RFK calls “Make America Healthy Again” - a push to make healthcare more accessible and less wasteful.

COVID supercharged telemedicine, for obvious reasons. What’s notable is how sticky it has been: as of 2024, one survey has found that 54% of Americans had used telehealth services. Many companies rushed in during the boom to cash in on the hype and a new remote-first future for healthcare, and most of their stocks collapsed when reality set in.

One of those “fallen angels” is American Wellness Corp, or Amwell (AMWL).

At first glance, Amwell looks like another casualty: its stock down over 90% from the highs, revenues bumpy, profitability elusive. But beneath the surface, the company has transformed its business model, trimmed costs, and just landed a critical government contract. On top of that, it trades below its net cash balance, making it one of the rare “hidden net-nets” in today’s market.

From Pay-Per-Visit to SaaS

Amwell started with a simple model: reroute patients to healthcare professionals and take a cut of each visit. This pay-per-visit model was unpredictable, low margin, and reliant on competing for partner contracts, with ever expanding costs.

A couple of years ago, Amwell pivoted. It launched an integrated software-as-a-service (SaaS) platform that healthcare providers pay for on subscription. The transition wasn’t painless: some customers balked, revenues dipped, and the post-2021 tech bust crushed the stock. Adding onto that, AMWL has begun disposing of different business units in its Visits segment, optically hitting its revenues in the short term.

But today the benefits are showing:

• Subscriptions now dominate revenue. In Q2 2025, 57% of revenues came from SaaS, the highest yet

• Margins are expanding. Subscription gross margins run 75–90%, compared to just ~25% for the old visits business.

• Revenue mix keeps shifting. Even if total revenues grow slowly, quality is improving quarter after quarter.

Traction and the DoD Contract

Beyond the business model, Amwell has begun to win meaningful deals. Most notably, it secured a Department of Defense contract to provide telehealth services for U.S. soldiers in the field.

That deal took nearly two years of trials and integrations, but the DoD has now renewed it. Government agencies are notoriously sticky once systems are in place, especially in sensitive areas like healthcare. The contract may only be for a year, but the odds of unwinding it soon are slim. If the system works, and the renewal after a near two-year trial suggests it does, inertia and switching costs become a moat.

Management is also targeting other agencies, leveraging its new credibility as a trusted government partner. While I am sceptical whether success with the DoD automatically opens the doors to other agencies, it does show that Amwell has the political chops to deal with these agencies and the staying power to handle such long trials.

The DoD signing a company that isn’t even cashflow positive is another interesting point. I suppose this is another reason why AMWL is holding onto its cash-pile so tightly – it likely contributed to the DoD’s confidence in Amwell as a going concern.

Cutting Costs, Building Leverage

Meanwhile, Amwell has slashed expenses across the board. R&D as a percentage of revenue has fallen from 41% in FY23 to 29% in FY25. Headcount and operating expenses are down too, helping the company move closer to cash flow breakeven.

Management is explicit: the goal is to get lean, hit cash flow positive, and then shift back to growth mode. Based on current trends, I model them crossing that milestone in 2026, in line with company guidance.

This discipline matters. Many busted IPOs in healthcare tech simply burned through cash until markets gave up. Amwell, by contrast, banked cash early and has preserved it. The company still has no debt, making it one of the few players positioned to benefit from operational leverage rather than drown in costs.

A Hidden Net-Net

Here’s where things get interesting.

Amwell trades below its net cash position, even after adjusting for deferred revenues (listed as a liability). In theory, a net-net implies the market thinks the company is worth more dead than alive.

That’s often true of companies with shrinking revenues, weak products, and no plan. But Amwell is the opposite: revenues are growing again, margins are rising, and cash burn is narrowing. The DoD deal shows its product has staying power.

In value investing terms, this is a hidden net-net: a business with more cash than market cap, but unlike typical “cigar butts,” it has genuine prospects.

Valuation

Valuing a pre-profit company is tricky. I’ve modelled Amwell to adjusted EBITDA as a rough proxy for cash flow. In Q2 2025, they beat both my estimates and consensus. Guidance for FY25 remains conservative, and I expect further beats, even after accounting for less than expected services within the DoD contract.

A few key numbers:

• Subscription gross margins: ~83% based on management commentary in Q1 2025, where they stated subscription margins range between 75%-90%. I took the midpoint.

• Visits gross margin: ~25%. Taking the Company’s total gross margins for that quarter of 53% and the 83% margins in the subscription business implies 25% margins for their Visits business

• I forecast Subscription revenues growing double-digits buoyed by the DoD contract and new signups such as Florida Blue

• I forecast Visits declining double-digits (by design). Amwell has reiterated their guidance for full year visit numbers. Taking that multiplied by their long-term average revenue per visit gets me to a sequential double-digit decline. Remember the sooner they dispose of their visits, the sooner they can enjoy SaaS-style margins

I model overall revenue growth in the mid-single digits, with an inflection in cash flow from operations profitability by 2026. The more interesting story to me is 2026, when the company becomes cash flow positive, which is when I believe the market will have to wake up and rerate the company.

Comparables:

• Teladoc: larger, EBITDA-positive, but suffered year-on-year revenue declines trades at 0.53x P/S.

• Privia Health: similar SaaS healthcare model but aimed at physicians, profitable, trades at 1.4x P/S.

• Amwell: currently trades at 0.41x P/S.

If Amwell proves cash flow positive in 2026, with continued strong revenue growth and margins getting closer to their overall subscription margins, a re-rating closer to 1.0x P/S seems reasonable, but it will take a string of continued successes for the market to finally warm up to AMWL.

As the company is pre-EBITDA, perhaps a better way is valuing AMWL at Net Current Asset Value (NCAV) (taking out deferred revenues), which gets us to a target $15.1/share, 119% upside and a 46% discount to NVAC. This is also an implied P/S of 0.89, which allows us to triangulate a ‘reasonable’ multiple’. As I mentioned above, I don’t think AMWL deserves to trade at a 46% discount to NCAV.

Why the Opportunity Exists

Why does the stock still trade this cheap? A few reasons:

1. Burned investors. Amwell was a pandemic darling. Institutions bought at inflated valuations, lost big, and swore off the name. PTSD keeps them away.

2. Painful transition. Revenues fell and expenses ballooned during the SaaS pivot. Many assume the business will never be profitable.

3. Cash skepticism. The company has guarded its cash pile closely, raising concerns it’s just hoarding. But in context, it’s part of a deliberate strategy to survive the transition.

This combination creates alpha: the market is still treating Amwell like a broken story when in fact the fundamentals are improving. Investors getting in now will get the benefit of jumping in after their major R&D round and after the painful roll out of their new business model.

Final Thoughts

Amwell is a classic “fallen angel”: a stock everyone loved during the pandemic, hated after the bust, and now ignored despite improving fundamentals. The business has pivoted to SaaS, expanded margins, landed a marquee government contract, and slashed expenses.

At the same time, it trades below its net cash. That disconnect won’t last forever. If management executes and hits cash flow positive by 2026, the market will have to rerate the stock.

The beauty of investing in fallen angels is that expectations are already so low. You don’t need perfection - just proof of survival and progress. Amwell looks like it has both.

Thanks For Reading!

If you’ve found my work valuable, here are a few ways to support it, any help is greatly appreciated!

• Subscribe to my Substack - just sign up below

• Share feedback - let me know what I missed or could improve

• Spread the word - pass this write-up along to friends or colleagues

• Send me ideas - if I like them, I might write them up (win/win)

Disclaimer: This article is for informational and educational purposes only. It is not investment advice or a recommendation to buy or sell any security. Please do your own research before making investment decisions.

A few months ago, I closed a position that checks all the boxes for a great short: clear catalysts, good risk/reward, and room for error.

Subscribe now

Containerships - On Top of the World

I’ve always liked shipping as an industry to analyze as it’s a pure supply-and-demand story. Supply is the number of ships (or container slots), demand is the global need to move goods. Small imbalances swing rates wildly, which makes for big stock moves.

By late 2024, container shipping was on top of the world.

• Suez Canal closure: Houthis in Yemen had forced ships to reroute, adding days to voyages.

• China’s export push: To offset a weak domestic economy, China was exporting at record levels.

I estimated global container ship utilization at 96–100% in 2024, essentially full capacity. Earnings were booming, stocks were rallying. And like every cyclical market, the boom was already sowing seeds for the bust - as was seen in the hangover post-Corona.

(Clarksons’ East-West Container Freight Price Index)

Three Catalysts

The definition of super-natural profits implies that the situation is temporary. But markets often extrapolate things far longer than they should and that breeds opportunity. The case for shipping was no different.

The Suez Canal Reopening

In October 2023, after Hamas attacked Israel, the Houthis blocked the Suez Canal in solidarity. The canal handles ~22% of global container traffic, and shipping companies, backed by insurers, refused to risk taking the route.

The closure hit container lines hardest, since they rely heavily on Asia–Europe trade. Tankers and other segments were less affected.

The Trump administration had made reopening the canal a priority, this was partly for regional stabilization but also because Trump wanted to lower inflation by any means that didn’t involve raising interest rates, mainly through supply-side intervention.

This included subsidies for energy production and repeated complaints of the Panama Canal setting fees that were too high, with the consumer footing the bill. Panama ultimately relented and lowered its fees.

Considering he had taken forceful aim at one canal, it wasn’t a stretch that he would soon turn his attention to the other one. On Dec 21, 2024, the U.S. launched airstrikes on Yemen’s capital, their first direct intervention. At the same time, geopolitical pressure was building:

• Israel was shifting military resources towards other fronts after a wind-down in the kinetic stage of the Gaza war

• A Lebanon peace deal freed up air force capacity.

• Iran, Yemen’s backer, was weakened by setbacks in Gaza, Lebanon, and Syria.

I figured these factors would push the canal to reopen by Q2 2025.

How much did the closure matter? I calculated it was equivalent to 6% less global supply:

1. 22% of container trade used the canal

2. Rerouting added ~29% to voyage distance (Clarksons data)

3. That’s roughly a 6% “artificial demand” boost (or supply cut, tomato tomato)

   1. ( (22%*(1+29%) – 22% = 6% uplift)

In any commodity market, a 6% supply cut is huge - imagine 6% of oil supplies disappearing overnight.

As with all exogenous shocks, the question was when, not if, this would end. Sell-side analysts were assuming closure into 2026 while my view was that there would be an opening by mid-2025.

Sure, in the 70s it stayed closed for eight years, but this wasn’t Egypt, it was a rebel faction in Yemen, a country that the West would probably feel more comfortable intervening in.

Chinese Exports Slow

China’s real estate deflation made 2008 look tame. The CCP’s answer: export its way out. Even at the cost of overcapacity and price dumping, they kept factories running by any means necessary.

Evidence was everywhere:

• China overtook Japan as the world’s top car exporter

• Steel and other heavy industries hit records

• Shanghai port processed 50 million TEU in 2024, the most in the world

But this wasn’t sustainable.

Either China stabilized and normalized exports, or the downturn deepened, forcing a pullback. On the other side, trading partners were pushing back:

• U.S., Mexico, Brazil put tariffs on Chinese cars.

• The EU hardened rhetoric on overcapacity.

China grew exports 6% in 2024 to $3.6 trillion, adding $200 billion in a slowing global economy (Germany lurching into a recession by end of 2024). Repeating that looked unlikely.

(My containership supply-demand model, assuming Suez would open up in Q2 2025)

The Tariff Threat

Container shipping has benefited from a positive exogenous shock (Suez closure), but negative shocks were possible too. A trade war is about as direct a hit to shipping demand as you can get.

Trump campaigned on steep tariffs for China, even floating a 60% blanket tariff. Post-election, he was quiet on the topic, which made it a wild card.

I didn’t want to discount this fully as he has a tendency to surprise, so I kept this as a ‘bonus’ catalyst - if it hits it’ll just juice the trade that much more. Besides, consensus wasn’t modelling any tariff impact at all (Oh how things change in half a year).

As a rough estimate, I assumed Trump’s threatened 60% blanket tariff on China would reduce demand by 1.6%. The calculation:

• US makes up 20% of container trade

• The 2019 15% tariffs led to a -2% drop in US imports

• I assume a x4 higher tariff will lead to a 4x higher drop (-8%)

• 20%*-8% = -1.6%

Obviously, these things don’t work linearly - but this was a bonus scenario that could have gone so many different ways, it was better to just understand directionally what could happen.

Downside Protection

Looking at the the catalysts above, I believed we had three ways this could play out: My base case of the Suez opening up in Q2 2025, the Suez opening up and a 60% tariff on China, or we remain the same:

And this is where in my mind, this went from a good short to a great short.

Even if I was wrong on all three catalysts, container rates were likely to fall anyway because of supply.

Unlike demand, supply is predictable. Ships take 4–6 years to build, and orders placed in boom times arrive regardless of current market conditions. After the 2020 COVID boom, companies flush with cash had over-ordered.

By 2024, new capacity was hitting the water:

• Global fleet growth was ~10% YoY

• Deliveries were set to stay high for two more years

(Source: Clarksons)

Without the Houthi disruption, we should have seen a wave of bankruptcies. Maersk even warned as much in its Q3 2023 call: “We could see a pretty dire situation in 2024.”

That’s the beauty of this trade - even if my forecasts were off, oversupply alone would loosen the market, leading to lower rates and reduced revenues.

China broke all export records and it still wasn’t close to keeping up with the new supply that hit the shipping fleet (6% vs 10% yoy).

All together, the probabilities looked in my favor and I made this short a concentrated position.

(Quarterly Breakdown of Supply-Demand)

Epilogue

Writing this today in August 2025, having closed the position in April for a ~35% return, we can see how things ended up developing:

• At first I aimed to short Maersk as the largest player but I couldn’t get short borrow so I moved to ZIM, which is smaller and more exposed to rate swings

• In hindsight, this made the trade much smoother - ZIM’s fundamentals steadily deteriorated while Maersk was able to insulate itself to a degree until Trump’s tariff announcement crushed the stock

• Ironically, the biggest driver was my “bonus” catalyst: tariffs. Trump went harder than anyone expected, and container stocks swung wildly on the news. That was my signal to take profits while the easy money was made

• The Suez Canal didn’t fully reopen as expected. Transits improved, but mostly for smaller vessels

• China continues to try and export its way out of its ills - but the EU and the US are getting more hawkish by the day

• Maersk now sees growth of 2%-4% in demand for 2025, below the expected supply increase of ~6% over the same period, leading to a 10% decline in freight rates year on year

In the end, I was right on two of my calls, but neither was the big one. And that’s the point: I didn’t need perfection. One catalyst landing was enough.

Thanks For Reading!

If you’ve found my work valuable, here are a few ways to support it, any help is greatly appreciated!

• Subscribe to my Substack - just sign up below

• Share feedback - let me know what I missed or could improve

• Spread the word - pass this write-up along to friends or colleagues

• Send me ideas - if I like them, I might write them up (win/win)

This content is for informational and educational purposes only. It is not investment advice or a recommendation to buy or sell any security. You should do your own research and consider your own financial situation before making any investment decisions.