Equifax calls this database the ‘Work Number’. It holds more than 823 million pay records which have been sent in by ~5 million employers. It is used by mortgage lenders, landlords, and state agencies giving out food aid and health benefits. In 2024, The Work Number processed 149 million verifications, nearly 400,000 a day, supporting 6.4 million auto loan applicants, 4.5 million job candidates, and 25.5 million people seeking government benefits. When a platform functions at this scale and delivers critical information to its customers, it stops being a simple data provider and instead becomes the basic functional layer, without which its customers will struggle to run their business.

Equifax is better known for its credit bureau. But the employment database is the most important segment in the company and the core reason why Equifax functions as a tollbooth for the business run by its customers. And the customers are happy to pay the toll, as it accomplishes some very important but basic checks for them in seconds. They pay so little compared to the value they get that they don’t mind.

Equifax has fallen 36% from its 52-week high of $272 to $174. The market is pricing this as a cyclical company in a mortgage drawdown and pricing it as if the information services business is itself under threat from AI. This is a premature conclusion, at least as of now. In the coming years, Equifax’s free cash flow will inflect higher as it completes its technology spend ($3B multi year cloud build). A business that the market believes will be disrupted by AI is entering its best cash-generating years. This is the key dilemma in this story, and we will test it. While researching the stock, we found another risk that we thought was innocuous at first but is material to the thesis.

Welcome to Rebound Capital. We study beaten-down stocks and businesses that have made successful comebacks. Subscribe for free to join our growing community of 25,500+ subscribers and make sure you don’t miss our next briefing.

Subscribe now

How Equifax makes money

The main segments within Equifax are the Workforce Solutions (~43% of revenue), USIS (~34%), and International (~23%).

These 3 segments share one thing in common. These are businesses built on proprietary data and function as the foundational data provider for their customers. USIS and International are bureau businesses, and Workforce Solutions is the leading provider of employment data in the US and is the crown jewel of Equifax’s business. Across the three, FY25 revenue was $6.07B, adjusted EPS $7.65, and free cash flow $1.13B.

Workforce Solutions compounds fastest at more than 50% margins, which is why its share of revenue rises. The last column is from our base-case estimate.

Workforce Solutions: The best business within Equifax

Workforce Solutions made $2.58B in 2025 at an adjusted EBITDA margin above 50%. It is the biggest part of Equifax and by far the most profitable.

The mechanics are simple. Payroll systems send employee pay records to Equifax automatically, every pay cycle. When a lender, landlord or government agency needs to check what someone earns, it queries the database and gets an answer in seconds. Equifax charges a fee for each check.

So why would an employer hand over its payroll file for free? Because answering these requests is a chore it does not want. A large employer fields thousands of them a year, from mortgage lenders, landlords, benefit agencies and background screeners. Each one takes an HR person’s time, and there is legal risk attached if the answer is wrong or goes to the wrong party. Handing the file to Equifax once, automatically, makes the problem disappear. Equifax also runs related services for the same employers, such as managing unemployment claims and filing tax credits. The data is effectively the payment.

That is what makes this a two-sided network. Employers contribute because it saves them work. Verifiers pay because the answer is instant. Neither side has a reason to leave. The alternative for a verifier is to call or fax the HR department and wait two or three days, or use a smaller database that may not be cheaper and may not have complete or current data.

Workforce Solutions has 2 unequal segments. The first is Verification Services (84%), which is ~$2.2B in 2025 and consists of Mortgage, Diversified Markets (card, auto, personal, and direct-to-consumer), and Government. The fastest-growing part of this is government. When someone applies for Medicaid or SNAP, the agency has to verify their income, which is the same query a lender makes. Equifax signed roughly $300M of government contracts and renewals in H1 2026.

The other part is Employer Services at ~16% with ~$0.4B in 2025 revenue. This consists of talent screening, onboarding services, unemployment claims, and tax credits. Employer Services is expected to shrink in 2026 due to Congress not renewing the Work Opportunity Tax Credit (a US federal tax credit for employers who hire from targeted disadvantaged groups), which cost the segment a couple of points of growth.

The Moat: Workforce Solutions is a 2-sided data network. This is the deepest moat among Equifax's 3 divisions.

New competitors cannot offer the same breadth of service as Equifax as it does not have the scale of data. Also, in a 2 sided network like this, the problem faced by new competitors is that they have to onboard data providers and customers at the same time - and both of them have no incentive to shift away from Equifax (which has >823 million records). Equifax has a large number of automated data partnerships and integrations like the Workday partnership. Such partnerships are sticky once put in place.

The moat may be even deeper for a mortgage provider's workflow. Equifax has the requisite compliance and audit trails needed for such work. In fact, Equifax’s systems will be integrated into the loan-origination workflows. So if a lender has to switch, they will need to change their very processes.

Equifax has the strongest pricing power in this division, as can be inferred from the >50% adjusted EBITDA margins. In an antitrust complaint filed in 2024, the allegation was that the per-transaction pricing rose by ~270% between 2012 and 2024 (from $18 to $66). We will not comment on the merits of the case, but this does show pricing power.

USIS: the bureau everyone associates Equifax with

This segment holds the traditional US credit bureau. It made $2.08B in revenue in FY25 with a ~35% adjusted EBITDA margin. This segment provides data on consumers for consumer and commercial lending, along with the credit scores (either its own or 3rd party). Other products for decision-making, identity, and fraud prevention that lenders require are also part of this segment.

This segment also makes money by charging a certain fee for each data pull across mortgage, auto, card, personal, and commercial lending. USIS also resells the FICO score, computing it on its own file under license and remitting a royalty to FICO. We will come back to this mechanism as it explains the reason for lower margins in this segment.

The Moat: The key moat for this segment is regulatory. The US government has mandated that all conforming mortgage lending (loans sold to Fannie Mae and Freddie Mac) be done by pulling data from all 3 credit bureaus: Equifax, Experian, and TransUnion.

The way this system reinforces the moat is:

• It is mandatory to pull data from all 3 bureaus on conforming loans, which makes the 3 companies complements rather than rivals. Worth noting that this is a regulatory choice, not a law of nature. The FHFA proposed dropping to two reports in 2022, shelved it in early 2025, and the current director has kept tri-merge in place. The industry is now pushing further. In December 2025, the Mortgage Bankers Association asked the FHFA to allow a single report for borrowers scoring above 700. We have modeled for a bi-merge scenario in the bear case [model shared at the end].

• To pull bureau data, you have to give your own account data. In this way, a new entrant will not have any data to furnish, and hence lenders will not contribute to it, so building a competing file or data is extremely difficult.

• FCRA (Fair Credit Reporting Act) compliance, ability to handle disputes, and years of regulatory compliance record and relationships are a deep moat.

The key factor to remember here is that since there are 3 players in the tri-merge mortgage file and since this system is maintained by regulatory action, there is little scope for continued pricing increases.

International

This is the credit bureau and data business running internationally. This includes Canada, Brazil via Boa Vista, Argentina, Chile, Europe (UK, Spain, Portugal, Ireland), and Australia, New Zealand, India.

This segment made ~$1.41B in FY25 with 28.5% adj. EBITDA margins. Latin America is the key growth driver here. This segment has the weakest moats of the 3 segments. In some markets, Equifax is in a good position, and in others it's weaker. Equifax is strong in parts of Latin America, behind Experian in the UK and in India, and in a duopoly with TransUnion in Canada. We model this segment to grow at 5%-6% CAGR till 2035.

Our thesis: A mix shift towards Workforce Solutions

Workforce Solutions has the strongest moat amongst Equifax’s segments. We would classify both Workforce Solutions and USIS as basic infrastructure, and hence these 2 segments have the deepest moat.

In our model, we expect the share of revenue from Workforce Solutions to rise to ~46% in 2035 vs. ~43% in 2025. So the company’s most important and defensible division is expected to increase its share of the total revenue. This is a major shift and the core of our thesis on Equifax. This is also the answer to the question we opened with. The mix shift will drive the margins higher, and this will in turn drive the cash flow inflection (coupled with the technology build getting complete).

What went wrong?

Multiple factors came together leading to Equifax’s ~36% drawdown since its 52-week high.

• EFX entered 2025 at north of 30x earnings and was priced as a compounder. This premium multiple was challenged as the business’s positive catalysts got deferred.

• The expected mortgage recovery has been deferred due to multiple issues one after the other. Compounding this was the fact that information services companies have been de-rated by the market due to worries about the effect of AI.

• The Q2 results on 21st July beat on EPS while revenue was in line. But the Q3 guidance was short. This caused the stock to fall ~12% that day, on fears that the mortgage market has not troughed.

• On 2nd October 2025, Equifax fell ~8% after FICO launched its Mortgage Direct License Program. This let the tri-merge resellers license FICO’s algorithm directly and compute the scores themselves on bureau data. This replaced the previous system where each bureau, including Equifax, would give its data along with the computed FICO score (after paying a royalty to FICO) and give this file to the tri-merge resellers.

  The data still comes from Equifax and the bureaus, but this was a direct hit. The revenue that disappears is the difference between FICO’s royalty and the price at which the file was being sold. Analysts put the earnings hit to bureaus at 10%-15%. It is a real attack on the USIS revenue line. But the defense is that the bureaus can still reprice their data.

The key reasons for the drawdown are driven mostly by sentiment, plus slowing revenue growth for Equifax. The business continues to perform, though, with Q2 revenue growth of 7% YoY (ex FICO royalty dynamics) and EPS growth of 13% YoY. FY25 free cash flow generation was $1.13B. The multiple derating drove the price decline. This is an interesting setup for us as we think the business still has decent runway to grow in the coming years.

The Bureaus and FICO: From partners to competitors

The credit bureau business inside Equifax gets the most attention but matters less than the Workforce Solutions segment. It is worth understanding anyway, because it explains why the reported margins look worse than the business is.

The above chart shows how the bureaus and FICO interact with each other. When a financial institution orders a tri-merge report, it does so from a reseller. This reseller is a company that buys the bureau data from Equifax, Experian, and TransUnion (they have to buy all 3). Each bureau has already computed a FICO score on its data for the customer and then handed the file to the reseller (with their markup). The bureaus directly pay the FICO royalty.

The History: For decades, Classic FICO was the only score Fannie Mae and Freddie Mac were allowed to accept on a loan. This was an unusual situation where the regulator had handed a monopoly to a private company (FICO). Congress tried to end it in 2018, but for ~7 years nothing happened.

This changed when FICO raised prices by ~16x, from $0.60 per score in 2018 to ~$10 in 2026. This led to political pressure to allow a second score. The result was that VantageScore, jointly owned by the 3 bureaus, was allowed to be accepted by both Fannie and Freddie.

Why does it matter to our thesis: The first is margin optics. Equifax books the full FICO fee as revenue, then pays it straight to FICO. When FICO raises the royalty, the revenue goes up while the profit does not. This leads to a lower margin. This is just optics. That is why USIS adjusted EBITDA margins fell to 32.8% from 35.0% YoY in the latest quarter. Nothing about the business changed.

The second is that the bureaus own VantageScore. Every FICO score replaced by a VantageScore removes a royalty payment. Equifax’s 2026 guidance assumes 100% FICO scores, and the CEO said that as mortgage customers convert to the lower-priced Vantage scores, the company expects significant margin expansion. Revenue down, margin up. So both the USIS revenue and margin growth rates have nuances in them that an investor must be conscious of.

VantageScore adoption: VantageScore is already widely used outside mortgage, with ~42 billion scores pulled in 2024. Adoption in mortgage has been slow, and will stay slow. Classic FICO is wired into MBS pricing and investor risk models, and investors price VantageScore-backed paper as higher risk. Management has sized the profit opportunity at $100M-$200M.

From partners to competitors: For 30 years, this was a partnership. FICO built the score, the bureaus held the data, and both coexisted. Then FICO wanted more: it increased royalties by sixteenfold and created a direct-licensing program that cut out the bureaus' markup entirely. The bureaus answered with a rival score. Neither side can walk away. FICO needs the data from the bureaus and the bureaus’ own score is not as trusted by the market as the FICO score. The bureaus have a slight upper hand in the medium term (if VS adoption picks up).

The Compounding Opportunity

A few things compound here. The revenue mix moves toward the highest-margin segment, which lifts consolidated margins. And with the technology build finished, those margins now generate free cash instead of funding capex.

We expect EFX’s free cash flow to inflect in the coming years. Free cash flow was $866m in 2021, collapsed to $133m in 2022 at the peak of the cloud build, then recovered to $516m, $813m and $1.13bn in 2025. These are reported figures. In the model, we deduct stock compensation, which takes 2025 to $1.06bn. Capital intensity should now normalize to ~7% of revenue. This heavy spending on technology was the key reason Equifax was not generating cash in the last few years.

The mix shift described above does most of this work. Consolidated EBITDA margins will move towards ~38% from the current ~32%. We expect that EFX can compound revenue and FCF at ~7% and ~10% CAGR in the coming decade.

Capital returns will be another important driver of returns. In H1’26, the company purchased $560M of shares, with roughly $1.5B of authorization remaining. EFX is currently at ~25x EV/FCF, which means they can keep reducing up to ~3% of shares every year. At current low prices, this is a source of marginal buying and adds ~3% CAGR to our expected return.

Mortgage Recovery as a coiled spring: Mortgage originations were $4.4T in 2021, $1.6T in 2023, and expected to be ~$2.2T in 2026 (MBA forecast). The company is forecasting $2T. The key reason for this lower level of originations is that interest rates have repriced materially in the last 5 years. Roughly 75% of loans are priced under 6% so moving means repricing the largest expense for a household. Another reason is the historic unaffordability of housing. We expect that this lull in mortgage originations cannot go on indefinitely. There are roughly 860,000 new households formed a year on Harvard's projections, and sooner or later they need somewhere to live. When the market recovers, we expect the USIS division to benefit directly.

In our base case, we are not embedding a mortgage recovery. It is an optionality.

Does AI break the business? Can startups disrupt it?

The market fears that AI makes information businesses less valuable. This may not play out. A model can think about things, but it cannot manufacture a payroll record an employee did not send. The barrier here is permission and access. It is not intelligence. Also, with the advent of LLMs, companies with proprietary data are well positioned to monetize it in multiple ways. We expect Equifax to increase the share of revenue from new products and services. The company tracks this metric as the Vitality Index. It is currently at 16% (so 16% of current revenue is from newer products). This is higher than the ~10% goal set by the company. This metric is important as a company innovating for its customers and having critical proprietary data is unlikely to be disrupted.

Another fear is that as AI drives white-collar job losses, what happens to the usage of the Work Number? This is again not a direct threat, as the Work Number is used for lending decisions and benefit eligibility and not for hiring. The hiring-sensitive services are talent screening and onboarding, which make up ~16% of the segment’s revenue. So the risk is capped here as well. Also, unemployment claims can be expected to rise with rising layoffs. Government verification is the bigger offset. Job losses mean more benefit applications, and each one needs an income check.

In any case, this assumes AI will lead to job losses, which hasn’t played out yet.

Disruption threat from startups

Argyle, Truv, and Truework attack verification by asking the borrower to log into their own payroll account. Argyle markets itself at up to 80% below a legacy report, and Argyle and Truv are now GSE (Fannie & Freddie) accredited. These startups have been around for more than 5-6 years and, to date, have not been able to dent Equifax much and are still subscale. The reason is structural. Equifax already has the data, so a lender types in a name and gets an answer instantly. The startups have to wait for the borrower to log in, and many never do. So lenders use both: Equifax first, and a startup only for the applications Equifax cannot answer. That leaves the challengers with whatever is left over. Industry estimates put Equifax's instant hit rate on mortgage files at 60% to 70%. EFX also carries FCRA certification and an audit trail a lender can defend in an examination. The startups win where Equifax cannot answer: gig workers, thin files, and price-sensitive buyers like benefit agencies and tenant screeners. The competitor that should worry a shareholder is Experian, which collects payroll data the same way Equifax does and has the balance sheet to keep going.

So neither AI nor the startups break this business. There is another less cited threat that we are tracking.

Legal and Regulatory Threat

Equifax is facing a regulatory threat on its strong access to payroll data from ADP, Paychex, and other payroll processors. The way these multi-year deals and access are structured makes it more expensive to get this data from anywhere else. That is why the startups have to ask customers to log into their payroll database. The threat to this moat is coming from courts. 2 mortgage brokerages sued Equifax in 2024. Their argument is that exclusive deals give it control of >40% of the payroll data anyone needs to verify income electronically, and that it used that position to raise the price of a verification from about $18 in 2012 to about $66 in 2024. Equifax asked the judge to dismiss the case and was refused in February 2025. It then tried to move the dispute out of court and into private arbitration, and was refused again in February 2026, with the court holding that Equifax had waited too long and given up that right.

The remedy was already imposed in December 2025 in Australia. Equifax entered into a binding agreement with the regulator not to sign new agreements that would block rivals from accessing payroll data. This is the obvious thing a US court would ask for. This is not the first time. The FTC put a ten-year consent order on this business in 2008, limiting its acquisitions of verification competitors. That order expired in 2018, and since 2021 Equifax has completed 14 acquisitions, a third of them in Workforce Solutions.

Separately, senators opened an investigation in Feb’26 into Equifax profiting from Medicaid and SNAP work requirements, which is the fastest-growing part of Workforce Solutions. So the threat is legal and regulatory. Losing exclusivity would not kill The Work Number, as 5 million employers and the largest set of records do not disappear, but it turns a monopoly into a scale advantage, which changes what the business is worth long term.

We modeled this risk. The exclusivity remedy and bi-merge together only cost ~11% of revenue by 2035, but we expect a much larger effect due to the lower margins Equifax would be able to charge if the Work Number is disrupted. That combination is what produces our bear case.

Valuation & Rebound Capital Rating

Read more

Subscribe now

July was an interesting month, with a sharp pullback in the Semis/AI trade and the poster child of the AI trade, Situational Awareness, getting margin called and its public portfolio suffering heavy losses (and being sold to a hedge fund) due to high leverage. We had derisked our portfolio for such a pullback over the last month. We trimmed our holdings in semiconductor stocks as prices rose over the last 2 months.

We think the AI trade is still strong. We previously wrote about playing the AI trend through cloud service providers. The market is coming around to this thesis as well, demonstrated by the strong post-earnings reaction to Amazon and Microsoft earnings.

Rebound Portfolio Performance

The Rebound portfolio is now beating the market (S&P 500) by ~5.8% since inception in Aug’2025. Below is our portfolio snapshot from RC’s Schwab account.

On a year-to-date basis, the Rebound portfolio is beating the market (S&P 500) by ~3.2%. We are quite satisfied with our returns in 2026, given how polarized the markets have been - with only AI/Semi names doing well. This is a strong performance given our preference for value picks. In July, the key drivers for the RC portfolio’s performance were the strong earnings from Amazon and Microsoft. Names like Constellation Software and Mercado Libre were also major drivers of the outperformance. Year to date, Semis were a large driver of the performance.

Portfolio Holdings Update

My target for the RC portfolio is to achieve a ~15% CAGR over the long term. At that rate, the portfolio doubles roughly every 5 years. This is a reasonable goal to reach without taking on significant risk.

I am focused on compounding my wealth and treat the S&P 500’s return as a benchmark to beat, but I don’t expect my strategy will beat it in all conditions. Especially as the market is driven by momentum and trends in many quarters.

Read more

In this quick note, I will lay out how we see the current drawdown in Semis and whether there are any interesting rebound opportunities yet.

What is causing this drawdown

This drawdown is not due to a lack of fundamentals, as the semis are, on aggregate, delivering >100% earnings growth in Q2 as per FactSet. As per the SIA, the global semiconductor sales were ~$110B in April (up 11% sequentially) and ~$120B in May (up 9% sequentially).

The AI buildout is now being funded by Debt

The biggest change in the last few months or the last 2 quarters has been how the AI buildup is being paid for. The 4 Hyperscalers have, to date, guided to ~$725B in capex in 2026. This is up ~75% from the capex spent in 2025. For the first time in the history of the hyperscalers, they are funding the data center buildout using debt. Another key factor is that capex is growing faster than revenue growth, raising concerns about the sustainability of capex growth. The numbers are staggering:

• Amazon, Meta, Nvidia, Google, Oracle, and SpaceX have issued ~$182B in investment-grade bonds in 2026. This is roughly 15% of all US corporate debt issuance.

• Alphabet just reported its first free cash flow negative quarter ever in its listed history since 2004. We estimate that all of the other large capex spenders except Microsoft will also be FCF negative in 2026 and are most likely to be FCF negative in 2027 as well. The largest companies in America have gone from buying back shares to issuing shares (Google) and issuing debt to fund their capex plans. This is a massive technical shift in the market.

• The Credit Default Swaps for Oracle and other large companies like Nvidia are at ~7 year highs. These are like insurance contracts to protect against the risk of default of these companies’ debt. Higher prices of these contracts show that the debt market is ready to fund the AI capex (Nvidia’s bond offering was oversubscribed by ~4x) but is looking to hedge its risks as well.

Uncertain AI ROI

The market worries that the capex spending done by the hyperscalers may not earn a good ROI due to:

• Competition might make the LLM models a commodity. This would hurt the revenue growth expected at OpenAI and Anthropic. These 2 companies are responsible for a large part of the AI backlog at the hyperscalers.

• The hyperscalers currently depreciate their GPUs over a 5-6 year horizon. Many industry analysts say that this is too long and the real depreciation life cycle is only 2-3 years (as GPUs have a yearly innovation cadence).

Circular financing fears

We have written previously about circular financing in the AI industry. A simple example is Amazon investing in Anthropic and then, in turn, Anthropic uses Amazon’s AWS cloud service. This is also called vendor financing.

There is another version of this circular financing. Nvidia invests in a multitude of AI companies, like the Neo-clouds and various other semiconductor bottlenecks (that word got very famous). Now, as a bear, you can say that these companies are being propped up. Nvidia and Jensen say that they are supporting innovation. And the truth lies somewhere in between.

Yesterday, news broke that Nvidia is in talks to backstop up to $250B in data center leases for OpenAI. This is vendor financing at a sovereign level. And the number is too big to be ignored. The problem with all of these deals is that any hiccup in revenue growth for the frontier LLM labs, like higher-than-expected competition from China, or a pause in revenue growth as companies look to extract maximum value from their tokens, or a moratorium by the politicians, will lead to a domino effect (negatively).

Semiconductor stocks were not priced for any of this. The recent weakness is the market beginning to price these risks in, amplified by profit booking and deleveraging.

Potential peak in the memory cycle

I am not a memory bear. But the simple observation is that the rate of change of the growth rate in DRAM pricing has been coming down.

From 10-15% in Q3 2025, to 18-23% in Q4, 93-98% in Q1 2026, 58-63% in Q2, and to 13-18% guided for Q3 (TrendForce). Even though prices are still rising, the rate of change in price increases has slowed. The market is worried that this will be negative soon.

Even though memory will have much higher margins than historically due to its long-term agreements and the AI buildout, that is not enough to prevent the kind of drawdown we have seen in memory stocks.

The market is trading the memory stocks on the rate of change of memory prices. And in the coming quarters, we will get closer and closer to the expected 2028 increase in memory capacity, where demand and supply will match a little more evenly. As the market starts taking that into consideration, these stocks will trade at seemingly very cheap multiples. But they may not be cheap.

The thing that could restart this trade is another explosive use case for AI. Like AI coding led to an explosion in the use of tokens at the start of 2026.

A look back into the past: In 1995-1996, there were 50 fab plans announced by the memory players. This is quite similar to how fabs are being planned by memory players (including CXMT in China). DRAM prices fell for 2 consecutive years in 1996-1997 while the PC and internet industry and use cases kept scaling. Secular demand may not protect memory prices (this would be the first time) as supply is ramped up.

Our take on a few stocks

In this section, we will put out our research on a few stocks.

Read more

I will periodically update the portfolio as the earnings season progresses. I will keep these very short and focused on explaining the incremental data points that cause me to make changes to my portfolio.

Read more

We initiated our position in Sep’25 at ~$245. This update asks 3 key questions:

• Is Google falling behind in AI? What are the implications for the Search and GCP business?

• How big can the cloud business get?

• Is the risk-reward attractive at the current levels?

We have also rebuilt the valuation model from scratch.

The bear case is credible

Read more

That’s how the story of a potential rebound stock starts. What we do at Rebound Capital is try to understand the second half of the story. Will the stock rebound or keep dropping more? Today UNH is back near its 52-week high, up more than 70% from that low (We bought UNH at $310 when its stock price was cratering due to excessive pessimism and sold it at ~$425 recently).

Here are the top 6 stocks we are researching as rebound candidates for July 2026.

6. AeroVironment ($AVAV)

AeroVironment makes drones and counter-drone systems. Switchblade is their most famous drone. It’s a tube a soldier carries on his back that can be launched to deliver overhead streaming video until a target is selected, at which point it flies into the target and detonates. Ukraine has used a lot of them. Puma is a scout drone used to see over the next ridge. Titan knocks enemy drones down. That is ~70% of the revenue. The other 30% is satellites, lasers, and cyber and electronic warfare work for the Pentagon and the spy agencies. Fiscal 2026 revenue was ~$2 billion, and fiscal 2027 guidance calls for roughly 10% growth.

The Moat

The military cannot just swap drones quickly. The system has to be certified, thousands of soldiers have to be trained on it, and allied forces build their tactics and spare parts around it. Ripping out a proven drone to save money on a cheaper one takes years, and nobody wants to be the officer who signed off on it. The moat is the fact that it is already out there and everyone knows how to use it.

What went wrong

• Growth is slowing, and the stock is still expensive: Last year, revenue more than doubled, but most of that was due to the takeover. This year, the company expects 10% overall growth. At $145, you pay 45 times adjusted earnings, which is not cheap.

• The BlueHalo deal went bad: Last May, AeroVironment paid $4.1 billion in stock for BlueHalo. Its biggest contract was building the antennas that talk to America’s military satellites. In January, the government told it to stop work. In March, it canceled the contract. The company had to take a ~$151 million writedown due to this.

• Investor day made it worse: Management promised $3.5 to $4 billion in revenue by 2030, which isn’t exactly a high growth rate. This year, growth slows to 10%, and margins are flat.

Our Take

We always look for a margin of safety, and revenue stability is a big input into that. Here we see very little. The company has had to restate its own numbers. Growth slows to 10% this year, and the stock still trades at 45 times adjusted earnings.

The Army just signed a contract worth up to $500 million for Titan. There is a large opportunity or TAM available for $AVAV. The thesis is interesting, but we will wait and watch for now.

We are studying the drone and missile defense space to identify opportunities. Let us know any interesting stocks you think we should focus on.

Leave a comment

5. Coinbase ($COIN)

Coinbase is a platform where retail and institutional investors buy crypto and, more importantly, store it. Holding your own coins means guarding a password that, if lost or stolen, will cost you access to your own money, and institutions will not run that risk. So they pay someone to hold it for them. When BlackRock launched the world's largest bitcoin ETF, it did not hold the coins itself. It hired Coinbase. This custody business builds institutional trust and is a foot in the door for Coinbase.

The Moat

Coinbase is not the cheapest place to buy crypto (its brokerage is higher than most). But it is the one that has never misplaced crypto. There are hundreds of crypto exchanges, but none with the trust and scale of Coinbase. What nobody builds quickly is more than a decade of clean custody in a business where the competition has been unable to prevent fraud against its customers. There are so many instances of crypto wallets being hacked or of customers losing their crypto due to fraud.

Look at what that lets Coinbase charge. Buying bitcoin on the Coinbase app costs between 0.5% and 1.49%. The same bitcoin on Binance costs 0.10%. People, and especially institutions, pay it anyway because they trust Coinbase and not any other platform.

What went wrong

The stock is down about ~60% from its 2025 high. Bitcoin itself has nearly halved from its October peak. The last crypto crash was caused by crypto lenders blowing up, FTX turning out to be a fraud, etc. - basically the industry’s own missteps. This time, the fear is that crypto has not performed or held its value under the conditions for which it was designed: global uncertainties, potentially higher inflation due to out-of-control US budget deficits, and regulatory approval for crypto and blockchain use cases.

• Crypto crash and drop in volume: Volume fell to $202 billion in the March quarter, from $401 billion a year ago. Coinbase gets a commission on every trade, so fees dropped to $756 million. Retail was down 23% QoQ and institutional was down 27% QoQ.

• Lower subscription revenue: Subscriptions and services are the stickier part of the business and less cyclical than trading revenue. In the March quarter of 2025, it brought in $698 million, or 35% of what Coinbase earned. In Mar’26, subscription and services revenue was $584 million.

  Subscription and services revenue is primarily composed of stablecoin income and staking income. Stablecoins are digital dollars backed by real cash and Treasury bills, which earn interest. Staking is a savings account for crypto where you leave your coins and the network pays you interest in more coins.

Our Take

Coinbase’s business model is derisked to some extent from the volatility of its underlying assets. Coinbase takes a cut on every trade, interest on the cash parked with it, and a fee for holding the coins. It is a tollbooth on institutional usage of crypto and stablecoins.

The question we are working on is how big can the stablecoin business get? Digital dollars are doing the useful work of moving money across borders in seconds for a fraction of the cost of current cross-border payment systems. $19 billion of USDC already sits at Coinbase, more than a quarter of the total (~$75 billion), and Coinbase keeps all the interest on it in its wallets and half of the rest. The whole stablecoin market is about $300 billion. If governments let banks and companies settle real business in digital dollars, Coinbase will own the regulated pipe it runs through. It’s not exactly cheap, but it has a large opportunity ahead of it.

Share

4. Intercontinental Exchange ($ICE)

ICE owns the New York Stock Exchange. It owns ICE Futures Europe, home of Brent crude, the contract that prices most of the world's oil. It owns ICE Futures U.S., where coffee, cocoa, and sugar are priced. And it owns the clearinghouses that back these exchanges.

The company makes money three ways. First is its exchange business. Someone buys a share on the NYSE, an airline locks in jet fuel, a utility buys gas for winter, and ICE takes a slice. $5.4 billion last year, about 55% of the total.

Second, subscriptions for the data generated by those markets. Banks and asset managers pay every month whether they trade or not. $2.4 billion, about 24%.

Third, the plumbing behind the American mortgage. The application, the documents, the closing all move through ICE’s software, and lenders pay per loan. $2.1 billion, about 21%.

That is $9.9 billion in revenue in 2025, up 7%, and $4.9 billion in operating profit. Twentieth record year in a row, and the company is worth about $77 billion today.

The last quarter was the best in its history. Revenue was $3.0 billion, up 20% YoY. The exchanges grew by 30%, with energy trading up 46%.

The Moat

The exchanges: Traders go where the other traders are. A crowded market gives you a better price and lets you get out when you want to. That is why exchanges end up as near-monopolies and why a rival can’t just undercut on fees. Classic network effects.

The data: Brent is the world’s oil benchmark because it's where the most oil is traded. A rival cannot license or copy its way to a benchmark. It has to build a market first. And once a bank has wired an ICE feed into its risk systems and written it into thousands of contracts, ripping it out is hard.

The mortgage software: This is the weakest of the three moats. A lender’s loan officers, compliance team, and every downstream partner all work inside the same system. Switching means retraining staff, rewiring integrations, and disrupting business. Nobody wants to make this switch, which is why the business has kept growing even as the housing market has slowed.

What went wrong

The business has continued to grow even as the stock has derated.

• The starting price: At $189 last summer, ICE traded at 31 times earnings. That is a steep price for a company that grows revenue at 7%. It now trades at 19 times, which is fair for a company growing revenue in the mid single digits.

• Money left quality compounders and went to AI: Investors have sold quality, dependable, cash-generating names to buy into the AI trade.

• A regulatory action: On May 29 the CFTC approved America's first perpetual futures contract, a bitcoin perp, on the prediction market Kalshi. A perpetual never expires, so all its liquidity is in a single order book rather than being split across September, October, and December months. That is how perps beat regular futures in crypto: not on price but on convenience of not having to roll over contracts. The market is worried that the old exchanges will lose business as perpetual contracts gain traction. But there has been no approval for a perpetual oil contract (not yet, though Kalshi is pushing for exactly that), and an airline hedging fuel in Q4 wants a contract that expires in Q4.

• The mortgage software got caught in the software sell-off: Housing is frozen, so fewer loans are being written, and ICE gets paid per loan. That segment grew 4% last year. Meanwhile, the market is selling every vertical software business on the theory that AI makes disruption cheaper.

• Receding energy tailwinds: Energy revenue jumped 46% as everyone panicked about oil and needed to hedge. The markets are now pricing in lower revenue from these exchanges, as the war recedes.

Our Take

The market has derated ICE due to fears that trading volumes will decline. But half the revenue comes from data, indices, and mortgage software. The business has grown for 20 straight years. The trading half is fine anyway. Open interest, the number of contracts customers keep open in ICE's markets, was up 20% in June. Companies list on the NYSE for the same reason traders trade there. Everyone else is already there.

The stock has fallen from 31 times earnings to 19. At $130, it trades at 16 to 17 times next year’s free cash flow, offers a 6% cash yield (about $6 of cash per $100 of stock), and we expect that cash to grow around 9% per year through 2030. Six percent yield plus nine percent growth is a mid-teens annual return, assuming the multiple stays the same.

Welcome to Rebound Capital. We study beaten-down stocks and businesses that have made successful comebacks. Subscribe for free to join our growing community of 23,500+ subscribers and make sure you don’t miss our next briefing.

Subscribe now

Now, on to our top 3 stocks: one each from the power, SaaS, and data provider sectors.

Read more

We have now covered 30+ companies at Rebound Capital. Our screener is as simple as it gets:

• It should be a high quality business

• It should be in a drawdown

With markets at all-time highs, the task is harder now.

So I thought I’d just ask you.

What’s a stock you’ve been watching that’s taken a beating lately? Something solid where you think the market is overreacting.

Hit reply with the name. Even just a ticker and one line is fine. If you have a thesis or a catalyst in mind, even better.

I read every reply. The most interesting name gets the next deep dive. I’m picking it this Friday and you can expect the full deep dive next week.

And even if your pick doesn’t win, you still get something. I’ll share back the list of most common names, so you can see what everyone else is watching :)

Thanks,

RC Team

P.S. If this landed in Promotions or Spam, drag it to your Primary inbox so you don't miss the coming deep dive.

We are publishing this Amazon update without a paywall

Quick recap before we begin. In November, we bought Amazon at ~$236 with one simple claim: AWS isn’t losing the AI race. It was out of capacity. Growth would re-accelerate the moment new data centers came online.

In December, we doubled down on our thesis and made Amazon our largest position.

Two quarters later, AWS grew 28% YoY last quarter (its fastest pace in 15 quarters). The backlog has nearly doubled, and Amazon has built one of the largest custom-chip businesses in the world (Graviton, Trainium, and Nitro), worth ~$50B annually at external prices.

And the stock? Back to ~$245. Right around our entry. When we are right about the business and the stock goes nowhere, it’s usually a good time to re-check our thesis.

A year ago, the fear was that Amazon was losing cloud market share due to weak AI offerings. Today, the fear is that Amazon is doing well in the AI data center market, but this business itself is not attractive from an ROI perspective (Amazon has committed to spend up to $200B in CapEx this year). The market is worried that Amazon and other hyperscalers will trade like utilities in the future due to heavy capex, low pricing power, and little to no free cash flow for investors.

We think the market is wrong for the second time.

Before you proceed, we highly recommend reading our original thesis on Amazon, as well as our December follow-up.

Welcome to Rebound Capital. We study beaten-down stocks and businesses that have made successful comebacks. Subscribe for free to join our growing community of 23,100+ subscribers and make sure you don’t miss our next briefing.

Subscribe now

We got the AWS thesis right

When we laid out this thesis in November, the overhang was specific: AWS was well regarded but widely seen as losing the AI race, ceding cloud share to Azure and GCP without a clear lead in foundation models or silicon. We took the other side, arguing that AWS, together with three underappreciated high-margin engines (Advertising, the 3P marketplace, and Prime), would drive the economics. Two quarters later, all four have accelerated.

• AWS: grew 28% last quarter, its fastest in 15 quarters, vindicating our call that ~20% growth was due to supply issues. Backlog has reached $364 billion (before adding a separate $100 billion commitment from Anthropic).

  

• Advertising: revenue crossed $70 billion in the trailing 12 months, still compounding above 20% (up from the >$60 billion we flagged in November), with margins better than most software businesses.

• Silicon: the chips business (Trainium, Graviton, Nitro) has reached a $20 billion run rate, growing in the triple digits.

• Subscription services (including Prime): revenue up 15% to an annualized ~$54 billion, with 15% worldwide unit growth, the fastest since Covid.

With the re-acceleration confirmed, the debate now centers on whether AWS will earn a decent ROI on $200 billion in annual capex. Our answer is yes. We will list 4 structural reasons the market is not pricing in. Let’s dig in.

The New Wall of Worry

On the February call, management guided to ~$200 billion of 2026 capex, roughly $54 billion above what the street had modeled. Shares slid from ~$240 to about $197 in about 10 days! The market feared that the AI buildout is outpacing demand and that years of record spending will yield an uncertain return.

The bears have two arguments why ROI on hyperscaler Capex will be poor:

• Depreciation: hyperscalers assume GPUs have an economic life of 5 to 6 years, but bears argue the true economic life is closer to 2 or 3 years, overstating profits and free cash flow.

• The counterparty question: what if the labs can't pay? The largest names driving the cloud backlog for hyperscalers are loss-making businesses that depend on continuous outside financing to honor multi-year contracts. OpenAI is estimated to have a ~$14 billion loss for 2026 with no free cash flow before the end of the decade; Anthropic is closer to breakeven but still burning capital. A backlog is only as good as its counterparty, and if funding for frontier labs tightens, a portion of that $364 billion (plus the $100 billion-plus Anthropic commitment) would be tied to companies that need to raise capital again to pay it.

We take both points seriously, but neither negates the thesis.

Regarding depreciation, the 2-3-year clock assumes a GPU is worthless once it is off the frontier (as newer GPUs have better compute per dollar). In practice the retirement decision is economic, not functional: an already-paid-for chip has near-zero remaining capital cost, so it stays in service as long as the inference revenue it earns beats its power and rack cost. Chips retired from training cascade into inference and other workloads that don't require the latest hardware, and Amazon's breadth of price-sensitive demand keeps that older silicon utilized rather than stranded, extending useful life and softening the write-down.

We watch the profitability of the LLM labs closely, and they seem to be trending well (with Anthropic rumored to be heading towards profitability), but it does not break the thesis because Amazon is selling infrastructure, not betting on who will succeed. If a lab stumbles, the workloads route (toward cheaper open-weight models that still run on AWS silicon, billed by AWS) rather than evaporate. The bear case requires the demand itself to be fake. It is not. What would genuinely concern us is a broad, simultaneous pullback in enterprise AI spending that neither open-source nor rerouting could absorb, and we see no evidence of that today.

Share this report with a friend

Share

Key changes in AI that the market is overlooking

The rise of open source: orchestrators will keep a larger share of the pie

As open-source models become more capable and the gap to the leading models keeps narrowing, especially for the day-to-day, mundane tasks that fill most enterprise workloads, the LLM labs will lose pricing power. The value will then migrate to whoever orchestrates the workload. This will be the cloud companies, such as AWS, GCP, and Azure.

Think of payment networks. Merchants compete brutally on price. Brands rise and fall every decade. But the network in the middle clips every transaction at the same rate, no matter who wins.

In AI, the models are becoming the merchants. AWS (and other cloud companies) is the network. We are not saying that Anthropic's and OpenAI’s margins will be as low as we see for merchants. But this is a good illustration of how the value will shift.

Since ChatGPT was launched, enterprises pointed every workload at the best model on the leaderboard and ignored the bill. The bills have now arrived. For example, Uber reportedly burned through its 2026 AI budget in four months, then capped employees at $1,500/month for coding tools. Coinbase’s CEO laid out a step-by-step method by which his company reduced its AI spending while increasing its token usage.

The rational response is not to use less AI. It is to route the routine 80% of requests (classification, extraction, summaries, boilerplate code) to cheap open-weight models like DeepSeek, Qwen, Kimi, Llama, Nemotron, and only the genuinely hard 20% escalates to a frontier model. This is already happening and will accelerate.

Here is a good chart showing how much open-source model usage has increased.

Routing reduces the model layer’s pricing power and increases the total TAM for cloud providers as usage explodes:

• As the workload shifts from a frontier model to open-source models, model providers’ margins decline because many models are now ‘good-enough’ to handle most tasks. But the token still runs on someone’s chips and security architecture. That’s how the cloud providers will produce value.

  

  Cheaper tokens don’t shrink usage. They multiply it. This means higher revenue for cloud providers that help enterprises deploy AI. The hyperscalers are investing in forward-deployed engineers for the same as well.

Switching costs are rising for cloud services and lowering for LLMs

As models become easy to swap, the cloud beneath them becomes harder to leave. Bedrock (AWS’s AI models library) hosts 100+ models from ~18 providers, can route prompts to cheaper models when they clear the quality bar, and provides the agent scaffolding (memory, tools, identity, observability) that an enterprise builds once and keeps. The fine-tuning, data, compliance, and billing all sit in the cloud layer, so switching models becomes a dropdown, not a migration. The lab is interchangeable. The cloud owns the customer.

June showed why no enterprise will want to rely on a single LLM lab. A new executive order creates a voluntary framework under which the most advanced models spend up to 30 days with the US government before wider release. Days later, Anthropic’s newest models were suspended worldwide for 19 days under an export-control directive. When model availability can change overnight, driven by a government decision, enterprises will want the option to route their workloads to different models and, ideally, use open-source models within the safe confines of a large cloud provider (which can ensure there is no backdoor data transfer and can host the model).

Slower than hoped expansion of data centers in the US

The key shortage in AI has shifted from chips to power, which favors whoever already has capacity in the ground. Interconnection queues in the markets hosting most of 2026’s announced capacity (Northern Virginia, Phoenix, Dallas) now run four to seven years, and of the ~16 gigawatts slated for the US in 2026, only about 5 are under construction, with ~11 stuck behind transformer, switchgear, and interconnection backlogs.

Amazon has the most capacity, locked in over years via nuclear offtake, its own substations, and ~$200 billion of 2026 capex. Every year the buildout runs slow, Amazon’s already-energized capacity grows more valuable, its pricing firmer, and its $364 billion backlog more defensible.

And even as those backlogs clear, a second barrier is coming up: community opposition. Sightline calls local resistance a material driver of project delays or cancellation. Maine has passed a statewide moratorium, and similar measures are moving in ten-plus states. That makes new sites harder to permit, which only widens Amazon’s edge.

Amazon’s successful internal custom ASICs

In a world where models commoditize, the lowest cost of compute is the moat. Designing your own chip is how one gets to the lowest cost. Trainium3, generally available since December, runs at roughly 4x the performance of Trainium2 and about 50% lower cost than GPU alternatives. The custom-chip portfolio (Trainium, Graviton, Nitro) has crossed a $20 billion run rate, growing triple digits.

The demand is strong: most Bedrock inference runs on Trainium today, Trainium2 is effectively sold out, and Trainium3 is nearly subscribed, anchored by Anthropic’s 5-gigawatt and OpenAI’s ~2-gigawatt commitments. AWS is vertically integrated: every dollar of inference on Amazon’s own chip is margin that Nvidia does not earn. That is the key to the open-source shift. As routine tokens move to good-enough models, whoever owns the cheapest chip will get the most workloads, and increasingly that chip is Amazon’s.

Why is Amazon our leading choice rather than Azure/GCP?

Fair question, since Microsoft and Google also have cloud businesses. The difference is that both compete with the LLM labs, whereas Amazon does not:

• Google wants Gemini to win the race for the best LLM model. It directly competes with Anthropic and OpenAI. This means that the leading labs will not want to give Google too much of their cloud business (unless the compute constraints are too strong).

• Microsoft is not neutral either. It owns roughly 27% of OpenAI and is now building its own MAI models to cut its AI bill, so its economics tilt toward specific models winning rather than staying indifferent across all of them. Microsoft competes directly with Anthropic and OpenAI for productivity software.

• Amazon is the only one of the three without its own frontier model in the race, which is exactly what allows Bedrock to act as a neutral router. It has no model whose success it needs to defend, so it can send each workload to whatever wins on price and quality. (The natural objection is Amazon's stake in the labs, but Amazon now backs both leading names, not one. It led OpenAI's $122 billion March round with a $50 billion commitment, bundled with two gigawatts of Trainium and AWS as the exclusive third-party cloud for OpenAI's enterprise platform. Amazon is not betting on a winner. Amazon is ensuring its cloud platform is used irrespective of who wins.)

To be precise, Amazon does build models. Its Nova family exists, but it is a cost lever, not a must-win. Google and Microsoft need a specific model to win because their economics are tied to Gemini and to winning in productivity software. Amazon does not. Whether Nova wins or loses, the tokens still run on AWS, on Amazon's chips, billed by Amazon.

That neutrality is important as it prevents any conflict of interest for the LLM labs. Amazon’s AI business rides on compute infrastructure, not model IP, so it does not need to back the right lab:

• If the frontier labs keep winning, the tokens run on AWS.

• If open source wins, the tokens still run on AWS (AWS will prefer this scenario).

• If any single lab stumbles, Amazon is selling the cloud beneath all of them rather than being caught out on one bet.

Key Risks

The risk is simple: we are wrong about demand and the margins AWS will earn. Amazon will spend ~$200 billion this year on data centers, chips, and power. This spend will not deliver good returns if AWS’s operating margins reduce from current levels. If the demand fails to materialize (we are confident that this won’t happen), then, again, the CapEx would have been ill-timed and would need time to digest.

The following key risks materializing would change our view:

1. AWS growth back below 25% for two consecutive quarters. In fact, we expect AWS growth to exceed 30% in the coming quarters.

2. AWS’s operating margins reduce in the coming quarters

3. The backlog stalls, or Trainium slips a generation against Nvidia’s roadmap.

4. AWS raises Capex without giving further data to show customer commitments against which it is raising capex. Or if it doesn’t communicate clear ROI on the $200B in spending (at least in theory).

Amazon Valuation

Our base case puts Amazon’s intrinsic value at ~$310, about 25% above its current price. The build runs to 2035, discounted at 8%, and rests on four things:

• Third-party seller services compound in the double digits early and fade toward 6%, growing from ~$172 billion to nearly $400 billion.

• Advertising, now a >$70 billion business still growing north of 20%, roughly doubles over five years as retail media and video keep scaling.

• AWS decelerates from the 28% it just printed toward the high teens by 2030, then to ~10% by 2035. Even fading, it roughly triples this decade and carries about two-thirds of company profit by 2030.

• Operating margin climbs from ~11% to ~17% by 2030 as the mix tilts toward the high-margin engines and automation deepens.

Access our full model here:

Amazon Model

Rebound Capital Rates Amazon a ‘Buy’

Amazon stays our largest position in the Rebound Portfolio. We hold a full position, so we are not adding here, and we will look to trim as the stock approaches our ~$310 base case.

Two honest observations. We hold a full position from ~$236, so this is not a fresh buy. At ~$247, the ~25% upside to fair value meets the margin of safety we require, and we are content to sit and let the business close the gap.

The bull case values Amazon near ~$487, close to a double from here. The bear case puts Amazon at ~$184, a roughly 25% drawdown. In this scenario, the multiple is down modestly and we assume AWS growth and margins disappoint. It does not assume the business breaks. The downside is a dented compounder, not a broken one.

A decade ago, cloud was a sliver of IT budgets. Add inference, agents, and the wholesale re-platforming of enterprise software onto that base, and the runway runs to the trillions. AWS enters that race in the lead, with the installed base, the power, and now the silicon to serve it at a lower cost than anyone else. Open source is commoditizing Amazon’s suppliers, model makers, and merchant chip vendors, not Amazon. It only widens the moat around the company that owns the rails.

Rebound Capital’s work is provided for informational purposes only and should not be construed as legal, business, investment, or tax advice. You should always do your own research.

On the 2nd of July, Microsoft went even bigger: it committed $2.5 billion and 6,000 engineers to embed with customers.

In May, Anthropic and OpenAI had done the same. They raised billions for a near-identical venture. The labs had decided they wanted to be Palantir on top of an LLM lab.

That is a strange compliment, and the most important development for anyone valuing Palantir today. The thing that made it special, the willingness to do the hard, unglamorous work of putting software/AI to work inside a real organization, is being copied at scale by the richest companies in technology.

So, what is a forward-deployed engineer?

A Forward Deployed Engineer (FDE) is a software engineer who embeds directly inside a customer's environment to solve that customer's specific operational problems. They then feed what they learn in the field back into the core product. Unlike a traditional engineer who builds one capability for many customers, an FDE builds many capabilities for a single customer.

Note: Scroll down to the bottom to see our latest portfolio update.

Welcome to Rebound Capital. We study beaten-down stocks and businesses that have made successful comebacks. Subscribe for free to join our growing community of 22,500+ subscribers and make sure you don’t miss our next briefing.

Subscribe now

How Palantir makes money

Palantir was founded in 2003 by a group that included Peter Thiel and Alex Karp and was seeded by In-Q-Tel, the CIA’s venture arm. Its first product, Gotham, was built with intelligence agencies to consolidate scattered, siloed data into a platform that analysts could act on. For most of the next two decades, the company lost money, roughly $6bn in cumulative losses, because the work of installing and customizing the software was expensive and each early contract was small.

Palantir has 4 major products today:

• Gotham remains the operating system for defense and intelligence.

• Foundry is the operating system for enterprise data used by large businesses. This is the key growth engine (as government contracts are lumpy and can be renegotiated on a whim).

• Apollo installs and continuously updates the software across cloud, on-premise, classified, and edge environments.

• AIP (Artificial Intelligence Platform), launched in 2023, is the layer that integrates third-party large language models into Foundry and Gotham, enabling customers to run AI agents on their own governed data.

Tying all four together is what Palantir calls the ontology. It’s a map of your business. It turns a row in a spreadsheet into the actual customer it stands for, links that customer to their orders and shipments, and lets software and AI agents act on those connections instead of just reading numbers. This is the layer Palantir says is its moat.

Palantir sells these platforms on multi-year contracts, with forward-deployed engineers who embed with the customer. It reports revenue in two ways: by customer type (commercial or government) and by geography (US or international). Government is a handful of very large, contract-driven relationships. Commercial is a land and expand model across many smaller accounts, and since 2023 it has been the growth engine.

The picture over 5 years is a company rebalancing. Total revenue nearly tripled, from $1.5bn in 2021 to $4.5bn in 2025. Government, once 58% of revenue, has drifted to 54% because commercial growth has outpaced it. In 2025, commercial revenue grew by 60%, compared with the government’s 53%. The market’s attention now rests on the commercial line. This is the segment where Palantir faces more competition, but it is the larger, higher-margin TAM.

Zoom into US commercial, the heart of the story, and AI’s effect is unmistakable. It grew more than sevenfold from 2021, but the path was not smooth: growth decelerated to just 36% in 2023, the year before AIP, then re-accelerated to well over 100% by 2025 as demo boot camps pulled in new companies. That surge is concentrated in the United States. International commercial revenue grew only about 2% in FY2025, even as international government, the relationship-driven piece, grew 47%.

But Palantir is more like a consulting business than a SaaS. If this were a usual software business, the gap between US and European commercial growth would not be so large. But Palantir’s business is high-touch. Forward-deployed relationships that exist in the US and not in Europe (Palantir’s own chief executive conceded the point, calling the European problem a matter of bandwidth) are a consultancy’s constraint, not a software company’s. This matters, as we will discuss in the following paragraphs (consulting businesses usually receive lower valuations than SaaS).

Why is it down?

Palantir peaked at ~$207 in November 2025 after a sharp rise, then fell roughly 50% to briefly touch $106 in late June, before rebounding on the NVIDIA partnership:

• SaaS derating: The market is rotating out of expensive software amid fears that AI-driven coding will disrupt software moats.

• Increasing competition: as we will argue, many large and prominent firms are coming for Palantir’s market. What Palantir offers, ontology, is now table stakes for all large AI deals.

• Nosebleed valuations: Palantir traded at a >200 NTM PE ratio at its peak. This has now moderated to ~80x NTM PE ratio (GAAP), but is still very expensive.

Competition: the enterprise AI market is getting crowded

Before the AI era, Palantir's commercial segment was not growing as fast as the headline numbers imply. There are a few red flags for us.

In 2021, the company put $405.5 million (per RBC) into nearly two dozen startups that were then going public via SPACs, which promptly signed on as customers. Management conceded the revenue was small, about 2% of its first nine months' sales, but Citi's Tyler Radke found that the headline commercial growth appeared to jump to 37% year over year in Q3 2021, while once you strip out the SPAC revenue it had slipped to ~22%, from ~25% a quarter earlier.

Palantir wound down the program in 2022, reportedly booking ~$300 million in losses as the SPAC shares collapsed.

Palantir’s differentiator was never the software. It was a business-model choice that everyone else found unappealing: send elite engineers to camp inside a customer's organization for months and wire AI into the workflow. The forward deployed model was the moat, and big-tech did not care much for it. It worked for a decade because no one with real scale wanted to copy it. In the last two months, that has changed.

The threat is that these giants can afford to give away deployment work because each earns its money on a different layer, whereas Palantir cannot. Deployment is the product Palantir sells.

Amazon earns through the cloud. Every system its engineers build runs on Amazon’s servers long after they leave, so it can treat the work as a near free way to sell more computing. And its offer beats old-style consulting. It can build the system within the customer’s own cloud in weeks, hand over working software plus staff trained to run it, and charge for a business result and not billable hours.

The labs sell the AI models. Every Palantir AI deployment that runs on a model from OpenAI or Anthropic means Palantir is really just a middleman marking up their work (and making >40% operating margins). They would rather cut out the middleman, do the deployment themselves, and keep the full margin, rather than hand a slice to Palantir. And they have an easy way in; their consulting ventures are backed by private-equity firms whose own companies become the first customers, so the sale arrives as an introduction from the owner.

Microsoft's advantage is distribution. It already has enterprise contracts with nearly every large company, and it wants those companies to spend more on its cloud and AI tools, so running their deployments is a way to pull more of that spend onto its own platform. Its edge is that it meets companies where they are, plugging into the software and data they already use rather than asking them to rebuild everything in a separate proprietary system, as Palantir's platform does.

The founder’s own behavior is a tell

Through 2026, Palantir’s chief executive has spent his airtime attacking the LLM labs. A CEO will only do this if he fears disruption from a particular company.

Recently, he called raw model quality ‘commodity cognition’ and said that companies using frontier models are furious about ‘paying for tokens that create no value’ while handing their edge to the model providers.

The LLM labs own the models inside many Palantir deployments, which makes Palantir a middleman reselling their work. Now they want to handle the deployment themselves. This is a direct swipe at Palantir.

The NVIDIA deal, allowing customers to keep their own model weights in-house, is a defensive move to prevent value from migrating away from Palantir (and NVIDIA!).

Commoditization from below: the warehouses close the gap

Another threat comes from data warehousing companies like Snowflake and Databricks. Palantir’s pitch is that raw warehouse data is inert and needs its ontology to become AI-ready. The warehouses are now building that layer themselves, on open formats, so a company gets a governed, AI-ready data layer inside the platform it already owns, with no separate system to buy and no lock-in.

This is not incremental. At its June 2026 summit, Databricks launched Genie Ontology, and its CEO told the room that AI does not have an intelligence problem; it has a context problem. That is Databricks attacking the semantic layer that Palantir calls its moat. Snowflake is doing the same.

Neither company will field a forward-deployed army, and many Foundry deployments still run on top of them. But they don’t need to. They only have to make the separate operating system look redundant or less important. It only has to make a premium layer look like a nice-to-have. For a bank whose data already sits on one of these platforms, that is the gap between months spent integrating with Foundry and weeks spent switching on the warehouse's own agent tooling.

The government business’s sovereignty ceiling

Management blames flat international commercial revenue (up ~2% in 2025) on ‘bandwidth,’ relationships built at home and not yet abroad. We read it the other way. Europe has looked at Palantir’s offering and, increasingly, said no.

The reason is legal, and hiring will not fix it. As a US-incorporated company, Palantir is subject to the US CLOUD Act, which can compel it to produce data wherever it sits, and no contract overrides US federal law. For a government to route citizen data through Palantir is unacceptable. With the trans-Atlantic relationship changing, a reversal seems implausible.

Swiss agencies rejected Palantir at least nine times over seven years on that exact concern, and a Zurich court in June 2026 threw out 22 of its 23 counterstatement requests against the reporting. Germany’s military excluded it. In the UK, its largest European market, lawmakers want the NHS to drop it, and the BMA, representing 190,000 doctors, has called for a complete break.

In June 2026, France said it would migrate from Palantir to the domestic firm ChapsVision over the coming years, framing the move as a break from 'strategic dependencies.'

The best counterargument

The best counterargument is that hyperscalers have tried and failed to disrupt most SaaS businesses. For example, AWS built CloudWatch to take on Datadog, DocumentDB to take on MongoDB, and Redshift to take on Snowflake. All 3 products thrived despite the competition. So why would Palantir be disrupted?

We would argue that previously the hyperscalers went after SaaS products. This time, they are disrupting the delivery model. On top of that, this time, Palantir is dependent on companies trying to disrupt it. The labs own the models sitting inside AIP, and Microsoft owns the enterprise agreement the deal gets bought through. Palantir also needs cloud services from the hyperscalers to run.

Many companies that beat their clones still trade 50 to 70% below their 2021 peaks. Surviving competition doesn’t protect from multiple contractions. Palantir trades at ~80x NTM PE ratio! Even if Palantir can fend off competition, how can an investor earn a decent return by buying it at such a high valuation?

If deployment is low-margin, how does Palantir achieve software margins?

A recent deal reveals how the market values the forward deployed engineering model. OpenAI reportedly guaranteed investors in its deployment company a 17.5% annual return. You guarantee a return floor for capital-intensive, lower-margin businesses, such as infrastructure or lending, not for high-margin software, which attracts capital for its uncapped upside. This 17.5% guarantee is evidence that this is more of a service-based business model. The smartest capital in the world is pricing AI deployment as low-margin work (not recurring SaaS).

Yet Palantir posts an 88% adjusted gross margin and a 60% adjusted operating margin. The key here is that Palantir counts certain expenses in its R&D, Sales & Marketing and other expense lines - which should ideally be counted as cost of revenue. This difference matters. A service-based business model should not be priced at a premium to the market - least of all at ~80x forward earnings. For reference, Accenture trades at an 11x PE ratio. Palantir is a more capable organization and is growing much faster, but the valuation gap remains too large to ignore. And just as Accenture is being disrupted, Palantir faces the same risk, albeit at a lower intensity.

A few pointers about Palantir’s high margins:

• First, the work becomes cheaper as it is repeated. Palantir spends heavily up front to win and build out an account, then earns years of high-margin add-on revenue on top of what it already built, so that early cost spreads thin over time. GAAP operating margin roughly doubled in a single year - from about 20% to 46% (Q1 2025 to Q1 2026), and that is the effect showing up. The catch is that this only works while Palantir is the one everyone pays for the deployment. Once rivals can do the same work, that add-on revenue will have to be offered at a lower price.

• Cost classification: A share of forward-deployed engineering costs is shown outside the cost of revenue, in R&D and sales and marketing. A services firm would book those consultants as cost of goods. Reclassify it, and the gross margin will fall materially. The market is potentially paying a software multiple on a services gross margin.

• Third, the SBC (stock-based compensation) add-back. The 60% adjusted operating margin is 46% GAAP plus ~14 points of SBC and payroll taxes, a real cost that dilutes owners.

We argue that not only is the market misclassifying Palantir as a SaaS company, but its pricing power will also decline in the coming years.

The 2029 risk to the government business

The government business has the strongest moat. In March 2026, the Pentagon made Maven a formal program of record, a protected line item in the multi-year defense budget. Its contract ceiling had already been lifted to nearly $1.3bn through 2029. Accreditation to operate on classified networks is a multi-year barrier that new commercial entrants will find difficult to overcome.

But government contracts come with their own risks. The Army’s $10bn enterprise agreement and the new $300m award from the USDA are ceilings, not commitments. The government may or may not fulfill them. Palantir’s own filings note that most government contracts are subject to termination-for-convenience clauses, cancellable at the government’s discretion, which is precisely why they are excluded from the backlog.

Palantir’s management is thought to be closely aligned with the current administration, and that alignment can cut both ways. At home, it is paying dividends in awards, ceilings, and access. But the association is unpopular in Europe and adds to existing sovereignty concerns about routing a government’s operational data through a US vendor. Even in the US, Palantir may have difficulty renewing contracts if the administration changes (for example, in 2029).

Program-of-record status protects the defense core across administrations, but a change of control in Washington could slow the pace of new contracts. This risk is being totally ignored by the market.

Valuation & Rebound Capital Rating

Read more

Last month, we derisked the portfolio from the AI theme. June gave us a preview of why that was important. More on that later.

Rebound Portfolio Performance

Happy to report that the Rebound portfolio is now beating the market (S&P 500) since inception (in Aug’25) by ~4%.

Our portfolio experienced a significant drawdown in Feb-March, driven by AI doomerism and the Iran War. Drawdowns are always tough. But we stuck to our core principles, and that discipline is why we are outperforming the market by ~13% since it bottomed on March 30th. Taking advantage of the downturn, we invested in mispriced stocks, with 1 investment ~3x for us (starting at ~8% of the portfolio and trimmed last month).

We pulled ahead in a market where a handful of AI names have been carrying the index, and we did it because we don’t buy cheap for its own sake. We pair value with embedded catalysts - a discounted stock only interests us when the market is also materially underestimating where its earnings and key metrics are headed over the next 6 to 24 months. Cheap can stay cheap forever. The re-rating comes when the market is forced to upgrade its consensus estimates for those material KPIs.

What We Are Focused On

Looking ahead, we are focused on three things:

Calibrating our thesis on Semis: The AI cycle is real, and companies are still early in AI implementation. But the stock prices of most AI and semiconductor stocks are, in our estimate, at the later stages of their bull run. Stock selection is key.

In late June, semiconductor stocks had a sharp, multi-day pullback as investors took profits in the most stretched winners. But Micron's earnings have reinvigorated the speculative rally. This is exactly the ‘any hiccup will lead to a sharp fall’ dynamic we flagged last month.

We had cut our portfolio’s exposure to AMD and Nvidia last month. We are keeping our reduced semi weights and are ready to trim further on any sharp bounce. The semiconductor companies’ revenues will keep compounding for many years - but the stock may not, especially since they are now so stretched in terms of valuation.

Calibrating our thesis on Big Tech: We are getting incrementally more bullish on megacap technology companies. These stocks have been laggards for 1-2 years now due to the large increase in capital expenditure for building AI data centers.

Their money has been spent on buying increasingly expensive semiconductor chips - whether it is compute from Nvidia, Broadcom, and AMD, or memory from Micron and SK Hynix. The semicap companies have benefited immensely as well, rising 3x-4x from just 18 months ago. Going ahead, we see the following happening:

• Big tech will cut costs by designing custom compute (ASICs) and finding more efficient ways to use memory.

• A pause or a second look at capital expenditure on building out AI data centers may be around the corner. Many executives at these companies are paid in stock, and there is no rationale in letting Micron make 80%+ gross margins at the expense of their own stock prices.

• Many companies are now trying to optimize their AI spending (read Coinbase’s CEO explaining how they are doing it) - and this will also be a catalyst for hyperscalers to revisit their capex plans. Of course, if their cloud revenue growth accelerates materially due to insatiable demand, they may not worry too much about capex. Either way, investors should do fine. Cloud providers are also well positioned to help customers optimize their operations. They can offer cheaper models for less-intensive tasks and frontier models only for the most difficult tasks (in an automated manner). This will shift more of the value in the AI stack towards cloud providers and away from frontier labs and semis.

Quality companies are overlooked amid the AI mania: Multiple names with years of strong growth ahead are now trading at their lowest multiples in a long time (including S&P Global, Visa, Moody’s, Netflix, etc.). We added to one of them this month (more on that below), and we will keep digging.

Portfolio Statistics

My target for the RC portfolio is to achieve a ~15% CAGR over the long term. I think this is meaningful compounding, doubling my wealth every 5 years.

I am focused on compounding and growing my wealth and treat the S&P500’s return as a benchmark to beat - but I don’t expect my strategy will beat it in all conditions. Especially as the market is driven by momentum and trends in many quarters.

Rebound Capital’s portfolio compounds faster than the index while trading at essentially the same valuation. To gauge it, we take a weighted average of each holding’s metrics. The headline numbers:

• Revenue growth, next 3 years: ~18%

• EBIT growth, next 3 years: ~19%

• EV/EBIT: ~26x

Here is the interesting part. Our portfolio is set to compound EBIT at around 19% per year, ahead of the low-to-mid-teens consensus for the S&P 500, and we pay no premium on the multiple to do so (S&P 500 currently trades at ~26x EV/EBIT).

Divide valuation by growth, and the portfolio screens materially cheaper than the index. In short, the RC portfolio compounds faster than the market while costing roughly the same per unit of growth.

Broader Market Commentary

AI remains the market’s driver

The demand for AI is strong. Micron’s late June print was a clear beat and raise, with revenue stepping up enormously year-over-year - a reminder that the AI infrastructure trade is backed by real spend, not just sentiment. We are not bearish on the buildout - we own semis. But we are concerned about the valuations.

So, our posture is unchanged. We still own the buildout through reasonably valued names and will keep trimming where price runs ahead of fundamentals.

Consumer distress is real, and the 2-tier economy persists

The top decile of earners still drives a record share of spending. Businesses catering to the lower-income customers remain under stress. The three sector implications we flagged last month all stand:

Restaurants: still under stress on a third straight year of weak traffic, price pushback, and GLP-1 headwinds. We have done deep dives on Domino’s and Wingstop. We will be on the lookout for opportunities here and remain very strict about the valuations we pay.

Medical devices: still see elective-procedure deferrals, which created our entry into Cochlear. This is a segment with many high-quality companies we are researching. The GLP-1 related headwinds are affecting this group as well (healthier people lead to lower need for medical devices).

High-end luxury: this is where the setup can quickly turn around now that the war has ended and if the trade war/tariffs are not revived.

Portfolio Holdings Update

Read more

Kelley Blue Book and Edmunds, the two pricing guides American dealers and shoppers rely on to value a used car, both grade a vehicle's condition partly on its paint, and chips, scratches, or a dull finish drop it a grade and knock down the price. Pristine paint is rarer than people think. Only about 5% of used cars clear Edmunds' top ‘Outstanding’ grade, and everything below it gets marked down for the cost of reconditioning the paint, which dealers subtract from their offer. Clear PPF keeps that factory finish intact, and because it is invisible it sidesteps the resale penalty a visible wrap can carry. XPEL is the name buyers and installers ask for, the most recognized premium film in a category it has out-grown for a decade.

XPEL started in San Antonio in 1997 to write software, not to make film. That software, called DAP (Design Access Platform), is a digital library of cutting templates. An installer enters a car's make, model, and year, and DAP drives a machine that cuts the film to the exact shape of each panel on a bench, before any of it touches the car. That removes the slow and risky part of a wrap, hand-trimming film on the vehicle, where one slipped blade can scratch the paint. DAP arrived in the mid-2000s, years before XPEL made film at scale. So this is a software business that grew into film, not a film company that bolted on software, and that is why it keeps taking share from rivals a hundred times its size. It helps that its customers, thousands of small dealers and installers, have no real leverage over price.

Welcome to Rebound Capital. We study beaten-down stocks and businesses that have made successful comebacks. Subscribe for free to join our growing community of 22,500+ subscribers and make sure you don’t miss our next briefing.

Subscribe now

The stock peaked above $100 in 2021 and now trades around $45, roughly 56% below that high. The de-rating was due to 1 soft year in 2024, a short report in 2023, and a US auto market that's been visibly weak into 2026. Based on our research, we believe the market is overlooking multiple catalysts over the next 2 years that could transform XPEL into a higher-quality business.

How XPEL makes money

XPEL earns revenue from 4 major product lines:

Paint protection film (PPF) is the flagship, about half of revenue. It’s a urethane film cut to a car's exact panels to stop chips, scratches, and staining. XPEL sells the most expensive film in the category. Despite higher prices, they have been gaining market share (XPEL’s growth rate has been higher than the PPF market’s mid to high single digit growth).

Window film accounts for ~16% of revenue, including automotive tint under its own brands (PRIME).

Installation services are about 18% of revenue and growing. But installing is not how XPEL mainly makes money. Most of its film goes out wholesale. XPEL supplies the film, the DAP software, and the training to thousands of independent installers and dealerships. They do the work and keep the labor fee. XPEL keeps the product margin. The installation line is where XPEL does the job itself, through 34 company-owned centers and the labor it runs inside dealerships and at OEM plants. On those cars it takes the labor margin too, not just the film, and it learns first-hand what buyers choose and what they come back for. We read it as a deliberate step up the value chain in select markets, not XPEL's main route to the customer.

DAP software is under 5% of revenue, but it earns far higher margins than the rest of the business - closer to pure software than to the low-40s gross margin XPEL earns overall. It is also the deepest part of the moat. This software provides the correct PPF film for any car model that requires protection. There are more than 90,000 configurations - due to the number of different car models, the different versions of the same car over the years, and so on.

Cut the wrong pattern and you have to scrap the film. That’s a few hundred to a couple thousand dollars wasted. Botch the install and the whole job gets stripped and redone, film and labor both, which on a premium wrap is $5,000 or more. And if a bad install or its removal lifts the paint, the owner is repainting panels at $500 to $3,000 each. That is the cost of getting it wrong, and why the software is important.

So the software that cuts the film (see video below) is a deep moat. Competitors have not been able to match the depth of configurations in XPEL’s DAP software.

The moat: why XPEL beats 3M and Eastman Chemical

XPEL goes up against 3M and Eastman, two industrial giants with R&D budgets that it can’t match. And yet it is gaining share driven by its DAP software, brand, and vertical integration.

DAP Software

When an installer wraps a car, the hard part isn't sticking film to metal. It's cutting hundreds of vehicle-specific shapes without scratching the paint, where one slip means a repaint that costs thousands. DAP lets an installer enter the make, model, and year and pull up a ready-made pattern. This cuts installation time by an estimated 25%-70%.

XPEL holds about 60% of the pattern-software market, a larger share than it holds in the film itself, and the software lead tends to pull film share up behind it.

The library is only the start. Once a shop runs DAP, it accumulates customer history, work orders, plotter calibrations, and warranty submissions within the system. Many shops carry several film brands but run DAP for all of them.

Brand

XPEL can charge higher prices for its products. This is its brand power in action. Installers quote SunTek and LLumar as a set percentage below XPEL. Dealers and installers charge more than 3M or Eastman product, protecting XPEL's pricing and its partners' margins. Like for like, XPEL runs about 15 to 20% above Eastman value films and is on par with or above 3M, the other premium name.

Vertical Integration

Over the past 6 years, XPEL has acquired distributors in Canada, France, Australia, and India, took a 76% stake in its China distributor in September 2025, and built its own installation network. In November 2025, XPEL said it would bring film manufacturing in-house with a $75 to $150 million investment. In May 2026, XPEL announced plans to spend ~$110 million on a 435,000-square-foot, four-building site in San Antonio that will become its North American manufacturing hub, and the purchase of a film manufacturing facility in China. Once both are running, XPEL will own software, manufacturing, distribution, installation, and a slice of retail. No other PPF brand is vertically integrated to this extent. This helps XPEL deliver higher profit margins.

PPF market’s secular growth

XPEL’s share gains would matter less if the pond were drying up. It isn’t. Independent estimates put the global PPF market near $500 million in 2024 and on track for roughly $650 million by 2028, a mid-to-high single-digit annual pace. The US sits around $105 million and grows about 6% a year. International runs faster. Europe is the largest single region today, and management likens it to where the US stood five years ago. Asia Pacific grows quickest, led by China and India.

One caveat we won’t paper over. These third-party figures scatter badly, from under $400 million to north of $1.3 billion, depending on whether a firm counts raw film, installed retail, or something in between. The low end is hard to square with XPEL’s own PPF revenue of roughly $240 million, which tells us the real installed market is a good deal larger than the headline numbers imply. So we anchor on the one thing every source agrees on: the direction. The category compounds in the mid-to-high single digits, pushed by longer vehicle ownership (the average US car is now past 12 years old), rising resale-value awareness, EV adoption, and film that heals its own scratches.

For XPEL the takeaway is simple. It has out-grown this market for a decade, taking share on top of a base that is itself expanding. That is two tailwinds, not one, and it is why we don’t read XPEL as a slice of a shrinking niche.

What went wrong

The 2024 air pocket

After four years of 22 to 63% growth, revenue slowed to about 6% in 2024. The slowdown was driven by a Chinese distributor de-stocking, post-COVID luxury digestion, and a transition in the Tesla and Rivian channel. By Q1 2026, growth was back to 13%.

The 2023 short report and Tesla scare

Two negative catalysts in 2023 pressured the stock. One was a short report by Culper Research & the second was fears that Tesla would do its own PPF and that Tesla is a large part of XPEL’s revenue. The shorts argued that Tesla accounted for 25%-35% of sales, that a PPG and Entrotech joint venture would launch an OEM paint product that would kill PPF, and that OEMs would insource and commoditize the category.

None of the elements of the short thesis worked out. Management put Tesla at about 5% of sales, a fraction of what the report implied. The PPG paint product is real, but it still has not displaced aftermarket film, and XPEL grew revenue 13% in 2025 anyway. OEM insourcing never showed up either. Wrapping a car is fiddly work, fit by hand panel by panel, and it does not drop cleanly onto an assembly line. If anything, XPEL has done the opposite of getting squeezed out. It now runs the OEM channel as a tailwind, with Tesla sending buyers straight to XPEL and Rivian offering the film factory-installed.

The macro

The US new-vehicle market has been soft into 2026, with year-over-year declines. Because the market still treats XPEL as an auto cyclical, a weak vehicle market weakens sentiment.

Re-rating catalysts over the next two years

The key business drivers are all heading the right way.

Read more

But by late 2025, Netflix was the one chasing legacy content. In December 2025, it agreed to buy Warner Bros. Discovery’s studios and streaming business for roughly $83 billion. The disruptor was now trying to become a consolidator.

The market did not like it. 2 things spooked investors: the roughly $59 billion of debt the deal would require, and the suspicion that this merger was being done to mask slowing organic growth. The deal finally fell through. Netflix was outbid by Paramount. The stock rose ~40% over a couple of months after this news, then gave back most of the gains. It now trades ~$81, down nearly 40% from a 52-week high of $134.

In January, we put Netflix in the too-hard bucket. We said we would revisit it once the merger was resolved or the multiple came down. Both have now happened. We sense an opportunity here.

The market has taken a maturing growth narrative, layered on AI fears (AI-generated content and short-form video eating into engagement), and event risk (the abandoned Warner Bros. bid and Reed Hastings leaving the board), and marked the multiple down.

What it has underestimated is Netflix’s strong free cash flow generation. Trading at ~3.3% NTM FCF yield, Netflix offers a rare combination of double-digit earnings growth and a mature level of cash generation, supporting continued buybacks and shareholder returns. With the advertising engine just beginning to gain momentum, FCF can grow at a mid-teens CAGR over the coming 3-5 years. This is not a demanding valuation, given that Netflix is a market leader and will continue to compound for many years.

How Netflix makes money

Netflix reports a single segment, but earns revenue from two distinct businesses. The first is the subscription business: about 325 million paid memberships at the end of 2025, monetized through a tiered pricing ladder. The second is advertising, still small but scaling fast. We model the two separately. Subscriptions are an annuity. Advertising is a high-margin opportunity where Netflix can leverage the platform it has built over 20 years and the nearly one billion people it now reaches.

Welcome to Rebound Capital. We study beaten-down stocks and businesses that have made successful comebacks. Subscribe for free to join our growing community of 22,400+ subscribers and make sure you don’t miss our next briefing.

Subscribe now

The core subscription segment

This segment earns revenue through subscriptions. Here, Revenue = (no. of subscribers * average revenue per subscriber). Several critical levers drive subscriptions:

• Pricing power: Netflix pushed through a US price increase in March 2026, and year-over-year retention improved across all regions during the cycle. This proves Netflix’s strong pricing power. The US Standard plan has climbed from $7.99 in 2011 to $19.99 in March 2026, roughly a 150% increase, while adding hundreds of millions of members along the way. The latest March 2026 move lifted all three tiers by an average of about 11%. Independent analysts at JPMorgan pegged the annualized revenue uplift at roughly $1.7bn and flagged minimal churn risk.

• Category-leading retention: Data from Antenna indicates Netflix has the lowest churn in the category, generally under 2% per month, compared with a premium-SVOD average of 4% to 5%. When a household decides to trim its streaming bill, we think Netflix is typically the last subscription to go, not the first. This structural habit creates a powerful lock-in effect.

The advertising segment

The ad tier has reached more than 250mm monthly active viewers by May 2026, up from 190 million in November and just 94 million in 2024. Ad revenue grew more than 2.5x in 2025 to over $1.5bn and is guided to roughly double to $3bn in 2026, with an internal ambition (leaked documents) of nearly $9bn by 2030. Most of that is incremental, high-margin revenue on content that already exists. We think the internal target may prove to be conservative.

In-house ad stack transition: Netflix has built its own first-party platform, the Netflix Ads Suite, and finished rolling it out across all of its ad markets in 2025, moving away from rented Microsoft systems. It now transacts programmatically through Amazon DSP, Yahoo DSP, The Trade Desk, and Google DV360, with >50% of impressions bought programmatically.

Netflix uses AI to build ad creatives matched to specific shows, with brands reporting significantly better execution. It has layered on AI media-planning tools, generative pause and interactive formats. The advertiser count has crossed 4,000, up about 70% year over year. Premium streaming commands premium CPMs (cost per 1000 views), and Netflix already sits at the top of the stack. It draws some of the highest ad rates in streaming, roughly $30 to $40 per thousand impressions, ahead of mid-tier peers like Hulu and Disney+. The edge is structural, with premium content, a light ad load of 4-5 minutes per hour, and scarce inventory that advertisers compete for. Pair that pricing power with the largest audience in streaming, and the ad business has a long runway.

Advertisers’ Feedback: To check if Netflix’s ad platform is working, we look at 2 different data points:

• Highest CPMs in the streaming industry.

• Netflix’s own data: at its May 2026 Upfront, it claimed campaigns run almost twice the TV norm for brand building and 23% above purchase-intent benchmarks, with 44% of its ad audience not reached elsewhere on TV.

• The more important signal is the revenue growth in advertising (2.5x in 2025 and 2x in 2026), and on the Q1 2026 call, the advertiser count crossed 4,000, up about 70% year over year, with programmatic nearing half of non-live sales. That is a vote cast with real money.

Geographic Distribution: UCAN, EMEA, APAC, and LATAM

• US and Canada (UCAN): The largest region, at roughly $20bn revenue, actually reaccelerated to mid-teens growth as price increases flowed through. This region can compound at rates in the high single-digit to low double-digit range in the coming years.

• Europe, the Middle East, and Africa (EMEA): Now nearly a third of revenue at about $14.5bn, this region grew in the high teens in Q1’2026.

• Asia-Pacific (APAC): This is the fastest-growing engine, compounding at ~20% (in 2025), driven by India, Japan, South Korea, and Southeast Asia.

• Latin America (LATAM): The most price-sensitive region grew in the high single digits on a reported basis, but constant-currency growth was roughly 30% in 2025. The drag there is due to currency translation, not customer demand.

We estimate that only the US, Canada, and Europe are close to being saturated. The rest of the regions can support a mid-single-digit to low-double-digit CAGR in subscribers till 2030.

What Went Wrong

The current drawdown in Netflix from its June 2025 high near $134 to around $81 (a ~40% decline) is driven by a fundamental narrative shift: the market has derated Netflix, fearing disruption from AI-generated content and maturing markets.

• Growth metric anxiety: Netflix stopped disclosing paid subscriber counts and churn at the start of 2025, leaving the market anxious about slowing headline net adds and shifting visibility.

• The engagement plateau: Short-form video content on YouTube and TikTok is competing for attention and capping engagement growth, with growth slowing to roughly 2% in H2 2025. Digital i’s measurement shows average daily use per Netflix account slipped from 101 minutes to 93 minutes, while YouTube account use rose to 99 minutes, indicating real cannibalization of hours.

• Generative AI structural fears: Bears worry that generative AI will flood the world with cheap synthetic video content and erode the long-term value of a premium owned-and-licensed library.

• Headline noise: Headline and event risks have weighed heavily on sentiment, including the abandoned $83bn Warner Bros. bid and Reed Hastings's leaving the board in June 2026.

Netflix’s Competitive Moat

Netflix’s moat is a system of interlocking competitive advantages - each reinforcing the others and proving highly resilient to the current environment.

Economies of scale

Read more

We are publishing this without a paywall

Back in 1998, a surfer named Chip Wilson took a Vancouver yoga class and walked out annoyed by the clothes: baggy cotton that sagged and went see-through. So he sewed better pants from technical fabric and gave them to yoga instructors to wear while teaching. They became walking billboards. The pants sold out, and Lululemon was born. Over 25 years, that one idea, quality workout clothes styled for exercise or yoga, grew into a company worth more than $60 billion at its peak. Lululemon did not join the athleisure category. It invented it.

Now, the founder is at war with his own board, the longtime CEO is gone, US sales are shrinking for a second year in a row, and the stock is down ~80% from its 2024 highs. The fashion press says the brand has lost its cool, that kids are buying knockoffs from Costco, and that Alo and Vuori are the names on everyone’s feeds.

If you’re in a rush

What they do: Lululemon sells premium athletic clothing, mostly through its own stores and website, with a lineup that runs from leggings and the famous Align pants to a growing men's range and a young footwear line. Because most of that product moves at full price through its own channels, gross margin is in the low to mid-50s, far above that of almost any other clothing company.

Why it crashed: The Americas, its largest and most profitable market, has not grown for about two years, with nine straight quarters of flat or declining same-store sales. In the latest quarter, America’s net revenue declined 3%, and America’s comparable sales fell 5% (-6% on a constant dollar basis) in the first quarter of fiscal 2026. Management has lowered its full-year outlook and now guides to a 0% to 1% decline in total net revenue for fiscal 2026. Other compounding factors were: margins squeezed by tariffs and discounting; a public proxy brawl with founder Chip Wilson that began after the company announced CEO Calvin McDonald's exit and only settled in late May; the leadership vacuum and the perception that Alo and Vuori are stealing the crown have caused Lululemon to fall out of favor with investors.

Possible catalysts: A product refresh already showing up in the spring numbers, a new CEO poached from Nike who starts in September, and a strange potential new tailwind: weight-loss drugs pushing more people into the gym and into new clothes.

How Lululemon makes money

The best way to segment Lululemon’s revenue is by geography. It reports three segments.

The most important metric is in the bottom row. The % of revenue from the Americas (the saturated market) fell from 85% of the business in FY2021 to 71% by FY2025, not because it shrank but because it grew more slowly, even as revenue rose from $5.3 billion to $7.8 billion. The two international engines simply grew faster. China Mainland revenue compounded at roughly 42% per year from a small base to $1.75 billion in FY2025, while the rest of the world tripled. So you are buying two businesses in one - a large, premium, very profitable Americas business not growing and in need of a refresh, and a smaller international one still growing like a startup.

Margins are under pressure. While gross margin was 56.6% last year, it dropped to 54.2% in the first quarter of fiscal 2026. Operating margin contracted sharply to 11.2% in Q1 FY2026, down from last year's 18.5% in Q1 FY2025, and well below its historical peak of 23.7%. The balance sheet remains resilient, showing $1.5 billion in cash and cash equivalents with essentially no debt. Meanwhile, management continues to aggressively buy back stock to defend the equity, purchasing 2.2 million shares for $358.3 million in the latest quarter, building on the $1.2 billion of stock repurchased last year.

A word on the margin. For all the talk of Lululemon losing its grip, the margin was remarkably steady till 2025, ending that full year at a robust 56.6%.

But now, gross margin has materially dropped. Gross margin plummeted 410 basis points to 54.2% in the first quarter of fiscal 2026. While a significant portion of the dip still stems from macro pressures, with tariffs causing a 280 basis-point reduction, markdowns have had a big effect as well. Markdowns are expected to rise 50 basis points in Q2 versus last year. Clear evidence is emerging that people aren’t chasing the product at full price like they used to.

Welcome to Rebound Capital. We study beaten-down stocks and businesses that have made successful comebacks. Subscribe for free to join our growing community of 22,000+ subscribers and make sure you don’t miss our next briefing.

Subscribe now

What went wrong

US demand rolled over

Americas’ revenue fell about 1% in FY2025. Americas’ revenue fell 3% in the first quarter of fiscal 2026, and same-store sales dropped 5%, with the full-year outlook dragged down to a 0% to 1% decline. The shopper’s complaint is blunt - the clothes got boring, and a $100 legging is a hard sell when it looks like last year’s. I’m sure the weak consumer sentiment isn’t helping.

Margin compression

In fiscal 2025, gross margin fell 260 basis points to 56.6% and operating margin fell 380 basis points to 19.9%, both down from their FY24 peaks, due to new US tariffs and heavier markdowns (discounts). The exit rate was worse, with fourth-quarter operating margin down 660 basis points year over year. In the first quarter of fiscal 2026, gross margin plummeted 410 basis points to 54.2% and operating margin collapsed 730 basis points to 11.2%, both down sharply from prior-year levels due to heavy US tariff headwinds, inventory provisions, and climbing store and proxy expenses.

The boardroom battle

Founder Chip Wilson, a major shareholder, ran a public campaign against the board for letting the brand go stale. In late May, they settled - Lululemon adds two of Wilson's nominees plus a third by October, Wilson stands down for 18 months, and Heidi O'Neill becomes CEO. Ex-CEO Calvin McDonald, who more than tripled revenue, left in January, and two interim co-CEOs are running things.

Increasing competition

Vuori and Alo Yoga have pulled younger, style-led shoppers toward their own premium athleisure, and a reviving Nike is pushing back into the category it once owned. Each is taking share that used to come to Lululemon without a fight.

Lululemon’s response is more concrete than the stock price would suggest. The problem it admits to is stale product, so it is lifting new styles from about 23% of the mix toward 35% by spring, cutting its design cycle to 12 to 14 months from 18 to 24, and pulling back on discounting to sell at full price again. It also hired Heidi O’Neill, a 26-year Nike veteran, to run product from September. The plan is sensible, but US sales are still falling, and the turnaround strategy is unproven.

The New CEO

The board hired Heidi O'Neill, who spent over 25 years at Nike running product and brand, most recently through the stretch that sent Nike into its own long slump. Investors wanted a fresh product-first fixer for the stale-product problem, not someone from the regime behind a rival's troubles. The shares fell on the news, and she does not even start until September.

Is the brand actually losing relevance?

This question decides everything, so I went looking for evidence.

The bear case is real. Younger shoppers increasingly call Alo and Vuori the fresher brands. Alo has built serious heat on TikTok and holds something like 14% of the premium direct-to-consumer athleisure market. Vuori, valued at around $5.5 billion and preparing for an IPO, owns the relaxed men’s corner where Lululemon has always been weaker. Brand-tracker BERA scores Alo highly on excitement and sophistication. And the dupes - Lululemon got so annoyed it sued Costco last summer over look-alike apparel. Add a couple of see-through fabric episodes/scandals, and the ‘Lululemon is over’ story writes itself.

Now, the other side, which gets less airtime. Even as the US activewear market softened, Lululemon’s sales have not collapsed. Its stores still achieve more than $1,400 of sales per square foot, which almost nobody in apparel matches, and it holds $100-plus price points, the clearest sign a brand has pull. The same BERA report that flatters Alo and Vuori shows their overall scores still sit in the 40s, the getting-to-know-you zone, while Lululemon is the benchmark. And dupes are a problem only aspirational brands have.

My read: the brand has cooled, not collapsed, and the cooling lines up almost exactly with a stalled product engine. When the clothes stopped feeling new, the energy drifted to whoever was shipping fresh ideas. A relevance problem caused by a stale product is fixable with a better product. Also, a huge part of the slowdown is due to weak consumer demand (most consumer discretionary businesses are affected).

But I must say that I have a contrarian bent of thinking, and that maybe is coloring my analysis here. Maybe Lululemon is a melting ice cube in the US. I’ll be humble and acknowledge that I cannot predict women’s (who drive Lulu’s sales) clothing preferences over any time period. Lulu has strong moats (brand and distribution), but these can only go so far. My research on whether the brand is becoming irrelevant in the US is inconclusive.

Will the US same-store sales turn positive?

The whole turnaround hangs on one lever: newness. Lululemon management has noted that new styles sell better than repeats and attract its highest-spending customers. So the number that matters most is the share of the lineup that is new. It fell to around 23% in mid-2025, roughly when US sales rolled over. The plan is to push it to 35% by spring 2026.

Lululemon has cut its main product development cycle from 18 to 24 months to 15 to 16 months and is targeting 12 to 14 months, while aiming to achieve about 20% more volume on its winning styles this year. It hired a Chief AI and Technology Officer to speed up design. So the strategy is not just a slogan.

So far the signals are mixed, not the clean win the bulls want. On the Q1 call, interim CEO Meghan Frank admitted that recent launches did not wow shoppers and that a wave of negative brand chatter dented traffic, though management still expects full-price selling to turn positive starting in Q2. The newer footwear push is one genuine bright spot - the Beyondfeel running shoe and Cityverse sneaker have seen solid sell-through since their 2024 debut - but it is still a small part of the mix. None of it has translated into any material improvement in the Americas numbers yet.

Coming to the biggest catalyst. The founder will now have 2 of his nominees on the board. And Heidi O’Neill, the 26-year Nike veteran who helped grow that business from $9 billion to $45 billion, takes over in September. Nike is primarily driven by sales to men, while Lululemon is driven by sales to women. Most brands tend to do well with either men or women. Even though Heidi helped grow Nike’s women’s division, I don’t think Nike has done very well with women’s apparel either. This is my pushback against this appointment, and I do agree with the market’s view that Nike itself is not a poster child for brand success in the last 5-10 years. So this appointment does not inspire much confidence.

A Strange Tailwind: GLP-1 drugs

Here is what makes this interesting. Weight-loss drugs (GLP-1), including Ozempic, Wegovy, and Zepbound, are changing how people shop, and their mechanism of action works in three ways. People lose weight, usually 10 to 20% in year one, which is one to five sizes, so they rebuild a wardrobe. They start exercising, and in one survey, the share of people exercising weekly roughly doubled, from about 35% to 71%, creating a new pool of buyers for workout gear. And people who pull off a visible health change like to show it, and premium athletic wear is how.

The scale is large. About 1 in 8 American adults has tried a GLP-1 and roughly 6% are on one today, and Bernstein estimates the wardrobe churn could add $3 billion to $13 billion a year in US apparel spending. Almost all of it is American, which lands it squarely on the 71% of Lululemon’s business the market has written off. But the catch is real - Bernstein’s bigger winners are the off-price and value names, not premium, since shoppers buy cheap clothes while their size is still moving, and only the higher-income end of the GLP-1 crowd is the kind who will pay $100 for a legging.

Examples of brands dealing with customer fatigue

Many brands have faced the same stale-product issues that Lululemon faces now, and the pattern in the table is not random. The ones that came back, Abercrombie, Crocs, and Lululemon in 2013, rebuilt the product and still had a wanted brand underneath. The ones that stalled either spent their brand equity on discounts (Under Armour, Gap) or missed a cultural shift they could not discount around (Victoria’s Secret).

Lululemon looks like a winner on what matters most - its gross margin is still in the 50s, so brand equity is intact, and it is moving early while remaining profitable. What I am wary of is the new CEO, hired from a weak (brand perception wise) Nike. Whether the new management will be able to pull off a turnaround cannot be answered by looking at historical precedents. And so even this is inconclusive. I have no confidence in predicting a turnaround at this stage. I will rather wait for it to happen.

Lululemon Valuation

We value Lululemon using a simple DCF model. One key caveat is that the model is my effort to put numbers around the story - Lulu’s model is very sensitive to the exit multiples, which will depend on the same-store sales growth for the Americas. This makes the DCF highly unreliable: as a small negative same-store sales growth versus a +2% figure will drive a material difference in the intrinsic value.

By the end of this decade (even right now), Lululemon is a maturing premium brand growing in the high single digits, so we set the multiple based on where mature branded apparel and footwear trade. The apparel group sits around 14 times forward earnings. Nike, the closest comp, trades ~25x-30x, but on depressed earnings, so normalized, it is lower. Ralph Lauren and Tapestry sit around 20 times, and Deckers, behind UGG and Hoka, at just 15 times. A premium, net-cash brand still growing 7% belongs in the upper half of that range, not above it.

So I conservatively use 15x FCF in the base case and 10x FCF in the bear case, where the brand keeps fading and the stock trades like tired apparel names.

The base lands near $153 versus ~$116 today, and the bear near $88. Unfortunately for Lululemon, unless they print a turnaround in the Americas, the market will not be interested in this name.

Lululemon Valuation

What could go wrong

Brand fade: If the US lineup fails to re-engage and comps stay negative, premium pricing will erode. This will be a slow death for the brand, and the stock will be a value trap. If it loses relevance in the US, the brand may lose its pull in China and the rest of the world as well.

Margins: Tariffs and markdowns could prove stickier than we assume, holding cash flow margin below our recovery path.

Leadership: Poor decisions by the new CEO will be difficult to recover from.

China’s growth falters: This would remove one of the bright spots from the Lululemon story.

Our take

We rate Lululemon a ‘HOLD’ with a ‘Sell on Rallies’ bias. This is in the too-hard bucket. Essentially, we will wait and watch the KPIs and revisit the thesis post-recovery (if any) in the US same-store sales.

The base case materializes only if the US stops shrinking (if and when that is) and starts growing again, with margins climbing back up. Take that away, and you fall straight back to the bear case. So the value is real, but it hangs on one thing, and that one thing has not shown up yet. That is why we are staying away. We will be patient here.

Share

Rebound Capital’s work is provided for informational purposes only, is intended solely for readers in the United States, and should not be construed as legal, business, investment, or tax advice. You should always do your own research.

There’s this one idea we always came back to: investing in quality companies in a drawdown.

Here’s what we wrote about Meta in 2022:

No matter how you look at the company, the financial metrics are incredible. 80% gross margin and 30% ROE with growing revenue. But the stock is down 55% in the last 1 year.

I am not pitching it as a great business. They certainly have some incredible challenges ahead of them in terms of privacy and attracting a young user base. No one knows directly the solutions for their hardest problems—that’s why they’re the hardest ones.

But writing off a company based on two bad quarterly reports is just too short-sighted.

The same theme kept recurring:

• Netflix lost 3/4th of its value because it did not grow subscribers one quarter.

• ASML lost ~1/2 its market cap on tariffs and export restrictions.

• Amazon was down 56% after its first quarterly loss since 2015.

So we finally pulled the trigger and launched Rebound Capital, putting our own capital on the line. We followed Charlie Munger’s advice of “take a simple idea and take it seriously” and spent the past year going through hundreds of stocks to find the optimal rebound setups.

Fast forward a year, and we are outperforming the S&P 500 while holding top 3 positions in relative value names such as Amazon, Constellation Software, and MercadoLibre. To put this in context, 79% of all active large-cap U.S. equity funds underperformed the S&P 500 last year.

While it’s easy to talk a good game when your portfolio is outperforming, sticking to the strategy when it's underperforming is the hardest thing to do as a public facing equity research platform. Q1 was hard on the Rebound Portfolio, where we underperformed the market by 9%. We continued to share our portfolio performance publicly throughout the period of underperformance (here are our Feb, Mar, & April updates). But sticking to our strategy helped us double down on quality names that rebounded strongly after the March dip.

Not to name names, but there are a lot of “gurus” who have Excel spreadsheets where the portfolio never goes down, and they magically exit at the right time. For us, we have maintained the policy of informing you of all our trades 24 hours in advance, and honestly, our biggest success till now is that nobody has ever said anything bad about us.

Overall, we are now in a good position with our portfolio compounding faster than the index while trading at essentially the same valuation. The headline numbers (weighted average of our portfolio holdings):

• Revenue growth, next 3 years: ~18%

• EBIT growth, next 3 years: ~19%

• EV/EBIT: ~27x, in line with the S&P 500

Here is the unusual part. We own a basket whose EBIT is set to grow ~19% per year over the next three years, compared with <13% for the S&P 500, and we pay no premium on the multiple to do so.

Divide valuation by growth, and the portfolio sits at ~27/19 = ~1.4, versus 27/11 = ~2.5 for the index (lower is better). In short, our portfolio compounds faster than the market while costing about the same, which on a growth-adjusted basis makes it materially cheaper than the index it is built to beat.

Your support has been incredible, and it's what makes this work feel meaningful. While most creators are hush-hush about how much they make, we are going to show you our real numbers. After all, this is the easy bit after sharing our portfolio performance :)

As of today, we have more than 21,000 investors who subscribe to us, of which 969 investors pay to read our research, which translates to revenue of $358K (ARR). The spike you see in December is when Michael Burry became a paid subscriber to Rebound Capital and endorsed our note on Birkenstock.

Consider this your nudge to put your own work out in public. You never know who’s reading.

Finally, of all the metrics at Rebound Capital, our retention rate is the one that we are most proud of. ~90% retention at 6 months is unheard of in the newsletter space and is the best testament to the quality of the research we produce and the value you derive from it.

Right now, the market is in love with AI. Money is pouring into a narrow group of names at prices that already reflect the next 5 years of growth. To be clear, we are bullish on what AI can do over the long run. But when an equipment vendor that has always been cyclical gets priced as if the cycle has ended, the reward for chasing it is small, and the cost of being wrong is high. With many indexes changing their decade-old rules to accommodate hyped-up IPOs like SpaceX, indexing is no longer the safe, neutral choice. The case for doing your own research only gets stronger from here.

And this is where being independent matters. A fund manager has to answer to a committee, hug the benchmark, and own the hot name so they don’t look wrong at the next quarterly review. We answer to no one but you. We never have to chase the story of the moment and can sit in an out-of-favor name and wait for the market to catch up, because no one is going to pull our capital for being early.

While investors flock to one part of the market, high-quality companies are being ignored. Some of the best compounders we know are at their lowest multiples in years. We will continue to focus here. The crowd overpays for what it loves and ignores the good businesses on sale. But the market eventually catches on.

So we will keep researching the names nobody wants to talk about. The rotation back to high-quality names will come. Our job is to be patient enough to hold the right names when it arrives.

Have a great weekend. We’ll see you Monday.

Subscribe now

Rebound Portfolio Performance

Happy to report that the Rebound portfolio is now beating the market (S&P 500) since inception by ~4.7%.

Drawdowns are always tough. But we stuck to our core principles during the February and March drawdown (driven by AI-doom and the Iran War). We kept researching, building DCFs, and looking for stocks in steep drawdowns.

This helped us invest in quality stocks at attractive prices and resulted in us outperforming the market by 14%, since the market bottomed on March 30th.

Looking ahead, we are focused on:

1. Derisking from the AI theme: This cycle is real, and we are still early. But AI-related stocks (especially Semiconductors) are now discounting multiple years of strong growth. Any hiccup, however temporary, will lead to a sharp fall in these AI names. The valuations that the new-age AI companies (SpaceX) are getting point to public market investors behaving like VC investors - and this will not end well. What I mean is that VCs can invest in companies that are not profitable, as they expect some of their companies to become 10x or even 100x. So they just pay for innovation. You can’t pay $2 trillion for a company and think like a VC (as the company cannot 10x). Just my 2 cents. We are now pivoting our portfolio and trimming exposure in areas with frothy valuations.

2. Quality companies overlooked amid AI mania: Multiple companies with years of strong revenue growth are now trading at their lowest multiples in a long time (Visa, SPGI, Moody’s, etc., to name a few). We will investigate such names to determine whether the lower valuations suggest a sound investment.

Portfolio statistics

Rebound Capital’s portfolio compounds faster than the index while trading at essentially the same valuation. To gauge the portfolio’s growth and valuation, we take a weighted average of each holding’s metrics. The headline numbers:

• Revenue growth, next 3 years: ~18%

• EBIT growth, next 3 years: ~19%

• EV/EBIT: ~27x, in line with the S&P 500

Here is the unusual part. We own a basket whose EBIT is set to grow ~19% per year over the next three years, compared with <13% for the S&P 500, and we pay no premium on the multiple to do so.

Divide valuation by growth, and the portfolio sits at ~27/19 = ~1.4, versus 27/11 = ~2.5 for the index (lower is better). In short, the portfolio compounds faster than the market while costing about the same, which on a growth-adjusted basis makes it materially cheaper than the index it is built to beat.

Broader Market Commentary

AI adoption is in early stages, but AI stocks price in 3-5 years of growth

The latest quarterly reporting season put the AI bubble narrative on the back foot. The hyperscalers are not slowing their spend - they are raising it:

• Alphabet lifted its 2026 capex guidance to $180-$190 billion and said 2027 would see a significant increase.

• Amazon reiterated roughly $200 billion in 2026 capex. AWS accelerated to 28% in Q1, its fastest pace in fifteen quarters, against a $364 billion backlog that is up 40% year-over-year.

• Google Cloud grew 63% to $20 billion (accelerating from 48% the prior quarter) with backlog nearly doubling to over $460 billion.

• Nvidia capped it off on May 20 with record revenue of $81.6 billion (+85%), data center revenue of $75.2 billion (+92%), and a $91 billion guide for the next quarter.

The takeaway from every management team was the same - the shift to agentic AI is increasing demand for compute. Adoption is progressing from experimentation into production budgets. Caution is warranted in AI related stocks due to valuations, but the underlying trend is fundamentally strong.

Panic Buying in Semiconductor Stocks

The move in semis since March has been indiscriminate. AMD is up ~330% over the trailing 12 months and now trades at 50x forward earnings, compared with Nvidia at 24x. The semicap names have rerated even harder: ASML at roughly 52x trailing earnings versus 38x at the end of 2024, Lam Research at 60x against a ten-year average of 20x, and Applied Materials at 42x against a ten-year average of 18x. These are cyclical equipment vendors being valued as if the cycle no longer exists.

The memory and optics complex is where the melt-up is cleanest. SK Hynix is up about 248% year to date and Micron 226%, with both joining Samsung in the trillion-dollar club. In optics, Applied Optoelectronics is up roughly 335% YTD, Lumentum 120%, and Coherent close to 80%. Some of the other Tier 2/3 suppliers are up similarly.

We are not bearish on the AI buildout - we own it. But price has run ahead of fundamentals across much of this group, which is what informs the trimming you will read about below.

Not All SaaS Is Created Equal

SaaS has bounced over the past few weeks, but we still view AI as a material risk to the revenue model and think deep discrimination between names is warranted. Rebound Capital currently has roughly 16% exposure to this segment, and we plan to keep that allocation roughly constant rather than chase the rally. We continue to dig into more SaaS names.

The broader picture has not improved much. Investors are repricing the risk that AI will compress seat-based pricing and let customers do more with fewer tools. Much of this derating is deserved. The easy land-and-expand era is over, and businesses whose moat was workflow inertia rather than genuine switching costs are exposed.

The nuance is that SaaS is not monolithic. Owners of mission-critical, deeply embedded software will do well and can even benefit from AI. That distinction is the investment opportunity.

Consumer distress is real: Implications for consumer stocks

The 2-tier economy is showing up everywhere in the data. The top 10% of US earners now drive 49.2% of all consumer spending, a record since 1989 and up from roughly 36% in the early 1990s. McDonald’s says lower-income traffic has fallen at a near double-digit pace for almost two years while higher-income traffic rises at a similar rate. Chipotle and Cava benefited from customers trading down. Delta’s main cabin revenue fell 5% year-over-year while premium ticket sales rose 5%, and luxury hotels like Four Seasons are outperforming budget brands.

Three sector implications:

• Restaurants: the sector has been brutalized. Names like Chipotle (down -54% from all time high), Domino’s (-45%), Wingstop (-63%), and DoorDash (-27% YTD) are posting historic drawdowns. BTIG analysts have noted that sentiment is at levels only matched by COVID and the 2008 crisis! The drivers are a third straight year of weak traffic, pushback on price hikes, GLP-1 drug adoption curbing food spend, and K-shaped pressure on chains over-indexed to lower-income guests. We are interested in this sector and have done deep dives on Domino's and Wingstop. Some names are now trading at interesting valuations, and we will be patient before deploying.

• Medical devices: the same K-shaped pressure has emerged as consumers delay elective procedures. That hesitancy has created an opportunity in one med tech name, in which we have built a position. The company is the global leader with roughly 60% market share, a recurring upgrades revenue tail, and a structurally underpenetrated patient base. The surgical deferrals cannot be extended too much. The company’s moat remains intact. Full thesis here.

• High-end luxury: even the top of the K is not immune. The Iran war disruption has pushed the long-awaited recovery further out. LVMH now trades at roughly 21x trailing earnings, against a ten-year average >26x, while the broader luxury basket is closer to 17-18x forward, against five-year averages in the mid-20s. The setup is starting to look interesting (more on LVMH below).

Portfolio Holdings Update

Read more

32 years later, Wingstop has 3,153 restaurants. Around 98% of the system is operated by franchisees - what the company calls ‘brand partners’. The company had 21 consecutive years of positive domestic same-store sales (2003-2024). Wingstop keeps the brand IP, the flavors, the technology platform, and a 6% royalty on every dollar of system sales, while taking on almost none of the operational complexity. This is an amazing business model, akin to Domino’s Pizza.

The brand partners are very happy too. Brand partner unit-level cash-on-cash returns are >50% for a ~$575,000-$650,000 buildout. The payback period is close to 2 years for most partners! No wonder Wingstop has a pipeline of 2,200+ committed restaurants, even in a weak consumer environment.

But the stock is down ~65% from its all-time high.

Welcome to Rebound Capital. We study beaten-down stocks and businesses that have made successful comebacks. Subscribe for free to join our growing community of 21,900 subscribers and make sure you don’t miss our next briefing.

Subscribe now

The drawdown is due to the 2025-26 deceleration in domestic same-store sales growth (SSSG%). SSSG% fell to (-3.3%) in fiscal 2025 (the first negative year since 2002). Q1 2026 was worse at -8.7%. The franchise royalty engine, by contrast, is not broken, and brand partners are still investing to open Wingstop stores (over 2,000 restaurants in the pipeline). The debate isn't whether Wingstop is a great business; it is. The debate is whether a great business trading at ~23x EBITDA, with its core customer under pressure (and weak consumer sentiment more broadly), is a buy today.

How Wingstop makes money

I would say that Wingstop is not really a restaurant company. This is a company that owns IP and a brand, which it licenses at very high operating margins, and it operates 3 businesses under that brand.

Royalties

Wingstop gives out franchise rights to its partners. These ‘brand partners’ then pay a 6% royalty on every dollar of domestic system sales (& 5% on international sales), plus franchise fees, technology fees, and rebates. Wingstop does not deal with food costs, labor, rent, or working capital. Wingstop has ~1,335 corporate employees versus 3,153 restaurants. Corporate complexity scales sublinearly, while royalty revenue scales linearly with system sales. FY2025 royalty/franchise/other revenue was ~$322M, with an estimated ~40% operating margin (excluding the marketing revenue, which is a pass-through). This revenue segment is what makes Wingstop an elite business.

The advertising fund

Brand partners contribute 5.5% of sales to a national advertising fund. This is a near-perfect pass-through - fees in, advertising expenses out, with a zero economic margin to corporate. The strategic value is the scale. A 5,000-store system contributes more dollars to the same Super Bowl spot than a 3,000-store system. It is a brand-building utility, not a margin business.

Company-owned restaurants

Wingstop owns 57 corporate stores, mostly in the Dallas Fort Worth area, generating ~$128M of sales at ~24% restaurant-level margin. This is small in absolute dollars, but strategic. Every new operating standard, technology platform, and menu innovation gets tested here first. Corporate AUVs (Average Unit Volume: sales in a year) are tracking toward $2.5M, a $500K premium over the system average, and serve as a proof point of what disciplined operations produce.

Wingstop is a royalty platform selling its IP and brand, an advertising fund, and a small laboratory of company-owned restaurants. The royalty segment is what creates the economic value. The other 2 segments are used to maintain the brand's vibrancy.

Unit economics

I will dig into the unit economics for both the brand partners and the corporate/company (which we will enjoy as investors).

Franchisee economics: A Wingstop franchisee typically spends $575,000 to $650,000 to build a single location. In return, the store generates roughly $2 million in annual sales while maintaining a 22% to 25% restaurant-level profit margin - allowing owners to recover their initial investment in just 2 years. These metrics represent some of the strongest unit economics in the Quick Service Restaurant (QSR) industry. This rapid payback is driven by a comparatively smaller restaurant size - locations average just 1,400 to 1,700 square feet, completely bypassing costly drive-thrus and large dining rooms to focus entirely on takeout and delivery. Ultimately, these elite-operator returns are driving explosive brand growth for Wingstop. As an operator, why choose any other brand when Wingstop delivers such stellar returns? This is a significant moat.

Company economics: The corporate economics are stellar. They are even better than the headline numbers. Wingstop reported a 25.7% operating margin on FY2025 revenue of $697M, except that about $248M of that ‘revenue’ is just the advertising fund passing through the P&L (fees in, ads out, no economic margin). Strip it out, and the real operating margin on the business Wingstop actually keeps is closer to 40%!

Why is it down 65%?

5 factors explain the move from $420+ in mid 2024 to $157 today:

• Same-store sales broke a 21-year positive streak: Domestic SSSG went from +19.9% (2024) to -3.3% (2025) to -8.7% (Q1 2026). Consumer distress has played a large part in this. It’s difficult to say when the consumer will bounce back, though management expects a bounce back in the second half of 2026.

• Multiple compression as growth slowed: At the mid-2024 peak, the stock traded above 40x EV/EBITDA, a multiple that only made sense against 20%+ same-store sales growth. As that growth turned negative, the market re-rated it toward a more normal ~23x (LTM). That compression accounts for ~50% of the decline, and it occurred only because of the slowing growth.

• Q1 2026 one-time shocks: Atypical winter weather temporarily closed 700+ restaurants in January. Then, the Middle East conflict pushed gas prices higher. Finally, the lower-income cohort is in consumer distress and spending less.

• Repeated downward guidance cuts damaged management's credibility: Wingstop entered 2025 guiding to low to mid single-digit domestic same-store sales growth. After Q1, it cut that to roughly 1%. After Q3, it cut again, this time to a 3-4% decline. Due to these frequent cuts, the market has stopped paying for the medium-term targets, including the $3M AUV goal, until they are proven.

• Secular fears about GLP-1 drugs: The longer-term effects of GLP-1 drugs on eating out are inconsistent, but many reports highlight a lower propensity to eat out (as we highlighted in our Domino’s Deep Dive). But still, there is now a strong possibility that eating out may be structurally inhibited by people consuming GLP-1 drugs. Mind you, these drugs can keep getting better, and so the full effect is still largely unknown.

These factors are correlated, not independent. None permanently impairs the underlying business engine, but a few of these can be long-term headwinds to growth.

Competition and International Expansion

Wingstop’s scale edge in the US is structural and widening. It ended Q1 2026 with 2,653 US restaurants and a pipeline of 2,200+ committed units, so the system keeps compounding no matter what any single quarter’s comps (same store sales growth) do.

Its closest legacy rival, Buffalo Wild Wings, plays a completely different game. BWW runs ~1,180 US sports bars at ~$3.6M AUVs (and a much larger restaurant area, on average, than Wingstop), and the chain hasn’t grown its unit count since 2018. Wingstop added 493 net-new restaurants in 2025 alone. BWW franchisees also spend much more on store setup ($2.5M to $5M).

Popeyes put wings on its menu in 2023, but it's still fundamentally a fried chicken and sandwich chain. And the long tail of regional wing shops has been consolidating for a decade, with Wingstop the main beneficiary.

The bottom line: at its scale, Wingstop doesn't really have competition. It has the most stores, offers the highest returns to its brand partners (likely in my estimates), and is the only growth engine still running, while everyone else is shrinking, defending their footprint, or is too small to matter.

Outside the US, international expansion offers great upside. Since the wing occasion barely exists abroad, Wingstop is building a brand new category instead of fighting established rivals. Beyond mature markets like the UK and Mexico, the brand is expanding its digital model into Canada, South Korea, and Saudi Arabia. These countries will drive near-term growth, but India is the real game changer. With a commitment to over 1,000 stores, India could be a key driver of long-term growth.

The Moat

Wingstop’s durability rests on four reinforcing advantages. Together, they explain why brand partners continue to invest in new Wingstop stores even as consumers are in distress.

Share this with a friend

Share

The franchise royalty model: with high returns on capital for brand partners

95% of 2025 new restaurant openings came from existing brand partners (management commentary), a self-reinforcing loop that strengthens with system tenure. The operators who best understand the unit economics are the ones signing up to build more. Because Wingstop’s model enables high returns for brand partners, they keep investing in the brand and opening more stores. This is a virtuous flywheel for Wingstop, enabling it to scale rapidly.

Flavor IP and a lock on the next generation

Lemon Pepper, Atomic, Mango Habanero, and Hot Honey are flavors that are culturally embedded among younger consumers, backed by 30 years of investment. This is not a vague claim. Millennials and Gen Z make up a large part of Wingstop’s customer base, with 25 to 34 year olds the single largest cohort. The brand earned that position through exactly what those consumers index on - bold customizable flavor, shareable formats, a digital first ordering experience, and ownership of the game day and late-night occasions. The Q1 2026 Hot Honey Trio sold out two of three variants within two weeks. Owning the next generation of restaurant spending is a durable advantage, though a softer one than the franchise structure. Brand affinity erodes if innovation slows.

Structural cost advantage due to small restaurants and sourcing materials at scale

A third moat is structural cost, starting with the restaurant itself. A Wingstop is 1,400 to 1,800 square feet with no drive-thru and barely any dining room. Roughly 90% of sales are off-premise, and more than 70% are digital. This reduces setup costs. The Smart Kitchen concept has recently been implemented, reducing ticket times from about 20 minutes to 10. A large-format competitor with a full kitchen, dining room, waitstaff, and drive-thru cannot get its cost structure down to Wingstop’s level.

Scale shows up in sourcing, too. There is no fixed price market for fresh bone-in chicken wings. Starting in 2024, Wingstop moved most of its system’s wing purchases off the spot market into contracted pricing, giving brand partners cost visibility no independent can match. When wings spike, Wingstop’s franchisees hold their margins while the independent down the street gets squeezed. Smaller formats face similar issues. They can’t get as good a deal as Wingstop.

The data and advertising flywheel

The fourth moat is the one that compounds with scale. Wingstop has a customer database of more than 60 million users, built on a digital mix above 70% that most peers cannot approach, and a national advertising fund that collects 5.5% of sales from more than 3,100 restaurants. A 5,000-store system funds the same Super Bowl spot with far more dollars than a 3,000-store system does, and the first-party data beneath it gets richer with every digital order. As Club Wingstop rolls out nationally (membership plan), Wingstop will monetize its data.

Short-term headwinds vs. long-term opportunity

Short-term Headwinds: Wingstop’s core customer makes the brand more sensitive to macro than its peers. The CFO has said lower-income customers account for roughly 25% of transactions, and gas prices, the labor market, and inflation all hit that customer disproportionately. Two things partly offset this, though. First, even under pressure, these customers are trading up into larger bundles rather than trading down (ordering larger sizes), so ticket has held steady even as people have cut back on ordering extras, etc. Second, management is deliberately focusing new customer acquisition on the $50K-$100K household income cohort, thereby reducing reliance on lower-income cohorts.

Read more

Each may have minor blemishes, but nothing that justifies the complete indifference they receive. Miss estimates outside the AI complex, and you get sold to valuations unseen in years. Beat them, and nobody cares. That disconnect is precisely where the opportunity lives. Mispricing tends to hide in the corners that the market has stopped bothering to look at.

The company we are looking at today controls north of 60% of a global market that grows at mid to high single-digits regardless of cycles, with gross margins above 70%, EBITDA margins in the mid-20s, and a recurring revenue tail that compounds for decades after the initial sale. Switching costs are effectively permanent.

New entrants need nearly a decade to clear regulatory hurdles, and demand is tied to secular demographic trends. And yet the stock is down close to 70% from its highs on a soft product launch and FX headwinds, temporary issues at worst. A rare combination of oligopoly economics, structural tailwinds, and a deeply mispriced entry point, hiding in plain sight.

Read more

The recovery since the lows has been swift. Aided in part by the better than expected earnings in the latest quarter. Since the early April bottom:

• IGV is up ~24% off the lows

• IGV logged its best week since October 2001 in mid-April as the panic broke

  

• Snowflake and ServiceNow have led the bounce, up ~46% and ~26% off their lows respectively, with Barron’s recently calling them the names pulling software ‘out of the abyss.’

• Q1 2026 prints have given the rally a fundamental backbone, not just a sentiment lift

Welcome to Rebound Capital. If you are new here, we conduct in-depth research on beaten-down stocks and study companies that have made successful comebacks. Subscribe for free and join 21,800+ other investors to make sure you don’t miss our next briefing.

Subscribe now

The single most important takeaway from this earnings cycle is that Software companies are monetizing AI. They are not being eaten by it. SaaS will and is adapting to the new reality.

The data points are hard to argue with:

• Microsoft: AI business run rate is now $37B (up 123% YoY)

• ServiceNow: Now Assist (its AI product) customers with Annual Contract Value (ACV) >$1M grew more than 130% YoY. Management is guiding >$1B in Now Assist ACV for 2026

• Snowflake: Closed FY26 with 9,100+ accounts on AI features, up from 7,300 just one quarter earlier. AI was a decisive factor in roughly half of the new bookings in H2 FY26.

• Salesforce: Agentforce ARR hit $800M just 15 months after launch (up 169% YoY), with 2.4 billion Agentic Work Units delivered and 29,000 deals closed.

• Cloudflare: Strongest quarter in years (revenue +34% and $1M+ deals up 73%), explicitly tied to a surge in agentic AI traffic.

• Toast: ARR of $2.2B (up 26% YoY), Adjusted EBITDA grew to $179M from $133M a year ago, full-year EBITDA guidance raised to $790 to $810M, and a fresh $500M buyback authorization.

The ‘agents will replace workflow software’ thesis is, in the field, playing out opposite of what the bears had predicted. Agents need workflow orchestration, audit trails, governance, and tool-calling rails. That’s exactly what Software platforms sell.

But here’s the catch. Not every cheap software stock is a buy.

What separates the survivors from the value traps in any major technology shift isn’t the multiple. It’s the intangibles. Brand. Customer trust. Regulatory standing. Embedded workflows. Network density. Proprietary data. Decades of installed base. None of this shows up on a P/E screen, and none of it can be conjured out of thin air by an AI startup. Competition in the industry will increase, but AI benefits the incumbents in this competition.

David Teece made this point back in 1986: innovators rarely capture the profits. The owners of the complementary assets do.

• EMI invented the CT scanner and lost to GE, which had the hospital sales and service network

• RC Cola was first to market with both canned and diet cola, then lost to Coke and Pepsi on distribution and brand

• IBM was late to the PC and won anyway, because it controlled the ecosystem around it

No 2 transitions are the same. Still, the above lens is helpful while analyzing SaaS. For Constellation, FICO, DoorDash, and ServiceNow, the code was never the moat. The workflows, the regulatory plumbing, the network density, and the decades of customer trust were. AI lowers software development costs for everyone. It cannot replicate any of the other moats.

One sober reminder. Roughly 78% of the US market cap is now exposed to AI disruption in some form, and a P/E screen will not distinguish survivors from casualties. The focus below is on names where the intangible moat does the heavy lifting, not on the valuation/multiple.

The selloff was indiscriminate. There can be an opportunity here. Let’s dig in.

Read more

Read more

We are publishing our first-ever 13F deep-dive without a paywall

One quirky aspect of the financial markets is that, by law, institutional investors must disclose their portfolios at the end of each quarter (with a 45-day lag). These public disclosures are called 13F filings.

This law applies to any fund managing over $100 million in assets. The government mandated this for transparency, and it gives us a look at what the best in the business are thinking about the markets.

Why track 13F filings

At Rebound Capital, we use these filings to get a sense of the market’s (institutional market’s) positioning and to track the best investors to see which companies they are investing in or divesting from. The aim is to understand both the bull and bear case thinking that leads these investors to bet on certain stocks.

As a naturally contrarian (value-oriented) investor, 13Fs help me make sure that:

• I am not brushing off any secular trend as a bubble (out of ego or ignorance)

• I am not being contrarian for the sake of it

Both of the above will lead to weak returns over time. Going through 50 13F filings forces me to face different perspectives on each stock I am researching. There’s always something to learn.

Welcome to Rebound Capital. If you are new here, we conduct in-depth research on beaten-down stocks and study companies that have made successful comebacks. Subscribe for free and join 21,800 other investors to make sure you don’t miss our next briefing.

Subscribe now

Elite funds spend millions of dollars on data, experts, and analysts to research a handful of companies. When they buy, they are committing serious capital with high conviction. Tracking their moves lets you go straight to evaluating their absolute best ideas.

The $220 Million Clue: Nvidia in 2023

In early 2023, Stanley Druckenmiller’s (one of the GOATs) filing revealed he had loaded up on $220 million of Nvidia. Nvidia was in a deep drawdown (~50%) when Druckenmiller took a position in the stock, and was still ~30% below its peak when the 13F filing was revealed.

Now, Mr. Druckenmiller is a macro and theme-based investor. What was he seeing in Nvidia? Especially when the company was facing a massive slump in its gaming chips? His 13F was a major clue to turn over the stones and study AI (the new use case for Nvidia’s GPUs).

If AI is as big as I think it is, Nvidia is something we’re going to want to own for at least two or three years, not for 10 months,” he said, adding, “And maybe longer.” — Druckenmiller in Jun’23

It wasn’t about blindly copying him. It was about letting a legendary investor offer you clues that we are early in a particular cycle.

Finding the next rebound

When a stock suffers a massive drawdown, it often stays beaten down and underperforms for a really long time. It can easily become dead money for months or years. We track the best value investors to see when they finally start showing interest in these names.

At the same time, we follow many specialized smallcap investors to identify their highest-conviction ideas in mid and smallcap companies. These managers focus on beaten-down small caps, for example, in the software space currently.

By combining these approaches, we can spot the best rebound opportunities.

Quality Investors we track

We currently follow ~50 investment funds (we keep adding). This is a mix of value and growth investors. Some of the ones we admire the most are:

• Berkshire (we have new management at the helm, but Warren is still consulted)

• Stanley Druckenmiller (best macro investor of the 21st century)

• David Tepper - Appaloosa Management (predicted the bottom of the financial crisis)

• Coatue Management (high-growth tech and consumer companies)

• Li Lu (value investor who Charlie Munger gave his personal money to invest)

• Atreides Management (one of the smartest technology investors)

• Baillie Gifford (long-term investors in quality companies)

• Ratan Capital (research-intensive fund focused on misunderstood businesses)

• Alyeska (deep fundamental research on both sides of the trade)

What was sold in the March 2026 quarter

A clear theme this quarter is that smart money continues to reduce exposure to any stocks with potential AI disruption. The selling clustered in horizontal SaaS and select megacap tech, with high valuations or capex-intensity concerns.

Software

Hedge funds reportedly made about $24B shorting the sector through early 2026, and the 13Fs confirm that large funds are also reducing their long positions. CRM, NOW, ADBE all show meaningful trims across our tracked list. Sands Capital nearly exited its previously significant ServiceNow stake. The fear is straightforward. Per-seat payment based contracts will change to usage-based agentic models, forcing incumbents into messy transitions.

Atreides, Gavin Baker’s fund (a technology investor we admire), exited HubSpot and Intuit entirely, sold most of its Snowflake stake, and cut its GitLab stake by another 23% in Q1. This is a fund that knows software cold. In fact, despite the sharp drop, very few funds bought ServiceNow and Salesforce (the 2 bellwether SaaS firms) with high conviction.

Our Take: I agree that most software firms will undergo a massive change in the coming years, and hence, investing in them is difficult. Valuations are not cheap either. We will stick to vertical SaaS names for Rebound Portfolio (Constellation Software is our top pick).

UNH

Berkshire under Greg Abel exited fully. This was mostly a sweeping rotation by new management rather than a UNH-specific call. That said, the broader institutional read is negative. Most quality funds trimmed, very few stepped in to buy during UNH’s drawdown in Q1.

Our Take: We agree with the consensus and will sell out our remaining 2% position soon (currently at a ~30% gain in 9 months).

Microsoft and Alphabet

TCI cut Microsoft from 16.78M shares to 2.73M, an $8B reduction, citing AI disruption risk in its investor letter. Appaloosa went even harder, from 500K shares to 90K. Druckenmiller exited his entire Alphabet position, one he had only built up in Q3 and Q4 2025.

But the same names are being aggressively bought elsewhere. Berkshire tripled its Alphabet stake. Third Point opened new positions in Alphabet and Meta. TCI itself opened a new 2.46M-share Alphabet position while exiting its Microsoft position.

Institutional investors are divided on whether the hyperscalers are AI winners or losers (due to low Capex ROI).

What was bought in Q1 2026

Funds were buying the following themes:

• AI adjacent infrastructure, where pricing power is strong (Micron, ASML, the optical stack).

• Quality compounders trading at reasonable multiples (SPGI, MELI, MCO).

• Beaten down vertical SaaS with embedded data moats (Toast, Axon).

• A basket of niche small caps with high conviction by small-cap focused funds.

Toast (TOST)

Toast is the operating system for North American restaurants. Its major business segments are: Point of sale, payroll, payments, and a growing financial services business (Toast Capital, lending, insurance). The moat is data and operational lock-in. Every order, every shift, every supplier payment runs through the platform, and switching out is brutal. The stock is down about 45% from its highs after a Q1 revenue miss and growth decelerating to 25-26%.

Why it is interesting: net location adds were 7,000 in Q1, the highest in three quarters, and Toast Capital is starting to flow through to ARPU.

Fund Ownership:

• ValueAct: 6% of portfolio (+61% QoQ)

• Alta Fox: 5% (new position)

• Shannon River Fund: 4% (new position)

S&P Global (SPGI)

SPGI has the closest thing to a regulatory moat in finance. It has a regulatory duopoly in the US over bond ratings (with Moody’s). Platts owns commodity price benchmarks. Capital IQ is entrenched in every finance workflow. The stock has lagged peers, weighed down by macro concerns over issuance volumes and AI disruption fears in its analytics businesses.

Why it is interesting: SPGI is the perfect defensive stock to hold to diversify away from the frothy AI theme. Multiple legendary investors have upped their stakes in SPGI (see below).

Fund Ownership:

• TCI Fund Management: 13.2% (+20% QoQ)

• Himalaya Capital (Li Lu): ~2% (new position)

We had initiated a 4% position in S&P Global after our deep dive two months ago.

MercadoLibre (MELI)

MELI is the Amazon and PayPal of Latin America. The e-commerce marketplace is growing robustly, and the Mercado Pago fintech is still in its infancy, with credit penetration in Brazil and Mexico still in its early innings. The moat is the density of merchants, payments data, and credit underwriting in markets with historically low banking penetration.

Read our deep dive on MELI here.

Why it is interesting: MELI will lead South America’s e-commerce industry for many years to come. It is a long-only favorite, and funds held onto their stock in the latest quarter.

Fund Ownership:

• Durable Capital (Ellenbogen): 6%

• Baillie Gifford: 6% (held constant QoQ)

• Tiger Global: ~1% (new position)

KKR

KKR is one of the four major alternative asset management firms. Its revenue sources are fee-related earnings from PE, credit, and infrastructure, as well as an aggressively growing insurance permanent capital base. The moat is the LP brand and the scale advantage in deal sourcing and credit origination.

Fund Ownership:

• ValueAct: ~5% (new position)

ASML

ASML is the only manufacturer of EUV lithography systems. No Nvidia, no TSMC leading edge, no Apple silicon without ASML. A clean monopoly on the most strategic capital equipment on the planet. The stock corrected on China export rules and a lull in front-end semicap orders.

Why it is interesting: a top-tier technology fund (Coatue) opened a brand new ASML position in Q1 at roughly 50x earnings. A separate, ultra-concentrated long-term fund, which holds names for years (Valley Forge), was also added. ASML rarely gets cheap, and these are not the funds that show up for a trade.

Fund Ownership:

• Coatue Management: ~2% (new position)

• Valley Forge Capital (Kantesaria): ~7% position (up ~35% QoQ)

The small-cap names worth digging into

This is where the 13Fs get genuinely interesting. The small and midcap space. A 5% position at a $500M specialist fund is real conviction.

XPEL

XPEL dominates the paint protection film oligopoly. The moat is the DAP software platform sitting inside every installer shop (workflow lockin) plus the brand premium with luxury and performance dealerships. US PPF penetration is still mid single digits, so the runway is wide. After a heavy 2022 to 2025 investment cycle, EBIT margins are guided from 13% today to 26% by FY28 through vertical integration. The stock has been crushed by short reports and macro fears.

Fund Ownership:

• Alta Fox Capital: 22.5% (largest position)

ACV Auctions (ACVA) and Wingstop (WING)

ACV runs the digital wholesale auto auction marketplace where dealers buy and sell trade ins. Has been taking share from Manheim and ADESA by going digital first. The moat is the two sided network plus proprietary AVM data on every vehicle. Beaten down on the auto cycle weakness.

Wingstop is a franchised QSR chain with AUVs above $2M and an asset light royalty model running at 25% operating margins. International expansion is just getting going. Both are the kind of high quality compounders that quietly drift lower in tough tape and re-rate fast when sentiment turns.

Fund Ownership:

• ACV Auctions: Atreides Management ~$100M (+125% QoQ in Q4)

• Wingstop: Atreides Management NEW (Q4 2025)

Worth keeping on the radar

A few more that fit the same pattern. Tencent Music (TME), where Li Lu opened his first new China internet position in years, signaling he sees the regulatory cycle past peak risk. Rocket Companies (RKT), where Atreides added 260% in Q4 on a mortgage refi thesis. And DNOW, Klarman’s rare new buy at Baupost, a small cap energy distributor with deep cyclical leverage.

Our Take

Quality investors are doing four things at once this quarter. Cutting AI disruption candidates. Leaning into AI infrastructure where pricing power is real. Stepping into beaten-down compounders at de-rated multiples. And bottom-fishing in small caps, where specialists are sizing up with conviction.

The megacap tension is genuinely interesting (TCI versus everyone on Microsoft, Berkshire versus Tepper on Amazon), but the trade we are most excited to research is:

• The small cap basket. XPEL, NCR Atleos, and ACV Auctions. These are the names where a sub $1B specialist fund taking a 5 to 20% position is real conviction, and the multiples already reflect years of pain.

• We are also interested in high-quality, non-semiconductor stocks that the market is selling off, like Toast, KKR, SPGI, and MELI.

We will be writing up two of these as full rebound deep dives over the next month.

Leave a comment

Rebound Capital’s work is provided for informational purposes only and should not be construed as legal, business, investment, or tax advice. You should always do your own research