Mastercard first, because it’s the least dramatic one.

$5.04 adjusted EPS against $4.77 expected. Revenue $9.28B against roughly $9.06B. Stock up about 2.7% the next morning. Nothing here should surprise anyone who’s followed this stock for more than a year, and that’s kind of the point.

What I actually paid attention to was the value-added services line, which grew 18% currency-neutral vs. 8% for the core network. People have been saying “Mastercard is more than a payments network” for years and usually the numbers don’t back it up as cleanly as they did this quarter. Operating margin went to 61.1% from 59.9%. Cross-border volume up 12%. All fine, all expected, all evidence the diversification story is doing what it’s supposed to do rather than just being a slide in an investor deck.

Two things worth a second look though. Buybacks added 14 cents to EPS this quarter not nothing, and it means part of the “beat” is capital allocation, not operations getting better. And the litigation charges keep showing up. $82M this quarter for ATM surcharge stuff, $151M same quarter last year, over $500M for all of 2025. They pull it out of adjusted numbers every time, which is standard practice, but four quarters running is not really a “one-time” charge anymore, it’s a cost of doing business that happens to get treated as an exception. I don’t think it changes anything about owning the stock. I do think it’s the kind of thing that gets glossed over if you only look at the headline adjusted EPS.

Guidance ticked up slightly for the year. Nothing here makes me want to do anything differently.

American Express is the one that actually annoyed me, in a good way.

Beat on EPS ($4.53 vs $4.40), missed slightly on revenue ($19.64B vs $19.69B), raised full-year revenue guidance to 10%, held EPS guidance exactly where it was. Stock dropped 5-6% the next morning.

That’s a beat. The stock fell 5-6% on a beat.

Card spending grew 9% FX-adjusted, the best in three years. Credit provisions actually fell, from $1.4B to $1.1B, and the write-off rate held at 2.0% flat. Platinum is the fastest growing product in their US consumer business right now. None of that sounds like a company that should get sold off hard the next morning.

So what happened? Expenses grew 12%, ahead of the 10% revenue growth, partly because of the Platinum refresh, which is a real investment, not waste. And management raised revenue guidance but explicitly said the extra money is getting reinvested into growth rather than flowing to the bottom line. That’s the actual sentence that moved the stock not the numbers in the table, the sentence on the call. Investors wanted proof that stronger revenue converts to stronger profit and instead got told “we’re going to go spend it.”

I think the market overreacted here. The tax rate jump (18.7% to 23.6%) is mostly a prior-year comparison quirk, not something structural. Everything that actually measures the health of the business — spend, credit quality, product mix — looks fine, arguably better than fine. The thing to actually watch is whether that expense growth rate comes back down toward the revenue growth rate next quarter. If it does, this selloff will look like an overreaction in hindsight.

Vusion is the one I spent the most time on, mostly because of what it doesn’t tell you yet.

Quick note before the numbers: this isn’t a full earnings report. French companies report full financials (margins, EBITDA, the actual profitability picture) twice a year, with revenue-only updates in between. Vusion’s full H1 numbers land September 21st. So everything below is about growth and momentum, not profitability, because profitability isn’t disclosed here at all.

Adjusted revenue for H1: €839.3M, up 29.3%, or 37% at constant currency. Stock up 4.4% on the release.

The number that actually matters here, more than the top-line growth, is the installed base. VusionCloud crossed 522 million connected shelf labels, up from about 220 million a year ago. That’s more than doubling. And recurring VAS revenue (the actual SaaS-like piece of the business) hit €61M, up 73%, with an annualized Q2 run-rate of €133M, up 83% year over year. If you’re trying to figure out whether this company is turning into a recurring-revenue software business riding on top of hardware, that’s the number that says yes.

Order intake for the half was down 22% year over year though, which sounds bad until you realize it’s being compared against an unusually large Walmart order from last year. Q2 orders alone were up 7% year over year and 16% sequentially against Q1. Reads more like a comp issue working itself out than actual demand weakening, but it’s worth checking again next quarter before assuming that.

Here’s the part I kept rereading. There’s a footnote about a US tariff refund — roughly $60M received in July, another $20M expected later this year — tied to a Supreme Court ruling that invalidated certain customs duties. That money gets credited back to customers under existing contracts, and it will retroactively lower both revenue and cost of goods in the 2026 accounts. No effect on margin, no effect on cash. But it means the headline revenue growth number could come in under the 15-20% guidance range for reasons that have literally nothing to do with the business slowing down. If that shows up later this year, this is the footnote to go find before assuming something broke.

And the thing I actually worry about with this one: they just agreed to finance the In-Store Media acquisition (around €120M in 2025 revenue) with debt, which isn’t something this company has leaned on much before, at exactly the moment we’re all being asked to trust that the margin story is real without a single EBITDA number to check it against. Two months from now we’ll know whether that trust was earned. Right now it’s a bet on a story that sounds convincing on paper.

Net: thesis getting stronger, not yet proven. The recurring revenue growth and the doubled label base are genuinely good signs. The profitability half of the argument is still unwritten, and September 21st is the date that actually settles it.

The connecting thread, if there is one

Mastercard and Amex are riding the same spending trend and both beat. The difference in stock reaction wasn’t about the quality of the quarter, it was about what each company chose to do with the upside, Mastercard let it flow into margin and guidance, Amex chose to spend it on growth instead. Vusion is a completely different kind of business, but the underlying idea is the same one that’s been working across retail and payments infrastructure for a while now: whoever owns the layer underneath the transaction, the network or the shelf label, keeps compounding while everyone argues about the headline print.

Let’s start with the numbers, because they are genuinely extraordinary.

In 2025, the S&P 500 returned 17.9%. Three consecutive years of double-digit gains. Impressive! Comforting! The kind of returns that make financial advisors look wise and index fund enthusiasts insufferable at dinner parties. But here’s what that headline number hides: just seven stocks generated 52% of that return, and a single chipmaker named Nvidia accounted for 15.5 percentage points of the full-year gain all by itself.

One company. One-sixth of the gains across 500.

I’ve spent more time than I’d like staring at these contribution charts, and every time I come back to the same thought: this is not how an index is supposed to work. The whole point of a 500-stock index is diversification of outcomes. What we have now is closer to a 7-stock index with 493 passengers along for the ride.

The tech and communications sectors alone, the home of Nvidia, Alphabet, Microsoft and their AI-adjacent neighbors, were responsible for 63.1% of the S&P 500’s total return in 2025. Strip them out and the rest of the index returned around 6% for the year. Not bad! But also: not what you were celebrating in January.

A market-cap-weighted index is not what people think it is

Most people who invest in the S&P 500 think they’re buying “the American economy.” And in some broad, impressionistic sense, they are. But the index isn’t 500 equal slices of America Inc. It’s weighted by market capitalization, which means the biggest companies have the most influence over the number you watch on CNBC.

As of early June 2026, the Magnificent Seven, (Apple, Nvidia, Microsoft, Alphabet, Amazon, Meta, Tesla), control 33.8% of the index. The top 10 companies, which add in JP Morgan, Broadcom, and Berkshire Hathaway, make up nearly 40% of the whole thing. The remaining 490 companies share the other 60%.

Now here’s the thing about that math. Nvidia alone has more influence over the S&P 500 than Clorox, Nordstrom, and approximately 350 other companies combined. It’s not manipulated or rigged. It’s just how market-cap weighting works when one company becomes extraordinarily large. Amazon’s market cap in April 2026 sat at $2.2 trillion, making it 3.8% of the entire index all by itself. These aren’t outliers distorting the system. They are the system.

That's a journey from 22% in 2020 to nearly 40% today. In six years. To put it another way: in 2020, buying the S&P 500 gave you roughly equal-ish exposure to the full economy. Today, it gives you a large-cap tech fund with some healthcare and bank stocks mixed in.

Why your stocks are flat, and why that’s the honest result

And yet. The average investor looking at their portfolio in 2026 isn’t seeing 12% gains. They’re seeing something like flat. Maybe mildly positive. Maybe some red. And they’re confused, because the headlines keep saying things are fine.

Here’s the explanation, and it’s almost insultingly simple.

If you own a diversified portfolio of stocks that isn’t extremely overweight on a handful of AI names, you own something close to the “equal-weighted” version of the market. The S&P 500 Equal Weight Index, where every stock gets the same 0.2% slice regardless of market cap, has actually outpaced the standard S&P 500 by roughly 5 percentage points so far in 2026. That gap is being driven by the largest tech names dragging down the cap-weighted index this year, after years of leading it higher.

Your portfolio of “regular” stocks, the industrials, mid-caps, healthcare names, financial companies that don’t yet have an AI story, behaved a lot like the equal-weighted index in 2023, 2024, and 2025. Which means it lagged the cap-weighted S&P 500 by a wide margin, because the cap-weighted index was essentially Nvidia with a hat on.

In 2023, only 27% of S&P 500 stocks outperformed the index. In 2024, 28%. That's the second narrowest market in nearly three decades, the first being 1999.

Let that land for a second. Two out of three stocks in the S&P 500 underperformed their own index in 2023. Nearly the same in 2024. One of those numbers is wrong, and I don’t think it’s the breadth figure. The index has simply been doing something that a portfolio of, say, 20 to 30 reasonably chosen stocks almost can’t replicate unless those stocks happen to be Nvidia, Alphabet, Meta, Microsoft, and a few friends.

This is not your fund manager failing. This is not stock selection going wrong. This is math.

The gap in 2023 was staggering. The cap-weighted index returned 26.3% while the equal-weighted version returned around 13.8%. A nearly 13-point spread. If you owned a portfolio that “felt” well-diversified, you were in the equal-weight world, not the cap-weight world. You didn’t lose money. You just didn’t happen to own the one thing that mattered.

The AI capex cycle is real, and it is doing the work

Now. Let’s be honest about why this happened, because “AI hype” is too easy and also not quite right.

The Magnificent Seven’s 2025 earnings grew more than twice as fast as the rest of the S&P 500. Forward estimates for 2026 show the same trajectory. This isn’t just narrative: the capital expenditure numbers are extraordinary. Microsoft, Alphabet, Meta, and Amazon collectively spent hundreds of billions on AI data center infrastructure in 2025, funding an entire ecosystem of chip suppliers, power companies, construction firms, and cooling systems.

Nvidia’s Q4 2025 revenue guidance sat at $65.7 billion. That’s a 67% year-over-year increase. Strip Nvidia out of the information technology sector and earnings growth falls from headline figures to something far more modest. One company is genuinely, measurably driving a huge portion of US corporate earnings growth right now.

That’s a useful data point. Because it means the concentration isn’t just about sentiment, and it isn’t just about multiple expansion. Most of 2025’s S&P 500 return was actually driven by real earnings growth. Over 75% of the 17.9% gain came from EPS growth rather than from investors simply paying more for the same earnings. Which is what made 1999 dangerous. Then, P/E ratios were expanding into the stratosphere on companies with no earnings at all. The Mag 7 today trade at roughly 28x forward earnings on average, elevated but nothing like the 66x multiple the largest stocks commanded at the 2000 peak.

Fidelity made this comparison explicitly. The current situation is fundamentally different from dot-com, they argued, because today’s AI leaders have real cash flow, real balance sheets, and real products with real enterprise demand. This is a legitimate argument. It doesn’t mean nothing can go wrong. It means the floor is probably higher than the skeptics think.

The risk that actually keeps me up at night (and it’s not the one you expect)

The obvious risk is the bubble risk. Everyone mentions it. The S&P 500 concentration chart looks eerily similar to the late 1990s. The equal-weight versus cap-weight spread has reached extremes not seen since 1999. This comparison gets made a lot in financial media, and it’s not wrong to make it.

I’ll be honest: I find the bubble argument less compelling than I did two years ago. Nvidia generates actual profits. Alphabet’s AI search integration is working. Meta’s ad business is humming. These companies aren’t burning through venture capital on unproven business models. They’re buying GPUs to serve paying enterprise customers who have signed contracts. When the dot-com bubble burst, many of the largest companies had no earnings path at all. That is not true today.

The risk that I keep coming back to, and that I think deserves far more attention than it gets in the standard “risks to the thesis” paragraph, is the rotation risk. Not whether the AI bubble bursts, but what happens if the AI trade simply moderates.

Here’s why this matters more than it sounds. If Nvidia grows “only” 25% next year instead of 67%, and if the Mag 7 go from growing earnings at twice the rate of the S&P to merely matching it, the concentration dynamic reverses without any crash. No drama. The index just starts returning less than the average stock instead of more. And every investor who spent the last three years being told they were “underperforming the market” discovers that, actually, their boring diversified portfolio was fine the whole time.

This rotation is already happening in pockets. The S&P 500 Equal Weight Index outperformed the cap-weighted index by 3% in the three months ending January 2026. The small-cap S&P 600 outperformed the full S&P 500 by 4% in a single month. The breadth indicators that analysts use to measure whether the rally is broad or narrow have been improving. Sectors that had nothing to do with AI, utilities, financials, industrials, started showing up in 2025 returns.

Now here’s what makes rotation genuinely dangerous for certain investors. If you’ve been chasing the AI trade by increasing your weight in mega-cap tech to “keep up with the index,” you’ve done exactly the wrong thing at exactly the wrong time. You bought concentration at the moment when concentration is most likely to mean-revert. The people who did nothing, who held their diversified portfolios and shrugged at underperformance, may end up being the ones who look smart by 2028. I’m not predicting a crash. I’m describing a dynamic where patience is the entire position.

There’s also the earnings revision risk, which is quieter but worth watching. The top ten S&P 500 companies represent 40% of the index’s market value but only 31% of total earnings. That gap has to close somehow, either through earnings growing into valuations or through valuations compressing and historically, gaps like this close more often through the latter.

The one genuinely reassuring data point in that chart is the economic profit figure. Russell Investments estimates that the Mag 7 generate nearly 70% of the total economic profit produced by S&P 500 companies, meaning returns above their cost of capital. On that measure, their weight in the index looks almost cheap relative to their contribution. The market is paying a 33.8% weight for companies producing 70% of the real value creation. That’s a different story than 1999, where index weight was entirely decoupled from any reasonable profit metric.

What this means if your portfolio is flat

Simply: it means your portfolio behaved like most stocks behave. Not like a portfolio that happened to be 30% Nvidia.

The S&P 500 you see on Bloomberg or Yahoo Finance is not the average stock. It is the market-cap-weighted average, which means it is, right now, a heavily AI-tilted large-cap technology index wearing the costume of a broad market benchmark. Comparing your returns to it is like comparing your jogging time to someone who took an electric scooter.

This doesn’t mean you should run out and buy Nvidia. The easy money in that trade was made in 2023 and 2024. What it does mean is that you shouldn’t misread your portfolio’s recent underperformance as evidence that you’ve done something wrong. There are about 370 stocks in the S&P 500 right now that have underperformed the index in 2025. If you own any of them, you’re in very crowded and very reasonable company.

The professionals aren’t immune either. In 2023, just 27% of individual S&P 500 members outperformed the index itself. That’s the narrowest market breadth since at least 1995. Any active manager who didn’t happen to be very overweight the right five or six names looked like they were failing at their job. Most of them weren’t failing. They were just not running a Nvidia index fund with extra steps.

The rotation: slow, painful, and probably already starting

By early 2026, something has quietly shifted.

The Mag 7 as a group have underperformed the S&P 500 so far this year, growing 5.4% collectively through mid-June versus the index’s 7.9%. Within the group there’s massive dispersion: Alphabet is still running, while Microsoft and Tesla have dragged. The concentrated leadership is fragmenting. Amazon was up 3.4% year-to-date as of late May. Alphabet was up 8.5%. Meta up 4.8%. But Apple, Microsoft, and Tesla were lagging.

The equal-weight ETF (RSP) has outpaced SPY by nearly 5 percentage points in 2026. The small-cap S&P 600 has been on a tear. The Wall Street Journal ran a piece in January headlined “Small Stocks Finish Week at Records.” Breadth is expanding. The concentration in the top 10 has dipped slightly below 40% from its peak.

This is what a slow rotation looks like. It doesn’t feel like a crash. It feels like: “Hm, my portfolio is actually doing fine relative to the index this year.” And then you realize the comparison finally makes sense again.

Goldman Sachs published its 2026 outlook predicting that the Mag 7 would trail the equal-weight S&P 500, framing it as a shift toward “broader market opportunities.” Ed Yardeni went on CNBC and told investors to be underweight the Mag 7 and overweight the “Impressive 493.” These aren’t crank calls. They’re mainstream reads from credible analysts who see the same mean-reversion math that the charts have been whispering about for two years.

The chart above is the one I find most instructive. The Mag 7’s gigantic outperformance in 2023 was basically a catch-up from the ugly 2022. After periods of extreme concentration, history suggests mean-reversion. The only prior period where the 3-year cap-weight versus equal-weight spread reached this level was late 1999 and early 2000. What followed wasn’t necessarily a crash in the leaders. It was a decade of the rest catching up.

The question you should actually be asking

Not “why is my portfolio underperforming?” but “what am I actually benchmarking against, and does that benchmark make sense for me?”

The S&P 500 is the right benchmark if you own an S&P 500 index fund. It is a deeply weird benchmark if you own a diversified portfolio of individual stocks or sector funds or anything that doesn’t replicate the index’s exact composition. For most individual investors, the equal-weighted S&P 500 is a much more honest comparison. And against that comparison, the last three years look a lot less alarming.

The practical portfolio question is whether to chase the AI concentration or wait it out. A lot of smart money managers are doing what Benzinga described: “keeping a core position in a low-cost cap-weighted fund for benchmark tracking, while adding equal-weight exposure to dilute concentration risk.” That’s a hedged, sensible position. It acknowledges both that the AI trade has legs and that the concentration is historically extreme.

What this looks like from inside a portfolio with zero AI stocks

Let me be specific. My portfolio holds no Nvidia, no Microsoft, no Alphabet, no Meta. What it does hold is a cluster of financial services companies: American Express, S&P Global, Mastercard, Brookfield Asset Management. The kind of names that show up on value investor reading lists, not on AI bubble watchlists. Businesses with real moats, steady earnings growth, and the slightly boring quality of being genuinely good at what they do.

In 2026, this portfolio is lagging the S&P 500. Not dramatically. Not in a way that should cause anyone to lose sleep. But lagging.

Now here’s the thing about that. I’m completely fine with it.

Not because I’m stoic or because I’ve somehow transcended the very human urge to compare. I’m fine with it because I’ve looked at what these companies actually do, how they’ve actually performed over the long arc, and what the math looks like when you zoom out past the current AI moment. The short-term gap between my portfolio and the S&P 500 headline number is real. It’s also, I think, close to meaningless as a signal about whether I’ve made good decisions.

Take Mastercard. Over the past ten years, MA has delivered an annualized return of around 18.7%, compared to roughly 13.3% for the S&P 500. That’s five percentage points per year of compounding outperformance over a decade, the kind of gap that turns a $100,000 investment into very different outcomes depending on which side of it you’re on. A thousand dollars invested in MA stock twenty years ago would be worth close to $121,000 today, good for an annualized return of more than 27%, versus about $8,600 from an S&P 500 index fund.

The catch, which any honest Mastercard observer will tell you, is that 2026 has been rough. Mastercard is down roughly 13% year-to-date. Berkshire Hathaway, which held both Mastercard and Visa since 2011, sold its stakes in both during Q1 2026. Make of that what you will. But the two-decade return is not a fluke. It’s what a near-monopoly on the plumbing of global consumer spending produces over time.

American Express is the portfolio’s clearest example of what patient ownership looks like in practice. The stock returned 60% in 2024 and has delivered 15%+ EPS growth annually. Q1 2026 results showed revenue up 12% and net income up 15% year-over-year, with EPS beating analyst estimates by over 7%. AXP’s consensus EPS estimates for 2026 and 2027 project 13.9% and 14.7% growth, respectively, with four upward revisions to estimates in recent months. This is a company that has found a way to keep growing its premium cardholder base while the broader consumer shows stress. The Centurion card network isn’t being disrupted by AI. Neither is the habit of well-off people spending money on travel and dining.

S&P Global is a slightly different story. It’s had a weak year. SPGI’s 10-year total return sits at 387%, against a far lower figure for the index, and its 20-year annualized return is 12.8%. But 2025 was only 5.7% and 2026 has been rougher. The company is spinning off its Mobility division, navigating macro headwinds, and dealing with the fact that bond issuance slowed. None of these are existential threats to a company that literally licenses the benchmarks every index fund in the world is measured against. You can argue about their short-term earnings trajectory. You cannot really argue that their product is going away.

Brookfield Corporation, the newest and most interesting addition, manages over a trillion dollars in real assets: infrastructure, renewable power, private credit, real estate. BAM has around $539 billion in fee-bearing capital and is expected to grow earnings 15 to 18% in 2025, with S&P rating it A-minus with a stable outlook. The business model is essentially a toll on global capital flows through alternative assets, which is a very good place to be when rates are elevated and institutional investors are desperately looking for yield outside public equities.

The common thread through all of these names is that they are, in different ways, infrastructure businesses. Not infrastructure in the literal pipes-and-cables sense, although Brookfield owns plenty of those. Infrastructure in the sense of being embedded in processes that don’t stop. Every time a payment moves through a card network, Mastercard gets paid. Every time a bond gets issued, S&P Global rates it. Every time an institution allocates capital to alternatives, Brookfield charges a fee. These aren’t exciting narratives. They’re not getting anyone on a fintech podcast. But they compound.

And yet. The honest version of owning these stocks in 2023, 2024, and early 2025 is that you watched the S&P 500 sprint past you at a pace that felt almost personal. When the index is up 25% and your portfolio is up 12%, you don’t feel like a patient long-term investor. You feel like someone who bet on the tortoise and is watching the hare disappear over the horizon.

The answer to that feeling isn’t to panic. It’s to remember what the 10-year numbers actually look like on these businesses, and to remember that the hare spent 2022 getting destroyed while the tortoise merely had a bad year. Over a full market cycle, quality compounders with durable moats and consistent earnings growth tend to close the gap. The S&P 500’s recent outperformance was heavily distorted by a small handful of names in a very specific moment. That moment will pass. These businesses will still be collecting their tolls.

The five-year game is the one I’m playing. Not the six-month game. Not the “why isn’t this keeping up with Nvidia” game, which is a game you can only win by owning Nvidia, and which you almost certainly would have lost at some earlier point by owning it at the wrong time.

Underperforming the S&P 500 in a given year, when that index is a de facto Nvidia fund with extras, is not a judgment on the quality of the underlying businesses. It’s just a statement about what the market chose to reward in that specific window. The market can be wrong about that for longer than feels comfortable. But it corrects.

Where I land

I’m long the broader market, and specifically long the idea that the next three to five years look better for the “Impressive 493” than the last three did.

Not because AI is going away. The data center build-out is multi-year. Nvidia’s demand is still “off the charts” per Jensen Huang. The hyperscalers are spending at a rate that has no historical precedent. The AI earnings cycle is real and durable.

But the easy outperformance trade in the Mag 7 required not just that these companies do well, it required that they do well while everything else treaded water. That dynamic is harder to maintain when the equal-weight index starts beating the cap-weight one. When small caps are breaking to records. When the breadth is expanding and the dispersion inside the Mag 7 itself has become enormous.

The most boring version of the next few years is that AI stocks deliver okay returns, the rest of the market delivers okay returns, and the gap between them shrinks. Your diversified portfolio stops looking like it’s broken. The S&P 500 headline number becomes a slightly less misleading representation of what you actually own.

That’s probably the outcome. Not a crash. Not a rotation into chaos. Just: mean reversion. Which, in markets, is usually the most likely thing and somehow always the most surprising.

Subscribe now

If this article resonated with you, you might like Investing with Eagles.

The portfolio I described above, financial services companies with durable moats, patient long-term ownership, ignoring the noise, didn’t come from gut instinct. It came from a framework. My book Investing with Eagles walks through exactly how I analyse companies, read financial statements without falling asleep, identify genuine competitive advantages, and estimate what a business is actually worth. It’s a value investing book, but not a stuffy one. Think less accounting textbook, more honest conversation about how to find great businesses and hold them long enough to let compounding do its job. If you’ve read this far and nodded along at the part about owning quality names and waiting out the Nvidia years, this is probably the book for you.

Get my book

Kinsale and APO are genuinely good businesses. Kinsale is a focused E&S insurance underwriter with a phenomenal track record; Apollo is one of the sharpest alternative asset managers on the planet. But both were being underwritten on the same basic assumption: 15–20% EPS CAGR over the next five years. And when I stress-tested that assumption, cracks appeared. Kinsale needs to avoid a bad underwriting cycle. Apollo needs credit markets to stay cooperative. These aren’t fatal flaws, they’re just the nature of businesses where the moat is more about execution than about structure.

On top of that, I already hold Brookfield Corporation which, if you squint, is playing a very similar game to Apollo: alternative assets, private credit, fee-earning AUM growth. Owning both was redundant. I was essentially doubling down on the same thesis, in the same sector, with correlated risk, without realizing it. That’s the kind of portfolio overlap that feels fine until the credit cycle turns and both positions move the same way at the same time.

So the question became: if I’m going to free up capital anyway, where does it go? And when I looked across the portfolio and asked where Mastercard fit by comparison, the answer was uncomfortable, I was taking on more execution risk and more sector concentration for the same expected return. Capital allocation isn’t just about picking good companies, it’s about ranking them. If you can own the global highway toll booth for roughly the same projected EPS growth as a regional insurer and a credit-cycle-dependent asset manager you’re already partially exposed to through BN, the honest move is to consolidate into the one with the near-indestructible structural advantage. Moat quality isn’t even in the same conversation, and over a 5-year hold, moat quality is basically everything.

With Vusion exploding in the last weeks we’re stile behind, but Portfolio is behind the S&P 500 only by 3,4% since the beginning of the year compared to -8% at the previous update. I still believe we’re going to underperform this year, SPGI is consolidated the business and MA is a new add. Earnings season will be key here. I’m investing for the long term and I know some years I’m gonna underperform, and this is one of those years.

1. The Machine

You already know what Mastercard does, so I’ll skip the basics and go straight to the investment case.

The moat here is a three-headed animal network effects (accepted everywhere because everyone has one), switching costs (banks embedded in the rails don’t leave), and regulatory licensing (operating permissions in virtually every country on earth, which took 60 years to accumulate). None of those erode quickly. None of them can be replicated with capital alone.

What makes it special as a compounder is the asset-light model. No factories, no inventory, minimal physical infrastructure. Return on invested capital runs above 30–40%, and the company reinvests some of that at high rates into new geographies, fraud AI, and tokenization, then returns the rest through aggressive buybacks. It compounds and cannibalizes simultaneously a rare combination.

The risks worth taking seriously: regulatory pressure on interchange fees is structural and never fully goes away. The stablecoin question is real over a 10-year horizon, even if it’s noise over 5. And the CFO change, while internal, adds short-term uncertainty during a period when Mastercard is actively pivoting its capital allocation toward digital assets. None of these are thesis-breakers. All of them are worth monitoring.

2. The Numbers

Current Valuation

Price at time of writing is around $483

Profitability Snapshot

Revenue (TTM): ~$33.9B, growing ~16% year-over-year

Net income (TTM): ~$15.9B

Operating margin: ~58%, one of the highest in the S&P 500, and still slowly expanding

2026E EPS: $19.60 (consensus)

Valuation Metrics

P/E (on 2026E EPS): ~24.6x

Historical P/E range: Low ~24x (2016) - High ~55x (2020)

10Y avg ~37.5x

5Y avg ~36.5x

Those were in the low interest rates world, and I do not believe we’re going to see MA at 37x in the near/medium term.

Earnings yield: ~4.1% (= $19.60 / $483) : for the first time in a long time, this is actually competitive with what you’re comparing it to:

10Y US Treasury: ~4.4% → minimal premium for owning a business growing 15%+ annually. That gap should not exist.

S&P 500 earnings yield: ~4.1% → Mastercard is trading at essentially the same earnings yield as the broader index, despite being a structurally superior business with a higher growth rate. That’s the anomaly.

Shareholder Returns

In 2024, Mastercard returned $13.4B to shareholders, $11B in buybacks, $2.4B in dividends, against $14.8B in operating cash flows. Essentially the entire free cash flow handed back to owners.

Dividend yield: ~0.6%, Buyback yield: ~2.5%

Share count shrinks roughly 2% per year from buybacks, a silent EPS booster most people underweight.

3. The Napkin Math

A. EPS Growth

Revenue growth: ~13% annually (conservative vs. recent 16% prints)

Margin expansion: ~0.5% per year

Share count reduction: ~2% per year (buybacks)

Total EPS growth estimate: ~14–15% annually

B. Valuation Impact

Current forward P/E: ~24.6x | 10-year avg: ~37.5x | 5-year avg: ~36.5x

Scenario 1 : Partial reversion to 30x: (30/24.6)^(1/5) − 1 = +4.1% per year tailwind

Scenario 2 : Multiple stays flat: 0% impact

Scenario 3 : Compresses to 21x: (21/24.6)^(1/5) − 1 = −3% per year drag

D. The Final Equation

Bull case: 15% EPS growth + 0.6% dividend + 4% multiple expansion = ~20% annually

Base case: 15% EPS growth + 0.6% dividend = ~16% annually

Bear case: 12% EPS growth + 0.6% dividend − 3% multiple drag = ~10% annually

Even the bear case competes with the S&P 500’s historical 10% average. The base case is 8 points of annual outperformance. And this is the core of the Kinsale/Apollo swap logic made numerical: I’m targeting the same EPS growth I was underwriting in those two positions, but now with a structural moat that makes the bear case feel survivable and the base case feel genuinely attractive.

4. My Proprietary Insight

On a forward earnings basis, Mastercard is sitting in territory it has occupied only once in the past decade, 2016, right before a multi-year re-rating. The current 34% discount to its own 10-year average P/E is not explained by deteriorating fundamentals. Revenue is accelerating. Margins are expanding. Buybacks are running hot. What’s changed is the narrative, not the business.

The part of the narrative I think is most mispriced: the stablecoin threat is being used as a discount on a company that is actively building the stablecoin infrastructure itself. Mastercard has already announced plans to expand settlement capabilities to include stablecoin options for issuers and acquirers. This is not a Kodak moment, it’s more like Kodak deciding to invent the iPhone. The market is applying a disruption discount to the company most likely to be on the right side of the disruption.

The second hidden angle is the services revenue mix shift. Most of the market still prices Mastercard as a pure transaction volume business, volume times rate. But value-added services (fraud detection, analytics, tokenization, AI consulting) grew 22% year-over-year in Q1 2026, well ahead of the 16% overall revenue growth. As that mix increases, the multiple on the business should structurally expand, not compress, that revenue stream is stickier, less regulated, and less cyclical than raw interchange. The market hasn’t priced that transition yet, which means you’re getting a business getting better priced as if it’s getting worse.

5. My Take

Sleep Well at Night Score: 8/10

What Excites Me

The earnings yield at 4.1% now matches the S&P 500 and nearly matches Treasuries, for a business growing EPS at 15%+ annually. That math doesn’t stay broken for long.

The services mix shift is a free call option on multiple expansion that isn’t in the consensus story.

On a forward P/E basis this is the cheapest Mastercard has been in a decade, driven by noise rather than any fundamental deterioration.

What Worries Me

Regulatory risk on interchange fees is permanent. One bad EU ruling and the revenue model takes a structural haircut.

The CFO transition during a critical strategic pivot (stablecoins, digital assets, agentic commerce) is the kind of timing that creates execution risk even when the transition itself is well-managed.

If EPS growth disappoints, say slows to 8–10% due to macro headwinds, the earnings yield argument evaporates and there’s limited valuation support from dividends alone.

Link to the full Portfolio

The market had been waiting for normalisation. Lower rates, softer inflation, a gentle glide path back to something resembling 2019. Energy just cancelled that flight. It has become the variable that no one in the rate-cutting camp wanted to account for, the one that keeps inflation alive long after the supply-chain shocks and the post-pandemic sugar rush were supposed to have faded.

The thesis here is simple: in a world of expensive energy and stubborn rates, you abandon capital-intensive businesses and move toward energy-efficient models and “commodity producers” sitting on the right side of the ledger. What follows is why, and how.

I. The Mechanics of Resilient Inflation

There’s a mental model I keep coming back to when thinking about energy prices: oil is not a commodity, it’s a tax. Every sector of the economy pays it, whether they know it or not.

Logistics pays directly, in diesel. Agriculture pays through fertilisers, most of which are synthesised from natural gas derivatives. Manufacturing pays through plastics, resins, and the energy cost of running factories. Real estate pays through heating. And consumers pay through all of the above, compressed into their grocery bills, their Amazon deliveries, and their utility statements. When oil goes up, it doesn’t spike in one corner of the economy. It seeps through everything, slowly, with a lag.

That lag is what makes it so treacherous. The first-order effect, the pump price, is visible and immediate. The second-order effects take six to eighteen months to fully materialise. And it’s those second-order effects that create the real problem.

Here’s how it works. Higher energy costs squeeze corporate margins. Companies facing squeezed margins do one of two things: they absorb the hit, or they pass it on. Most, eventually, pass it on. That means higher prices for services, not just goods. And higher prices for services mean workers start demanding higher wages to keep up. Now the inflation is no longer just “energy inflation,” it’s embedded. It’s in salaries. It’s in rent. It’s in the cost of a haircut. The Fed can’t target oil, but it absolutely has to respond to wages. And so the rate-cutting cycle everyone was pricing in gets delayed, then delayed again.

Now here’s the thing about central banks in this environment. They’re stuck in a genuinely unpleasant place. Cut rates, and you re-stimulate consumption, which increases oil demand, which pushes oil prices higher, which feeds back into inflation. Keep rates high, and you risk cracking the real economy, triggering a slowdown that destroys earnings but doesn’t necessarily bring energy prices down if the supply side remains constrained.

This is not a new dilemma. Cast your mind back to the 1970s, when OPEC embargoes and wage spirals created stagflation that neither rate hikes nor stimulus could cleanly resolve. The Fed under Volcker eventually broke it, but at the cost of a brutal recession. The 2020s version is playing out differently, with more complexity, more geopolitical moving parts, and a central bank that has already used a lot of its credibility. The historical comparison isn’t a perfect map, but it’s a useful reminder: energy-driven inflation has a habit of lasting longer than anyone projects.

II. Investment Strategy: Capital-Light vs. Heavy Industry

I’ve spent more time than I’d like on margin analysis for energy-exposed industrials, and the conclusion is consistently uncomfortable for anyone holding those positions.

Take a steel producer. Its cost base is directly tied to energy prices, through the blast furnace, through transport, through raw material extraction. When oil is at $60, the margin is manageable. At $106, you’re in a different conversation entirely. The same logic applies to airlines, to bulk chemical manufacturers, to any business that moves physical things across long distances. Their cost structures are not just exposed to energy, they’re defined by it.

Now look at what happens to valuation multiples in that environment. If margins compress, earnings fall. If earnings fall and rates stay high (because of said energy inflation), the multiple investors are willing to pay for each dollar of earnings also falls. You get hit twice. Once in the numerator of the earnings calculation, once in the denominator of the valuation. That’s a painful combination.

The alternative is what gets loosely called “capital-light.” Companies with minimal physical assets, low raw material dependency, and strong pricing power over their customers. Software businesses. IP licensing platforms. Professional services firms with irreplaceable expertise. The defining characteristic is simple: their operating costs are not correlated to the price of a barrel of oil. They don’t run blast furnaces. Their gross margins don’t move when diesel goes up.

Better yet, some of them actively benefit from the environment. If expensive energy forces every company in the world to optimise its operations, the software or advisory firm that helps them do it sees its value proposition go up, not down. Demand for efficiency solutions rises precisely when inefficiency becomes expensive.

And then there’s cash flow. In a world of high rates and high energy prices, free cash flow is the only metric that truly matters. Earnings can be managed. Revenue can be flattered by accounting. Free cash flow is harder to fake. A business that generates substantial FCF in this environment is one that can self-fund, can return capital to shareholders, and doesn’t need to refinance at punishing rates. That’s the screen I’d run first.

One of those numbers is wrong in most portfolio constructions right now, and I don’t think it’s the energy price estimate.

III. Sector Focus: The Oil Majors, Rehabilitated

The old mental model for oil companies went something like this: they’re cyclical, they’re dirty, they spend every dollar of windfall cash drilling new holes, and when the cycle turns they’re left with stranded assets and a balance sheet full of regret.

That model is outdated.

The majors that survived the 2014-2016 bust and the 2020 COVID collapse came out with very different capital allocation philosophies. They stopped chasing barrels at any cost. They cut costs structurally rather than cyclically. And when cash started flowing again, they returned it, through buybacks, through dividends, through debt paydown. This is not 2014, when $100 oil triggered a wave of speculative drilling and empire-building. Today’s majors are sitting on genuinely healthy balance sheets, with a discipline they didn’t have before.

Now here’s the thing about valuation. At $106 Brent, the cash flow yields on the major integrated oil companies are, in several cases, in the high single digits to low double digits. Compare that to a broad market trading at elevated earnings multiples with compressed yields, and the arithmetic becomes interesting. You’re getting paid substantially more to hold a major oil company than to hold most of the rest of the market, at a moment when that company’s revenues are directly indexed to the thing driving everyone else’s costs higher.

That’s an unusual combination of qualities. And the buyback programs amplify it: every share retired at today’s prices using today’s cash flows is a permanent improvement in per-share metrics for those who stay.

The counterargument is energy transition risk, and it’s real over a long enough horizon. But over a 12-to-36-month horizon, the transition is not moving fast enough to threaten the economics. At $106 oil, the producers are printing cash. The transition narrative is a long-term constraint, not an immediate one.

IV. Sector Focus: Uranium and the AI Bet

The artificial intelligence buildout has a power problem. A serious one.

Training large models and running inference at scale requires enormous amounts of electricity, continuously, reliably, with essentially zero tolerance for interruption. A data centre running LLM workloads cannot have the lights go out for four hours because the wind dropped. Intermittent renewables, however much capacity gets installed, cannot serve as the backbone of always-on, compute-intensive infrastructure.

Now. The tech industry has figured this out. Microsoft, Google, Amazon, and others have been quietly moving toward nuclear power agreements at a pace that would have seemed bizarre five years ago. The Three Mile Island restart, the wave of small modular reactor contracts, the sudden interest in dormant nuclear sites. This isn’t sentiment. It’s procurement.

Uranium is the input. And the uranium market is structurally short.

Production capacity was allowed to atrophy during the decade-long price depression that followed Fukushima. New mines take years to develop. Enrichment capacity is constrained. Meanwhile, demand projections have revised upward sharply, driven not just by AI infrastructure but by a broader reassessment of nuclear’s role in decarbonisation strategies across Europe and Asia.

The supply-demand mismatch in uranium today rhymes with what happened in semiconductors two years ago. Everyone knew the long-run demand story. Not everyone appreciated how long it takes to build new capacity, or how disruptive the shortage would be when it became acute. Uranium prices have already moved significantly. The question is whether the move is finished or just beginning. Given the structural deficit, I’d argue it’s closer to the beginning.

Build a Portfolio That Can Take a Punch

The 2026 investor doesn’t need to predict the price of oil. Trying to call whether Brent is $90 or $120 in December is a coin flip dressed up as analysis.

What actually matters is building something that works if oil stays elevated. That means balancing direct beneficiaries, the producers, with margin-protectors, the capital-light businesses that don’t bleed when energy costs spike. It means owning cash flows, not earnings stories. It means being sceptical of any valuation that assumes a return to the rate environment of 2021.

I’m aggressively underweight anything with a cost structure that treats cheap energy as a given.

The stowaway has been on the plane the whole time. The question is what you do now that you’ve found it.

Three months later, the Q1 2026 results are in. The machine is still running. The discount has actually widened.

Let’s update the numbers.

If you want the original articles you can find them here :

1. What Just Happened (Q1 Snapshot)

Brookfield reported $1.55 billion ($0.66/share) in Distributable Earnings for Q1 2026—essentially flat versus Q1 2025’s $1.549 billion. But the headline masks the important detail: DE before realizations grew 7% year-over-year, from $1.301 billion to $1.393 billion. The recurring earnings engine is getting stronger.

For the last twelve months (LTM) ended March 31, 2026:

The drop in total LTM DE is a simple prior-year comp issue: Q1 2025 included ~$1 billion in one-time disposition gains from the sale of a portion of BN’s stake in BAM (used to fund the AEL insurance acquisition). Strip that out and the underlying business is growing cleanly.

2. The Three Engines: Where Things Stand

Engine #1: Asset Management

This remains the crown jewel and it’s accelerating.

Fee-bearing capital hit $614 billion, up 12% year-over-year. Fee-related earnings grew 18% over the LTM to $3.07 billion. Year-to-date fundraising is already $67 billion (including $21 billion in Q1 alone), boosted by a $40 billion insurance mandate from Just Group—which Brookfield just closed the acquisition of. The seventh vintage flagship private equity fund is in its first close, targeting essential services and industrial businesses—what management calls “hard assets” for a cycle where everyone wants exactly that.

The asset management business now generates $1.953 billion in annualized DE at Brookfield’s share. At a 10x multiple on $2.7 billion net target carried interest, the carried interest component alone is worth $27.4 billion. Add to that the market value of BN’s 74% stake in BAM (1,193 million shares × $43.56/share = ~$52 billion at quarter-end), and asset management capital on a blended basis is ~$96.6 billion.

One new detail worth watching: Brookfield is quietly building a technology investment portfolio ($2.3 billion of balance sheet capital) that includes ~$1 billion in SpaceX shares, ~$500 million committed to Figure (humanoid robotics), a joint venture with OpenAI called The Deployment Company, and half ownership of Pinegrove Capital (a venture secondaries platform). These are still small relative to the overall business but reflect the “Watch, Learn, Invest” playbook that Bruce Flatt describes in this quarter’s shareholder letter early bets on secular trends before they become consensus.

Engine #2: Wealth Solutions (Insurance Float)

DE was $430 million for the quarter and $1.671 billion for the LTM growing but still building. The bigger news is what happened after quarter-end: Brookfield completed the acquisition of Just Group, a UK pension risk transfer platform. This added $40 billion to the insurance asset base, bringing total insurance assets to $180 billion.

Annualized cash flow is now $2.028 billion (including Just Group), supporting a $30 billion valuation at management’s 15x multiple. Return on equity is 15%, holding steady.

The gross spread remains intact at 2.24%: earning 5.77% on invested assets, paying 3.53% in cost of funds. Simple math: at $180 billion in assets, that spread generates ~$4 billion gross before operating costs. The Berkshire-playbook-with-real-assets analogy from my previous piece is even more apt now.

One strategic move worth flagging: Brookfield has announced the combination of BN and BNT (the Brookfield Wealth Solutions paired security). Shareholder votes are scheduled for July 16, 2026. The rationale is capital efficiency the combined entity unlocks roughly $145 billion of permanent capital to support the insurance business’s continued scaling. The company is also evaluating similar simplifications for its listed infrastructure and energy vehicles. Simpler structure = lower complexity discount = potential multiple expansion.

Engine #3: Operating Businesses

This is the quietest engine and intentionally so. DE was $360 million for Q1 ($1.536 billion LTM), down from $426 million in Q1 2025 due largely to reduced real estate distributions (BPG went from $215 million to $120 million quarterly). Infrastructure (BIP: +6% YoY) and Energy (BEP: +7% YoY) both grew.

The real estate fundamentals, however, remain solid: super core office occupancy at 96%, core plus at 95%, and net rents on new office leases running 15% above expiring levels. During Q1, Brookfield refinanced Two Manhattan West with a $1.9 billion non-recourse mortgage at 5.53%, repaid the prior $1.5 billion mortgage, and pocketed $400 million in net cash while keeping the building. That’s the asset recycling flywheel at work.

Monetization activity in Q1: $17 billion of asset sales, substantially all at or above carrying values. Accumulated unrealized carried interest stands at $11.8 billion (net $7.9 billion after costs). Management continues to guide toward $6 billion of carried interest realizations over the next three years.

3. The Updated Napkin Math

Management’s blended intrinsic value per share, as of March 31, 2026: $66.37 (down from $68.08 at year-end, primarily due to BAM’s share price declining from $52.39 to $43.56 during a volatile market quarter).

The components:

BN shares were repurchased at an average of $41 year-to-date management’s own characterization is a ~40% discount to intrinsic value. That’s not a marketing claim; it’s backed by the sum-of-the-parts math above. $470 million of BN shares and $575 million of BAM shares were repurchased in Q1 alone, in the open market, during the market volatility.

At $44/share (approximate current price):

• P/E on LTM DE before realizations: $44 / ($5.48B / 2.37B shares) = ~18x

• P/E on total LTM DE: $44 / ($6.01B / 2.37B shares) = ~17x

• Discount to blended intrinsic value: ~38%

• FCF Yield (DE before realizations): $2.32/share ÷ $44 = 5,2%

• FCF Yield (total DE): $2.54/share ÷ $44 =5,7%

The spread over 10-year Treasuries (~4.3%) remains over 100 basis points on a recurring basis, and over 150 basis points including realizations.

4. What Changed Since My Last Piece

The FY 2025 article used $44/share as the price. The stock has since sold off further, to ~$41 and came back to 44 now, ironically making the thesis more compelling since Earnings grew, not less. Let me be precise about what’s new:

What improved: Fee-related earnings growing faster (+18% LTM), insurance assets growing to $180 billion post-Just Group, BN/BNT combination announced (structural simplification = potential discount narrowing), $67 billion raised year-to-date already with a seventh PE fund in the works, management actively buying back shares at what they call a 40% discount.

What’s the same: The complexity discount is still there. Real estate distributions from BPG are softer. The stock hasn’t re-rated toward peers.

What’s slightly weaker: BAM’s share price declined ~16% in Q1 (from $52 to $44), reducing the blended intrinsic value per BN share from $68 to $66. This is mark-to-market noise on a publicly traded holding, not a deterioration in the underlying business.

5. The Structural Catalyst: BN + BNT Combination

This is the one genuinely new development worth examining carefully.

The proposed combination merges BN (the investment firm) with BNT (the Brookfield Wealth Solutions vehicle). The stated rationale: give the insurance business access to ~$145 billion of additional permanent capital sitting in BN’s balance sheet. This is the same logic that makes Berkshire Hathaway’s structure so powerful insurance float invested in a diversified pool of high-quality assets.

If approved at the July 16 shareholder meetings, the combined entity will trade under “BN” on NYSE and TSX. The company will also adopt US GAAP from Q1 2027, making it directly comparable to US-listed peers like Blackstone, KKR, and Apollo. That matters for valuation: if you’re running on different accounting standards, many institutional investors run you through a different (often lower-multiple) lens. Switching to US GAAP removes that friction.

Both changes simplification and US GAAP adoption are catalysts for multiple expansion that didn’t exist six months ago.

6. Risks (Still Real)

Nothing has changed structurally on the risk side, but a few deserve updated color:

Complexity discount: The BN/BNT combination and potential simplification of infrastructure/energy vehicles are explicitly designed to address this. Progress, but not resolved yet.

Real estate drag: BPG distributions were down significantly YoY ($642M LTM vs. $904M prior). Management characterizes this as timing and disposition-related, not fundamental deterioration 96% occupancy and improving lease economics support that view. But it’s worth watching.

Private credit sentiment: The shareholder letter addressed this directly. Market concerns about AI disruption of SaaS businesses have weighed on private credit broadly. Brookfield’s management explicitly states they have no material software exposure in their credit book, with the portfolio focused on real asset credit and opportunistic lending. Oaktree’s track record through cycles is the relevant benchmark here.

Capital deployment: $188 billion to deploy is a lot. Getting it wrong matters. The track record says they don’t $17 billion in Q1 asset sales, substantially all at or above carrying values, is reassuring but it’s a genuine risk at this scale.

7. The Updated Verdict

The Q1 2026 results don’t change the thesis they reinforce it.

The asset management business is growing faster than peers on a fee earnings basis. The insurance business just scaled to $180 billion with the Just Group acquisition. Operating businesses are stable. Management is aggressively repurchasing shares at a 40% discount. And for the first time, there are concrete structural catalysts (BN/BNT combination, US GAAP adoption) that could close some of the discount.

At $44/share, you’re paying:

• ~17x recurring earnings (peers at 22–25x)

• ~38% discount to blended intrinsic value ($66.37)

• 5.2% yield on recurring DE alone

The five-year return math is essentially unchanged from my prior piece: 12–15% DE growth + 0.6% dividend yield + ~2% buyback yield + potential multiple expansion from structural simplification = 15–18% CAGR scenario if things go right.

If you owned it at $47 per my last piece: the thesis is intact, the price is better, and you now have two concrete catalysts you didn’t have before. Add on weakness.

If you’re new: this is one of the more compelling setups in large-cap financials right now. The complexity that keeps most investors away is a feature, not a bug it’s precisely why the discount exists. The business itself is not complicated. It collects fees, earns spreads on insurance float, and owns world-class real assets. It does all three simultaneously, at scale, with a 30-year track record.

The Portfolio Update

Here is a look at my complete portfolio. As an investor, I am extremely concentrated by design.

I’ll be the first to admit it: I have significantly underperformed the S&P 500 since the start of 2026 (lagging 5%). Our two flagship holdings, Kinsale Capital and VusionGroup, have been absolutely torn apart by the market recently.

With this latest update, Brookfield now moves into the #3 spot in the portfolio by value. It sits just behind S&P Global (SPGI) which is my most recently added position and VusionGroup, a line where I continue to accumulate more and more shares with Kinsale and Apollo.

I am staying the course. I know the fundamental power of these companies and their long-term potential. I’d much rather be patient with high-quality businesses I understand than reckless by chasing the latest “trend” stocks trading at 35x earnings.

See the full portfolio

As always i’m updating the PDF, this time with a special offer, the report is a full equity deep-dive on Brookfield Corporation, updated with Q1 2026 results. It covers the investment thesis, a plain-English breakdown of how the Brookfield ecosystem works insurance float, asset management fees, and direct asset ownership and the key financial data: segment earnings, peer valuation comparisons, and scenario analysis from bull to stress case.

Get the Institutional version for 2.99$

The story nobody tells is colder. Literally.

Cold storage, the network of temperature-controlled warehouses that keeps your grocery delivery from arriving as a warm puddle, is undergoing the kind of infrastructure buildout that happens once a generation. And it’s happening right now, mostly out of sight, financed by institutional money that has been quietly accumulating assets for the better part of a decade. I’ve spent more time than I’d like going down this rabbit hole, and the more I look at it, the more I think most retail investors are completely missing the setup.

The Misconception Everyone Has About Grocery Delivery

Here’s what most people get wrong. When you think about the infrastructure behind your Instacart order or your grocery delivery, you probably picture a warehouse that looks like a regular warehouse but with some refrigerators in it. Maybe some guys in puffy jackets pushing pallets around.

The reality is a completely different category of real estate. A cold storage facility is closer to a precision instrument than a box with shelves. The temperature zones run from 55°F for produce to minus 10°F for ice cream. One building can have six or seven distinct climate chambers, each with separate mechanical systems, redundant backups, specialized racking, and floor insulation thick enough to prevent the frozen ground underneath from heaving and cracking the foundation.

The construction cost per square foot runs two to four times higher than a standard logistics warehouse. That’s before you factor in the energy systems. Now here’s the thing about that cost structure: It creates an accidental moat. Most of the cold storage capacity in the United States was built before 1980. It’s aging, inefficient, and increasingly unable to meet the food safety standards that regulators and large grocery clients now require. The modern facilities being built today are enormously expensive to replicate. And because the barriers to entry are so high, the operators who own the good assets have pricing power that would make a standard warehouse REIT jealous.

The Math the Market Has Been Sleeping On

Let me give you a number that I keep coming back to. The United States has roughly 5 billion square feet of total industrial real estate. Cold storage accounts for approximately 240 million square feet of that. Just under 5%.

And yet the demand growth trajectory for cold storage has been running at two to three times the rate of regular industrial space for the last several years, driven by three things that aren’t going away: online grocery, pharmaceutical cold chain, and the broader consumer shift toward fresh and prepared foods.

The supply side hasn’t kept up. New cold storage development is expensive, slow to permit, energy-intensive, and requires specialized operators to run. You can’t just convert a standard Amazon fulfillment center into a cold chain facility over a long weekend. The lead time from breaking ground to operational is typically 18 to 24 months, and the capital required scares off most developers who would rather build a cheaper, faster industrial box.

The gap between demand growth and supply response is exactly the kind of structural imbalance that tends to translate into sustained rent increases and high occupancy rates. Cold storage vacancy rates have been running in the low single digits nationally. For context, a healthy standard industrial market runs vacancy around 5 to 7%. Cold storage has been operating well below that for years.

The “Diva” Molecule: Why GLP-1s are a Logistics Nightmare

While everyone is focused on the grocery bag, the real margin expansion is happening in the medicine cabinet. The pharmaceutical angle isn’t just a “nice-to-have” tailwind; it is a structural shift in the quality of demand.

Standard chemical-based pills (small molecules) are relatively stable. They can sit in a standard warehouse for weeks without losing efficacy. Biologics including the GLP-1 blockbusters like Ozempic and Wegovy are different. They are “divas.”

These drugs are large, fragile protein molecules that require a strictly maintained temperature range, usually between 2°C and 8°C (36°F to 46°F). A single excursion outside that range doesn’t just reduce shelf life; it can render millions of dollars of product medically useless and legally unsellable.

This creates a level of “customer stickiness” that is almost unparalleled in real estate. A pharmaceutical giant isn’t going to switch providers to save $2.00 per square foot if it means re-auditing a new facility’s redundant power systems and cooling protocols. In the cold chain, the cost of the real estate is a fraction of the value of the cargo. That is the definition of a high-leverage pricing position.

A Framework for Thinking About Cold Storage Investments

When evaluating this space, the bears often argue that automation will compress margins by reducing the labor advantage of incumbent operators. That’s worth taking seriously, but the automation investment itself is a capital expenditure that only the largest, best-capitalized operators can afford.

Here is the four-filter framework I use when I look at this space:

1. Network Density: A cold storage operator with facilities in every major population center can offer national grocery and pharmaceutical clients a single contract. An operator with great facilities in three markets but gaps elsewhere loses national contracts to someone with 80% as good facilities but 100% coverage.

2. Temperature Range Capability: Multi-temperature facilities that can handle produce, dairy, frozen, and pharmaceutical in the same building complex are dramatically more valuable. The customer stickiness is higher because switching means re-qualifying multiple temperature zones with a new operator.

3. Energy Infrastructure: Cold storage is one of the most energy-intensive real estate categories in existence. Operators who have locked in long-term power purchase agreements, invested in on-site solar, or built relationships with utilities for favorable industrial rates have a cost structure that newer entrants simply cannot replicate quickly.

4. The Silent Barrier (Insurance & Liability): In a standard warehouse, if the roof leaks, you move the pallets. In cold storage, if the power fails and the backup generators don’t kick in, you’re looking at a multi-million dollar inventory total loss. The cost to insure these facilities has outpaced almost every other real estate category. Only the largest institutional players those with the balance sheets to self-insure or the scale to negotiate global master policies can survive this “insurance wall.”

Three Ways to Play It

1. Americold Realty Trust (COLD)

Americold is the most direct expression of this thesis in public markets. It’s a REIT, so you own the real estate directly, and it is the largest publicly traded cold storage operator in the world. Their customer list Unilever, Kraft Heinz, Tyson, Nestle represents long-term leases and high switching costs where Americold’s systems talk directly to the customer’s inventory management software.

The stock has had a rough few years, trading well below its 2021 highs as rate sensitivity hurt all REITs and some operational hiccups dented near-term earnings. That creates an entry point conversation worth having. The occupancy recovery story and the pharmaceutical demand tailwind are real catalysts that the market hasn’t fully repriced yet.

2. Lineage Logistics (LINE)

Lineage is the name you need to know. It went public in mid-2024 in what was the largest REIT IPO in US history. The scale is almost hard to process: over 480 facilities across 20 countries. Lineage is also the most technologically advanced operator in the space, having invested heavily in proprietary software that optimizes storage density and energy consumption in real time.

The Duel: Lineage vs. Americold With Lineage now trading publicly, we have a heavyweight battle of philosophies. Lineage is the Silicon Valley version of a REIT, leaning into algorithmic warehouse management to optimize their biggest variable cost: energy. Americold is the seasoned incumbent with a more traditional footprint and superior network density in high-velocity food corridors. If Lineage continues to trade at a “tech premium,” Americold becomes a compelling relative value play as its occupancy rates normalize.

3. United States Cold Storage (USCS) The Private Benchmark

USCS is private, owned by the Hong Kong conglomerate John Swire & Sons. The reason to know it is benchmarking. Private cold storage assets have been changing hands at 20 to 25x EBITDA in recent years. When you’re evaluating Americold or Lineage, understanding what private transactions look like gives you a valuation floor.

Why This Belongs On Your Radar (And the Risks)

The obvious reason is that cold storage sits at the intersection of three secular trends: the grocery delivery buildout, the pharmaceutical cold chain growth, and the aging existing infrastructure in the US. The less obvious reason is that institutional players Blackstone, Goldman Sachs have been accumulating these positions for a decade. They show up early; retail investors show up when there’s a magazine cover story.

The Rate Sensitivity Trap: Why the “Napkin Math” is Shifting

We have to be honest about the risks. REITs live and die by the spread between their cap rates and their cost of debt. In this mid-2026 environment, where the 10-year Treasury remains stubborn at 4.4%, the math has undergone a necessary repricing.

A 100 basis point move in interest rates directly eats into AFFO. However, because the replacement cost of these facilities has skyrocketed, the “economic rent” required to justify new construction has risen even faster than interest rates. This protects the incumbents. We are moving into a market where valuation isn’t driven by cheap debt, but by the ability to pass through capital costs to a tenant base that literally cannot afford to move.

The Concentration Risk

Finally, Americold and Lineage together control somewhere between 25 and 30% of total US cold storage capacity. Regulators haven’t looked at this market structure with any particular interest yet, but a world where antitrust scrutiny increases could complicate the pricing power thesis.

Conclusion

The window for getting in before this becomes consensus isn’t unlimited. The Lineage IPO was a signal. When the largest REIT IPO in US history happens in a sector and the mainstream financial press mostly shrugs, you have a narrow window where the smart money has arrived but the narrative hasn’t. In 2026, the performance won’t come from finding the next “box,” but from owning the infrastructure that keeps the world’s most sensitive molecules and its dinner frozen solid.

New report available in the library.

And don’t forget that I just dropped my 10 pages reports on Blackrock where I argue that BlackRock isn’t just an asset manager it’s the operating system of the global financial stack.

Go Get It

They’re missing the real story.

I just released a new 10-page Deep Dive on BlackRock for Waver Research Library subscribers. In this report, I argue that BlackRock isn’t just an asset manager it’s the operating system of the global financial stack.

The Moat Nobody Can Tear Out

While the world focuses on iShares, the true crown jewel is Aladdin.

• Scale: Aladdin processes $25 trillion in assets for over 1,000 institutions.

• Stickiness: It has a 98% client retention rate.

• Switching Costs: Replacing Aladdin takes 12 to 24 months of operational risk and retraining.

As I detail in the report, BlackRock has built “plumbing” that is nearly impossible to remove once installed.

The $28 Billion Pivot

In the last 18 months, BlackRock has spent over $28 billion to acquire GIP, HPS, and Preqin. This isn’t random spending; it’s a calculated move to migrate the business toward high-margin Alternatives.

The math is simple: every dollar shifted from a core index ETF into an alternative mandate is worth 10 to 20 times more in fee revenue.

The Valuation Gap

The most compelling part of the thesis? The market is pricing BLK at a discount to the broader index.

• S&P 500: Trades at roughly 26x trailing earnings while growing 8–10%.

• BlackRock: Trades at roughly 25x earnings while growing at nearly twice the rate.

What’s Inside the Full Dossier?

In this 10-page institutional-grade report, I break down:

• The Network Effect: Why Aladdin gets smarter as it gets larger.

• The Preqin Integration: How BlackRock is standardizing the “Bloomberg Terminal” of private markets.

• The Risk Matrix: A candid look at M&A integration risks and the “sleep score” for this position.

• Scenario Analysis: My Bear, Base, and Bull cases with projected annual returns through 2029.

  Get it for 6.99$

TriMas makes the pump on your lotion bottle. Not the lotion, the pump. Through brands like Rieke and Taplast, they supply dispensing components to consumer goods giants, and through Norris Cylinder they make high-pressure steel tanks for industrial gas. It’s unglamorous, it’s slow-growing, and nobody at a dinner party has ever asked about it. That’s kind of the point.

In March 2026, the company completed the sale of its aerospace division for $1.45 billion in cash, netting $1.2 billion after taxes. Against a total market cap of roughly $1.4 billion today, that one transaction was almost the size of the entire company. The result is a balance sheet that looks almost comically strong: $1.31 billion in cash, $397 million in debt, and a net cash position of $913 million sitting there doing nothing while management figures out what to do next.

The reason to look at this right now isn’t a growth story. There isn’t one. Revenue from the remaining businesses is growing at 3–6%, in line with inflation, essentially. The reason to look at it is that the stock appears to be priced as if the cash barely exists, the businesses are valued near zero, and you’re getting paid to wait. That’s a value play, not a compounder thesis.

1. THE MACHINE

The Simple Explanation

Think of TriMas as a toll road with a giant pile of coins sitting next to the booth. The toll road, the packaging and cylinders business, isn’t spectacular. Traffic grows slowly, pricing is moderate, and nobody’s building a better pump for Unilever anytime soon, but nobody’s disrupting it either. The giant pile of coins is the $913 million in net cash. The value play here is simple: you’re buying the toll road for next to nothing because the market keeps squinting at the coin pile and worrying about what management does with it, instead of just pricing the assets rationally.

The Moat

It exists, but let’s be honest about what it is. This isn’t a wide-moat business. The switching costs are real, changing a dispensing pump supplier mid-product-cycle means regulatory re-approval, compatibility testing, and supply chain headaches, but they’re not impenetrable. A large competitor with lower costs could chip away over time. The moat is really more of a “sticky mud” than a proper castle moat: not impossible to cross, but annoying enough that most customers don’t bother unless you give them a good reason to. For a value play, that’s fine. You don’t need a brilliant moat. You need enough durability to hold the business together while the discount closes.

The ROIC Story

ROIC is running around 12% on the operating businesses. Decent, not exciting. Critically though, with $913 million in net cash depressing the return on total capital figure, the underlying business assets are actually working reasonably hard. If you strip out the idle cash and look at what the operating businesses earn on the capital actually deployed in them, the picture is cleaner. This is a business that earns its keep, it just happens to have a massive cash balloon tied to it that’s distorting every ratio you try to calculate right now. For a value investor, that distortion is the opportunity, not the problem.

The Risks

The risks here are concentrated in one place: what management does with $900 million. If they overpay for acquisitions, which is extremely common when boards suddenly feel rich, they’ll destroy in eighteen months what took years to build. The packaging M&A market isn’t cheap; multiples for quality targets run 10–14x EBITDA, and a bad deal at the wrong price could turn a fortress balance sheet into a leveraged mess within two years. There’s also a subtler risk: management inaction. If the cash just sits uninvested for three-plus years, the investment thesis doesn’t break, but it also doesn’t resolve. You can be right about the value and still underperform because the catalyst never shows up.

2. THE NUMBERS

Current Valuation

• Price: ~$39

• Market Cap: ~$1.42B

• Cash: $1.31B

• Debt: $397M

• Net Cash: $913M

• Enterprise Value: ~$507M this is the number that matters. You’re paying $507 million for the actual operating businesses.

The EV is what the market is really placing on the packaging and cylinders operations. Against 2025 continuing revenues of $645.7 million, that’s an EV/Revenue multiple of 0.78x. For context, the closest comparable AptarGroup, which makes similar dispensing products for similar customers trades at roughly 2x revenue. Silgan Holdings trades at about 1.2x. TriMas at 0.78x is either a bargain or a warning sign, and the businesses themselves don’t have anything obviously wrong with them.

Profitability Snapshot (Continuing Operations)

• Revenue 2025: $645.7M

• Operating Margin 2025: 5.3% (guided to expand 300+ basis points in 2026 through cost-out actions)

• Adjusted EPS 2025 (continuing ops): $0.55

• Adjusted EPS guidance 2026: $1.50–$1.70 (midpoint $1.60)

• That EPS jump looks explosive, but a chunk of it is interest income (~$0.25/share) from the cash pile. Strip that out and the operational EPS is around $1.35. Still a big improvement on $0.55.

Valuation Metrics

• Forward P/E on $1.60: ~24x looks expensive at first glance

• But back out the cash. Net cash per share is $913M / 36.3M shares = roughly $25 per share. The stock is at $39. So you’re paying $14 per share for the actual business.

• $14 per share on ~$1.35 of operational EPS = roughly 10x P/E on the operating business alone. Now it looks cheap.

• Earnings Yield (on total price): ~4.1% barely above the 10-year Treasury. But the earnings yield on just the operating business, ex-cash, is closer to 9.6%. That’s meaningful.

• S&P 500 Earnings Yield: ~4.0%. The headline TRS yield barely beats it, but the ex-cash operating yield smashes it.

Shareholder Returns

• Dividend: $0.16/year → ~0.4% yield. Token.

• Buybacks: the company repurchased over $150 million of stock between November 2025 and March 2026, taking shares from ~40.7 million to ~36.3 million an 11% reduction in six months.

• At the current pace annualized, buyback yield is running around 10–12%. That pace will slow, but even at a more moderate 5%, total shareholder yield is north of 5%.

• Interest income on idle cash: ~$9M per quarter, roughly $36M annually. That’s 2.5% of the current market cap, coming in at zero operational cost.

Quality Indicators

• Balance sheet: net cash of $913M. Effectively unleveraged. The safest balance sheet in the room by a wide margin.

• Interest coverage: irrelevant, they’re a net receiver of interest, not a payer.

• Debt/Equity: negative in the best possible sense.

3. THE NAPKIN MATH

This is a value play, so the math works differently than a compounder analysis. We’re not modeling ten years of reinvestment at high ROIC rates. We’re asking: what happens when the gap between intrinsic value and market price closes? And when does that happen?

Scenario A: The Clean Re-Rating (Base Case)

The market is pricing the operating businesses at 0.78x EV/Revenue. If that re-rates to 1.2x still below Silgan, still well below Aptar, the operating business EV rises from $507M to roughly $775M. Add back the $913M net cash at face value. Total implied equity value: $1.69B. At 36.3M shares, that’s $46.50 per share. From $39, that’s a 19% return plus buybacks and dividends along the way. Timeline: 12–24 months if a catalyst emerges (acquisition announcement, further buyback acceleration, or just the market noticing the discount).

Scenario B: Buyback Machine (No Catalyst Needed)

No re-rating. No acquisitions. Management just keeps buying back stock at $39 with the cash pile. If they deploy $300M in buybacks over the next two years at current prices, shares outstanding drop from 36.3M to roughly 28.6M a 21% reduction. 2026 EPS of $1.60 spread over 28.6M shares becomes roughly $2.10. At even 18x a conservative packaging multiple the stock should trade at ~$38 on pure earnings math, but now you own a proportionally larger piece. Effective return to remaining shareholders: solidly positive just from the mechanics of shrinking the share count.

Scenario C: Management Blows the Cash (Bear Case)

They announce a $700M acquisition of a packaging company at 13x EBITDA, the market hates it, the multiple on the combined entity compresses, and the stock drifts back to $32–34. This is the real bear case and it’s not a small risk. The history of industrial conglomerates doing large transformational deals with freshly-received divestiture proceeds is not encouraging reading.

The Return Summary

• Base case re-rating: +19% price appreciation + ~4–5% annual shareholder yield = strong 1–2 year return

• Buyback-only scenario: mid-to-high single digit annual return, no catalyst required

• Bear case bad deal: –15% to –20% drawdown before recovery

vs S&P 500 at ~10% annually: base case wins on a 1–2 year horizon, loses badly on a 5-year horizon if growth doesn’t materialize.

4. MY PROPRIETARY INSIGHT

The $25 Bill Hidden in the Stock

Here’s the frame that most people aren’t using. At a $39 stock price, $25 per share is net cash money sitting in accounts earning 4.4% right now. That means you’re buying $25 in cash plus a real operating business for $39. You’re valuing the packaging and cylinders businesses at a total of $14 per share, or roughly $508 million.

Now look at what Blackstone and Tinicum just paid for the TriMas Aerospace business: 18x EBITDA. That was a fastener business with strong aerospace cycle tailwinds, sure, but it tells you what sophisticated PE buyers pay for TriMas-quality industrial assets at peak conditions. If TriMas Packaging were carved out and sold in a private market transaction, the floor for that business is probably $800M–$1B based on EV/EBITDA comps in the packaging space. The market is pricing it at $508M. That gap is the trade.

The historical pattern in small-cap industrials supports this. Post-major-divestiture, the “stub” that remains tends to trade at a discount for 6–18 months as the market processes the transformation, reassigns analyst coverage, and waits to see what management does with the proceeds. Then one of three things happens: a smart acquisition gets announced and the market re-rates, buybacks shrink the float enough that the per-share math forces the price higher, or an activist gets involved. All three are plausible paths here. None requires the business to suddenly grow faster.

The Interest Income Nobody Is Pricing Correctly

Management guided approximately $9 million in interest income per quarter from the invested cash call it $36 million annually. At TriMas’s current tax rate of 27–29%, that’s roughly $26 million net, or about $0.71 per share. The stock trades at 24x forward earnings, but $0.71 of those earnings per share is pure Treasury-rate cash income that disappears the day they deploy the capital. The market is either ignoring this or discounting it too harshly. Either way, you’re collecting it while you wait for the real catalyst, which isn’t a bad position.

5. MY TAKE

Sleep Well at Night Score: 6.5/10

The balance sheet is the best in the sector, the downside is genuinely limited, and the discount to intrinsic value is real and quantifiable. The 6.5 instead of an 8 is entirely management risk one bad acquisition undoes the whole setup and there’s no structural protection against it.

What Excites Me

The ex-cash P/E of roughly 10x on a stable, sticky packaging business with margin improvement underway is legitimately cheap for the quality of the asset. The buyback pace since November 2025 has been exceptional 11% float reduction in six months signals that management knows the stock is cheap and is acting on it. And the interest income from the cash pile means you’re getting paid ~2.5% of market cap annually just to sit and wait, before any operational improvement shows up.

What Worries Me

A single large ill-timed acquisition could turn a beautiful balance sheet into a debt problem in one press release. The operational earnings power without the interest income kicker is still modest $1.35/share adjusted is not a lot of cushion for a $39 stock. And this is a slow, under-covered small cap the discount could persist for years without a catalyst, and patient money isn’t always patient enough.

Bottom line, you’re paying $39 for a stock where $25 is cash and the business underneath is priced like it’s about to go away it isn’t, and that gap looks like it wants to close.

American Express posted the highest spend growth in three years and the market shrugged. Kinsale posted a 77.4% combined ratio, a 47% dividend hike, and $62.5M in buybacks at above-market prices, and got dinged for flat premium volume. Two different companies, same story: the headline lied, the filing didn’t.

This week’s breakdown is about learning to read the right number. Let’s go.

AMERICAN EXPRESS

0. THE SCOREBOARD

EPS of $4.28, up 18% year-over-year, beating consensus by roughly 4%. Revenue at $18.9B, up 11%. Guidance reaffirmed at 9-10% revenue growth and $17.30 to $17.90 EPS for the full year. Stock dipped slightly post-earnings on the reinvestment commentary, which was the wrong read.

Gut reaction: Strong quarter that accelerates the thesis. Spend growth at a 3-year high wasn’t in the base case.

1. DID THE STORY CHANGE?

Yes, and for the better.

Billed Business hit $428B, up 10% year-over-year. That’s the highest quarterly spend growth in three years. The premium cardholder base isn’t just holding, it’s reaccelerating. Net write-off rate improved to 2.0% from 2.1% a year ago, which means they’re growing balances and improving credit quality simultaneously. That combination is the thesis in its purest form.

The one thing worth watching is that expenses grew 11%, exactly matching revenue. Squeri made the deliberate call to reinvest the Q1 beat into marketing and technology rather than let it flow to the bottom line. Strategically correct. But operating leverage won’t appear until that investment cycle matures. Watch the margin trajectory in H2.

Provisions ticked up to $1.3B from $1.2B, driven by slightly higher write-offs and a smaller reserve release. Not alarming. But the direction is worth tracking.

Verdict: ⚡ Thesis Strengthening. Spend at a 3-year high, credit quality improving, NFL partnership globally, NBA extension, and an AI commerce developer kit all in the same quarter. Squeri is playing offense.

2. WHAT ACTUALLY MOVED

The number that matters isn’t EPS. It’s Billed Business at $428B, up 10%. Everything else revenue, NII, card fees flows from that line. When spend reaccelerates, the model compounds.

Net investment income also grew quietly alongside spending, as card balances build the float. Full-year EPS guidance midpoint at $17.60 on a 16-17x forward multiple puts fair value around $299. Buybacks continue to shrink the share count, which is doing roughly 2-3 points of the per-share math on its own.

3. THE ONE THING MOST RECAPS MISSED

The Amex Agentic Commerce Experiences developer kit got approximately zero coverage. Here’s why it matters: if AI agents become the transaction layer for daily life, whoever controls the payment authentication and trust layer wins. Amex is embedding their payment capability into AI ecosystems right now, before the platform winners are decided. It’s a 2028 story dressed up as a 2026 product announcement, and the market isn’t pricing it at all.

4. MANAGEMENT CREDIBILITY: 🟢 Delivered

They said spend would accelerate. It did. They said credit would hold. It did. No sandbagging, no pivots. Squeri also made a line worth decoding: when he said they’re increasing investments to “capitalize on long-term growth opportunities,” that’s not hedging language. That’s a CEO who sees the next three years more clearly than the market does and is spending accordingly.

5. MY TAKE

The spend reacceleration removes the biggest lingering doubt. The question was whether premium consumer spending could stay elevated. Q1 answered it definitively.

Bull case got stronger: Spend at a 3-year high, demographics still a tailwind, agentic commerce positioning is a free option the market hasn’t noticed yet.

Bear case is narrower but real: If macro softens, provision creep could accelerate from here. And at 16-17x forward earnings, any EPS guidance cut bites hard.

The one-liner: “The machine is running exactly as expected, and management decided to push it harder. Boring is beautiful, except when it’s actually accelerating.”

KINSALE CAPITAL

Flat GWP. Record underwriting profit. Market panicked at the wrong number.

0. THE SCOREBOARD

Diluted EPS of $4.88, up 27% year-over-year. Net income $112.6M, up 26%. Combined ratio improved to 77.4% from 82.1%. Gross written premiums essentially flat at $482M, down 0.5%. That last number is what spooked people. It shouldn’t have.

Gut reaction: The market read the wrong line. Flat GWP with a 77.4% combined ratio, a 66% dividend hike, and $62.5M in buybacks at above-market prices is not a struggling company. It’s a disciplined one.

1. DID THE STORY CHANGE?

No. And that’s the point.

The Commercial Property division fell 28.3% because standard carriers are flooding into E&S property at softened rates. Kehoe is refusing to write bad business at bad prices. Excluding Property, the rest of the book grew 6%. The flat headline number is management doing exactly what you want management to do.

Meanwhile, the actual business metrics moved in the right direction across the board. Underwriting income up 40.1%. Net investment income up 26.5% to $55.4M, as the $5.3B portfolio keeps rolling old 2-3% bonds into new money at roughly 5%. Operating cash flow at $248.9M. The platform is compounding quietly while the GWP headline creates noise.

The one thing that hasn’t resolved is construction liability adverse development in the 2018 and 2019 accident years. The overall prior-year development was positive at $18.7M net, but those older construction years keep producing adverse surprises. Not thesis-breaking yet. But it’s showing up for the third or fourth consecutive quarter, and bears will keep citing it until it stops.

Verdict: Thesis Strengthening. Kinsale is transitioning from a growth stock to a compounding machine. The flat GWP is the feature, not the bug.

2. WHAT ACTUALLY MOVED

The combined ratio at 77.4% is the headline number that matters. For every dollar in premium, Kinsale keeps 22.6 cents as pure underwriting profit before investment income. That improved from 82.1% a year ago, driven by dramatically lower catastrophe losses ($1.6M versus $22.6M in Q1 2025, which caught the Palisades Fire) and favorable prior-year reserve development.

Net retention ratio improved to 83.7% from 78.8%. Kinsale is ceding less to reinsurers and keeping more of its own risk, because it trusts its own pricing more than the reinsurance market. That confidence is showing up in the numbers.

The capital return picture is worth its own paragraph. $62.5M in buybacks at $378.64 average, which is above where the stock trades today. Quarterly dividend at $0.25, up from $0.17 a year ago. When management buys stock above market, that’s not a scheduled program. That’s a conviction signal.

3. THE ONE THING MOST RECAPS MISSED

Buried in Item 4 of the 10-Q: Kinsale implemented a new general ledger system in Q1 2026. This got zero coverage. Management says controls were effective throughout, and there’s no indication of any problem. But a mid-year general ledger transition is the kind of operational change that can surface accounting irregularities one or two quarters later. File this away and check Item 4 specifically in the Q2 filing. Non-event if nothing appears. Worth knowing it happened regardless.

The bigger picture nobody’s writing: this is the GEICO moment. Standard carriers entering E&S property at irrational prices will eventually take losses and exit. When they do, Kinsale will be standing there with its broker relationships intact, its balance sheet clean, and its technology platform ready to write Property at the rates the market should have been charging all along. The wait is the strategy.

4. MANAGEMENT CREDIBILITY: 🟢 Delivered

The construction liability adverse development was disclosed clearly and contextualized honestly. No surprises buried, no pivots on prior guidance. What Kehoe said last quarter happened this quarter. The discipline on Property pricing has been consistent across multiple earnings cycles, which means it’s culture, not a one-quarter call.

5. MY TAKE

The capital return posture removes doubt. When management buys stock at $378 and the market gives you the same stock at $340-350, they’ve told you something important with their own money.

Bull case got stronger: 77.4% combined ratio with room to improve further, NII compounding at 26.5% growth as the portfolio reprices, and a Property rebound coming when standard carriers eventually retreat.

Bear case is narrower but real: Construction liability adverse development hasn’t resolved after multiple quarters, which raises the question of whether it’s cyclical or something more structural. Property softness could persist longer than expected. And the general ledger transition is a minor risk worth verifying in Q2.

Why Flat Is the New Smart

Here’s the thing about insurance that most people outside the industry don’t fully appreciate: it runs in cycles, and right now we’re in the part of the cycle that separates the disciplined from the desperate.

The soft market is when competition peaks. Rates come down, terms get looser, and carriers start fighting for volume instead of profit. Standard insurers, under pressure from shareholders to show growth, start flooding into markets they’d normally avoid, including E&S lines like commercial property, offering coverage at prices that only make sense if nothing bad happens. The key word there is if.

This is exactly what’s happening right now. Standard carriers are cutting rates in commercial property and stepping into Kinsale’s territory. Kinsale’s response? Let them. The Commercial Property division shrank 28.3% this quarter not because brokers stopped sending submissions, but because Kehoe looked at the rates on offer and said no. Voluntarily. Repeatedly. Across an entire division.

This is where most insurance companies burn themselves. The soft market feels fine while it lasts. Loss ratios look manageable, new business is flowing, everyone’s growing. And then a hurricane makes landfall in the wrong place, or a wildfire season runs longer than the models predicted, or a string of large construction liability claims hits in the same quarter, and suddenly the policies written at 2023 prices are paying out 2026 claims. The carriers who chased volume in the soft market are the ones scrambling to raise capital, pulling back from markets, and posting ugly combined ratios for two or three years straight.

Kinsale has seen this movie before. So has every disciplined E&S underwriter. The hard market that follows a major loss event is where the real money gets made, and the only way to be positioned for it is to not have wrecked your balance sheet in the soft market trying to look busy.

When the next major catastrophe hits, and it will, the carriers who priced irrationally will retreat. Some will exit markets entirely. The brokers who placed business with them will need new homes for that risk, fast, and they’ll call the carriers they trust. Kinsale will pick up the phone with a clean book, a $5.3B investment portfolio, and the ability to write those policies at whatever price the market needs to charge to actually make money. That moment is where years of combined ratio discipline converts into years of premium growth.

The flat GWP isn’t a warning sign. It’s a reservation for a table that’s going to be very full, very soon.

The one-liner: Market panicked at the headline. The filing told a different story. Flat GWP with a 77.4% combined ratio, a 47% dividend hike, and $62.5M in buybacks at above-market prices is not stagnation. It’s a compounding machine choosing quality over quantity.

I have an important update to share with you today.

Since launching this Substack, my goal has been to provide high-quality, independent research on companies that often fly under the radar. However, I’ve realized that by keeping my most in-depth work behind a paywall, I’ve been limiting the very thing I value most: the growth of our community.

I want my research to be read, debated, and shared by thousands of investors, not just a few.

That is why I’ve decided to pivot: for at least the next 6 months, this newsletter is going 100% free.

The “Welcome Gift”: 6 Premium Analyses Unlocked

To kick things off, I have officially removed the paywall from 6 of my most detailed reports. If you missed these because they were for paid subscribers only, you can now access the full investment theses here:

• Kinsale Capital ($KNSL): A masterclass in specialized insurance and high-growth compounding.

• Burford Capital ($BUR): Decoding the world’s leader in litigation finance a complex but fascinating play.

• Brookfield ($BN): Everything you need to know about the king of alternative asset management.

• Verisk ($VRSK): Exploring the massive data moat protecting the global insurance industry.

• Lumine Group ($LMN): A vertical market software “compounder” with a unique acquisition model.

• Propel Holdings ($PRL): How this fintech is disrupting the lending space with AI.

What this means for you

1. Total Access: From now on, you will receive every single deep-dive and market update directly in your inbox, with no restrictions.

2. Pure Research: I am not a financial advisor. My mission remains the same: to provide the data and analysis you need to make your own informed decisions.

3. The Goal: I’m aiming to grow this base to 5,000 subscribers. If you find value in my work, the best way to support this project is to share it with a friend or colleague.

I’m excited to start this new chapter with all 520 of you. Stay tuned the next analysis is already in the works.

Happy reading,

Brookfield Corporation is no longer just a diversified alternative asset manager; it is a permanent-capital compounding machine powered simultaneously by three self-reinforcing engines: the BAM asset management flywheel (73% owned, $3.0B in fee-related earnings), the Brookfield Wealth Solutions insurance float (growing 24% annually, targeting 50% of distributable earnings by 2029), and $1.18T in inflation-linked real operating assets spanning renewable power, infrastructure, real estate, and private equity. The planned merger with its insurance entity (BWS) into a single listed security is the structural catalyst that should force a re-rating from the current 8.7x DEBR multiple toward the 15–18x historical range before a single dollar of earnings growth is applied.

The stock currently trades at a 37% discount to intrinsic value while posting its strongest financial year on record.

What’s inside the Research PDF:

• Institutional Layout: An 8-page structured dossier optimised for professional study and printing, built on the same framework as Waver’s Apollo deep dive.

• The Three-Engine Flywheel Analysis: A deep dive into BN’s unique permanent-capital structure BAM fees, insurance float, and direct asset ownership and why the compounding interaction between the three is structurally undervalued by a market that prices it as a simple conglomerate.

• The BN/BWS Merger Thesis: A proprietary analysis of why the planned consolidation of Brookfield Corporation and its insurance entity into a single listed security is the re-rating catalyst and why it maps directly onto the Berkshire Hathaway structural parallel.

• Proprietary Valuation Matrix: Detailed Bear, Base, and Bull case scenarios with explicit DEBR (Distributable Earnings Before Realisations) projections through 2029, including a stress scenario for severe recession and real estate impairment.

• Risk Scorecard: Analysis of the real estate complexity noise vs. the fundamental compounding reality including the $11.6B unrealised carried interest pipeline and $188B deployable capital base that consensus models consistently underestimate.

• High-Resolution Data: All charts and financial tables rendered in vector quality, including AUM growth trajectory, DEBR per share path to 2029, P/DEBR peer comparison, monetisation volumes, and fee-bearing capital build.

Get the Standalone Report

If you prefer to own this specific analysis without a monthly subscription, you can purchase the standalone Institutional PDF via the link below.

Get it for 9.99

Read more

Today, we’re diving into two companies that have absolutely nothing in common except for the fact that they both just spent the first quarter proving the doubters wrong.

• BlackRock is busy turning the global capital markets into its own personal sandbox, using a mix of massive ETF inflows and a private credit machine that refuses to slow down.

• VusionGroup is quietly installing the digital nervous system for the world’s largest retailers, proving that their “pipes in the ground” strategy is creating a software moat that’s getting wider by the minute.

Both stocks have been weighed down recently by “clouds”one legal, one macro but the underlying numbers are screaming acceleration. Grab a coffee; here is why the fundamentals are currently running laps around the stock prices.

BlackRock : The Beast is Bolder, But Lawyers Lurk

0. THE SCOREBOARD

• EPS (adj.): $12.53 (Beating the $11.70–$11.96 consensus by ~7%).

• Revenue: $6.70B (Beating estimates of $6.46–$6.62B by ~4%).

• The Vibe: The stock popped 3% at the open, but broader market “bad vibes” dragged it to a -0.6% finish. Management basically said, “March was a bit bumpy, but April is looking lush.”

One-sentence gut reaction: A powerhouse quarter that basically slapped the “private credit panic” across the face, though the lawyers are still hovering in the background like uninvited party guests.

1. THE THESIS CHECK: THE MACHINE IS HUMMING

• ETF Dominance 🟢: iShares isn’t just winning; it’s lapping the competition. Record Q1 net inflows of $132B(double last year’s fee growth). Organic base fee growth hit 8%, the highest in five years. The “passive is dead” narrative died again today.

• Private Markets Jackpot 🟢: $9B of fresh inflows this quarter. Total alternatives assets hit $687B (up 50% YoY). The HPS acquisition is already acting like a money printer, chipping in $230M in fees in its first full quarter.

• Aladdin’s Magic Flywheel 🟢: Tech revenue grew 22% to $530M. This isn’t just a “moat” anymore—it’s a massive, compounding revenue line that people can’t live without.

• Margin Creep 🟡: Operating margin hit 44.5%, slightly missing the 45%+ target. Integrating big fish like HPS and Preqin is expensive. We need to see these synergies actually show up in the bottom line soon.

• The HLEND “Drama” 🟡: The bear case was all about the “redemption gate” (limiting withdrawals). CFO Martin Small basically laughed it off: the fund has a 10.4% return and is actually seeing new money ($840M in subs). Operationally? It’s fine. Legally? The Pomerantz investigation is still a dark cloud over the ticker.

• Institutional Bleeding 🔴: Big institutions pulled $35B this quarter. It’s low-fee business, but it signals that the “big boys” are rebalancing away from BlackRock’s standard products. Not a dealbreaker, but worth a side-eye.

2. THE NUMBERS THAT MATTER

• The Heartbeat: Base fees hit $5.44B (+24% YoY). This is the recurring revenue that funds the jets and the dividends.

• Performance Fees: Shot up to $272M (vs. $60M last year). Thank HPS for that 353% spike, but don’t expect it every quarter.

• Capital Allocation: BlackRock isn’t a tech startup; it’s a cash cow. They returned $1.3B to shareholders this quarter ($450M buybacks + a 10% dividend hike to $5.73/share). That dividend raise is the real “we’re confident” signal.

3. MANAGEMENT SAYS VS. REALITY

• Fink’s Confidence: Larry Fink used words like “strongest start in history” and “built to compound.” Usually, CEOs hedge. Larry is practically shouting from the rooftops.

• The “Flight to Scale”: Fink thinks macro chaos makes people run to BlackRock because they’re the “safe harbor.” With $130B in fresh cash during a 4.6% market dip, he’s probably right.

• The Dodge: Analysts kept poking at the Pomerantz investigation regarding HLEND. Management answered with “operational facts” (returns/liquidity) but totally sidestepped the legal risk. That’s where the “multiple ceiling” lives for now.

4. PROPRIETARY INSIGHT: THE FOOTNOTE NOBODY READS

Look at the GAAP-to-adjusted math. Usually, adjusted earnings are higher than GAAP because management “forgets” some expenses. Not this time! GAAP was higher ($2.81B) than Adjusted ($2.67B). * Why? Because BlackRock’s stock price fell, the amount they might have to pay for their acquisitions (contingent consideration) dropped. Their own stock going down actually created a $549M GAAP gain. If the stock rallies next quarter, GAAP will look “worse.” Don’t let it confuse you; it’s just accounting voodoo.

5. MY TAKE

• Sleep Well Score: 8/10. The business is a tank. The only thing keeping the stock down is the legal overhang and general market grumpiness.

• The One-Liner: “The bears brought a knife to a gunfight, but the lawyers are trying to call a foul.”

Vusion : The IoT Monopoly is Loading

0. THE SCOREBOARD

• Adjusted Revenue: €294M (+26% YoY).

• The “Secret” Number: At constant exchange rates (ignoring the weak Euro), growth was actually +36%.

• The Vibe: The stock closed +3.7%. The market finally realized that even if the Euro is struggling, the business is on fire.

One-sentence gut reaction: Record Q1 with the high-margin software business (VAS) moving way faster than management’s own targets yet the stock is still priced like it’s 2024.

1. THE THESIS CHECK: THE FLYWHEEL IS SPINNING

• The Software Transition 🟢: VAS revenue (the high-margin stuff) hit €51M (+53% YoY). Management’s full-year goal was +40%, and they’re already smashing it. VAS is now 17% of total sales.

• The Infrastructure Moat 🟢: VusionCloud connected devices hit 435M (up from 188M last year). When a retailer has half a billion of your gizmos in their stores, they aren’t switching to a competitor over a 5% price difference. That is “pipes in the ground” dominance.

• The Whale Dominoes 🟢: Carrefour just signed a 3-year European exclusivity deal. Walmart expanded its partnership to Mexico. When the two biggest retailers in the world use you, everyone else has to follow or get left behind.

• EMEA Lag 🟡: Europe grew only 10% while the rest of the world grew 37%. Management promises an “acceleration” in H2. We’ll see. Q2 needs to show some European muscle.

• The Price Gap 🔴: The stock is lingering near 52-week lows (~€123) while the business is printing record numbers. Why? FX headwinds (they earn USD but report in EUR) and tariff fears. It’s a “valuation gap” that is frustrating but clearly driven by macro, not the company.

2. THE NUMBERS THAT MATTER

• Recurring Revenue: Annualized run-rate for Cloud/SaaS is now €111M. This is the “sleep well at night” money.

• Order Intake: €316M for the quarter. Note: The Walmex deal (signed March 30) isn’t even in this number. The real pipeline is much bigger.

• The September Wait: We don’t get full profit/margin data until the H1 results in September. This “visibility gap” is why big funds are sitting on their hands.

3. MANAGEMENT SAYS VS. REALITY

• The Gadou “Pivot”: CEO Thierry Gadou is reframing the business. He says stores aren’t just shops; they’re strategic assets. Translation: “I’m going to charge retailers more because my tech helps them do e-commerce and ads, not just change price tags.”

• The Walmart Showcase: Management is using Walmart as a “live reference.” They don’t have to explain why the tech works anymore; they just point at Bentonville and say, “They like it.”

• Conservative Guidance: They kept the +15–20% revenue target for the year. Given they just did +36% at constant FX, this is management-speak for “we’d rather beat a low bar than miss a high one.”

4. PROPRIETARY INSIGHT: THE VANISHING “WALMART DRAG”

There is a massive accounting distortion here that almost everyone is missing. To get the Walmart deal, Vusion gave them warrants (basically stock options). Under IFRS rules, the value of those warrants is subtracted from reported revenue.

• In Q1 2025: This drag was -€17.6M.

• In Q1 2026: It’s down to -€5.3M. As Walmart spends more, this “penalty” eventually hits $0. When that happens, Vusion’s reported revenue and growth will suddenly “pop” purely because of accounting rules. Investors following IFRS are literally seeing a worse business than the one that actually exists.

5. MY TAKE

• Sleep Well Score: 7.5/10. The Walmart/Carrefour combo basically eliminates “bankruptcy risk.” Now it’s just a question of when the market wakes up and stops valuing a high-growth tech firm like a boring hardware company.

  The One-Liner: “Vusion is currently a SaaS company trapped in a hardware company’s valuation, waiting for the September margin data to set it free.”

Not through debt. Not through equity raises. Not through some financial engineering trick that makes your CFO nervous. What if your customers millions of them simply handed you cash upfront, before you delivered the product, and while you held that cash over the following weeks, months, or years, you invested it, made a return, and kept the difference? What if the business model itself was specifically designed to use other people’s money as its primary fuel?

That’s the float. And it’s probably the single most underappreciated concept in all of investing.

Most people who hear the word “float” think of Warren Buffett, nod vaguely, and move on. That’s a mistake. Because float isn’t just a quirk of insurance companies it shows up in membership businesses, subscription platforms, asset managers, and retailers. It’s hiding in balance sheets that most investors read backwards. And the companies that generate the most durable float tend to compound capital at rates that quietly embarrass the rest of the market over a full decade. Today we’re going to pull back the curtain on exactly how it works, what it looks like in the numbers, and how to spot it yourself before the stock moves.

First, The Problem: You’ve Been Reading Balance Sheets Wrong

Standard finance education teaches you that liabilities are bad. Debt is bad. Obligations are bad. The less a company owes, the better. And for most companies, that’s directionally correct debt costs money, it amplifies downturns, and it can sink you when rates spike.

But there is a special category of liability that flips this logic entirely. It’s the money companies collect from customers before they’ve done anything to earn it. An insurance premium paid in January covering you through December. A Costco membership fee charged in January before you’ve touched a single rotisserie chicken. A software subscription billed annually upfront. On the balance sheet, all of this shows up under “deferred revenue,” “unearned premiums,” or “policyholder reserves” a liability, technically, because the company hasn’t yet delivered what you paid for.

Here’s what most investors miss: that liability cost the company absolutely nothing to acquire. No interest payments. No equity dilution. No covenants. No maturity schedule. You just gave them your money voluntarily, because you wanted what they were selling and while they hold it, they get to put it to work.

Now compare that to how a traditional manufacturer funds its operations. They buy raw materials on credit, pay workers weekly, carry inventory for months, and only get paid 30-60 days after they’ve shipped the product. Every phase of the production cycle requires capital. To grow revenue by 20%, they need to fund 20% more working capital meaning debt, equity raises, or retained earnings that could have gone to shareholders. Growth is capital-intensive by design.

A float business works in reverse. The customer funds the operation. Growth generates more float. More float generates more investment income. The machine feeds itself.

The Buffett framing of this is precise: “insurers receive premiums upfront and pay claims later this collect-now, pay-later model leaves us holding large sums that will eventually go to others. Meanwhile, we get to invest this float for our own benefit.” The genius isn’t in the insurance. The genius is in recognizing that the insurance product creates a structurally free source of permanent capital.

The Numbers That Will Make This Click

Let’s work through the math concretely, because this is where most explanations stop too soon.

Take a simple example. An insurance company collects $100 in annual premiums. Over the course of the year, it pays out $85 in claims and operating expenses. That leaves a $15 underwriting profit fine, not exciting. But here’s what the income statement doesn’t show you directly: that $100 in premium was sitting in the company’s investment portfolio for an average of six months before claims were paid. If the company earns 5% on its investments, that’s $2.50 in investment income on top of the $15 underwriting profit. And in a good year where underwriting is especially tight, you can have a situation where investment income exceeds underwriting profit meaning the insurance business itself is almost irrelevant. The real product is the float.

The metric that captures this is the combined ratio claims plus expenses, divided by premiums. A combined ratio below 100% means the insurer made a profit on underwriting alone, before touching the investment income. Above 100% means underwriting lost money, but investment income from the float may still make the overall business profitable.

The elite underwriters run combined ratios well below 100%, consistently, across years. Progressive, for example, more than doubled net income to $8.5 billion in 2024, running a combined ratio of 88.8 for the full year meaning they spent only 88 cents of every premium dollar on claims and costs. They pocketed 12 cents on the underwriting side, and then invested the entire float on top of that. Chubb’s 2025 results were even more striking P/C underwriting income hit a record $6.5 billion on a combined ratio of 85.7.

What makes this almost obscenely powerful is scale and time. Berkshire Hathaway’s float has grown from $46 billion in the early 2000s to $176 billion at year-end 2025. That’s $176 billion in permanently available, interest-free investment capital and Berkshire didn’t borrow a cent of it. Every year, millions of policyholders renew their coverage, new customers sign up, and the float grows. Buffett has been investing this pool for 50+ years, compounding it at ~20% annually, and the result is a company worth over a trillion dollars that started as a struggling New England textile mill.

The critical insight: over two decades, Berkshire’s insurance businesses generated $32 billion of after-tax profits from underwriting about 3.3 cents per dollar of sales after income tax. Meanwhile, the float grew from $46 billion to $171 billion. The underwriting profit is nice. The float is the point.

It’s Not Just Insurance: Float Is Everywhere Once You Know Where to Look

Most people, when they hear “float business,” immediately think of Berkshire and stop there. That’s leaving serious money on the table, because the float concept shows up in at least four other business models that generate it just as reliably often without the catastrophe risk that comes with property and casualty insurance.

The Membership Float

Costco is not a retailer. That’s the most important thing to understand about Costco, and almost no one frames it correctly. Costco is a membership club that happens to sell things at near-cost, using the membership fee as its only meaningful profit source.

Virtually all of Costco’s operating income derives from the $4.6 billion in membership fees, while the $237.7 billion in merchandise sales essentially just covers operational costs. The entire retail operation buying, storing, stocking, selling runs roughly at break-even. The profit engine is the annual membership fee, collected upfront every year, before members have spent a dollar on groceries.

But Costco has a second layer of float that most people miss entirely: the inventory financing float. Costco turns inventory 12.4 times per year selling through stock in just 26-27 days while negotiating 30-60 day payment terms with suppliers. The math here is quietly remarkable: customers pay Costco for the goods before Costco pays the supplier. At Costco’s scale, that gap represents billions of dollars in interest-free financing from the supplier network, growing automatically every year as revenues grow. It’s negative working capital as a structural feature, not an accident.

Costco’s US and Canada membership renewal rate sits at 92.9% a moat signal that tells you this float is essentially permanent. It doesn’t evaporate in a recession. It doesn’t require new customers. It just keeps renewing, quietly funding the operation year after year.

The Asset Manager Float

This one is more sophisticated, and it’s why certain alternative asset managers those that manage insurance company assets, pension funds, or annuity liabilities are genuinely different animals from traditional fund managers.

The traditional asset manager collects a 1% fee on AUM. Fine business. But the fee is earned continuously on assets that can leave at any time. The float here is thin.

The modern alternative asset manager that has cracked the float code does something smarter: it acquires or partners with insurance companies that generate long-duration liabilities annuities, life insurance policies, pension obligations. Those liabilities can last 20-30 years. The insurance company holds the float; the asset manager invests it in higher-returning private credit, infrastructure, or real assets; the spread between the liability cost and the investment return is pure economic profit. Crucially, that float is sticky. Annuity holders don’t switch managers. Pension obligations don’t run. The capital is as close to permanent as anything in modern finance gets.

Apollo represents the modern evolution of the float concept: Athene collects long-duration annuity liabilities life insurance float rather than P&C float and invests them in private credit assets originated by Apollo’s platform. This is why Apollo looks cheap on traditional PE metrics the market keeps valuing it like a cyclical fund manager when the insurance float model makes it structurally closer to a bank that can invest its deposits in higher-returning private assets.

The Subscription Float

Software companies figured this out quietly in the 2010s. Annual subscription billing, collected upfront, generates deferred revenue on the balance sheet technically a liability, practically an interest-free loan from customers. The best SaaS companies grow their deferred revenue faster than their revenue, meaning they’re collecting money faster than they’re earning it, and investing the difference in growth. The float isn’t as large or as durable as insurance float, but for high-retention subscription businesses with net revenue retention above 120%, it’s real and it compounds.

The Five-Point Float Checklist

Here’s your repeatable framework. Run any company through these five questions before you decide what multiple it deserves.

1. Does the company collect cash before delivering value? Look for insurance premiums, annual subscriptions, membership fees, gift card balances, advance deposits, or deferred revenue growing over time on the balance sheet. If any of these exist, float exists.

2. Is the float growing faster than the underlying business? Float that compounds faster than revenue signals that the company is collecting more upfront capital per dollar of eventual service pricing power and float growth together. That’s the holy grail.

3. What does the float cost? For insurers: look at the combined ratio. Below 95% consistently means the underwriting itself is profitable the float is cost-free or even negatively priced (they get paid to hold your money). For membership businesses: look at the CAC-to-LTV ratio and renewal rates. If 92% of members renew annually and the membership costs $65, the effective cost of float is near zero.

4. How does management invest the float? This is the art. Berkshire invests float in equities and operating businesses 15-20% annual returns over decades. Most insurers park float in short-duration investment-grade bonds 4-5%. The spread between those two outcomes, compounded over 30 years, is the difference between a good business and a legendary one. Find management teams with a clear, long-term investment philosophy for the float.

5. Is the float permanent or episodic? A one-time project advance from a client isn’t float. But a business with 90%+ annual renewal rates, growing its float organically every year, has something that is functionally indistinguishable from permanent equity capital except it shows up as a liability.

Three Float Businesses Worth Understanding Right Now

Berkshire Hathaway (BRK.B) The textbook case, but the numbers still shock

People think they understand Berkshire. They mostly don’t. They think of it as a stock portfolio with an insurance business attached. It’s the opposite: it’s an insurance float machine with an investment portfolio attached. Berkshire’s insurance subsidiaries generated about $11.4 billion in underwriting profit in 2024 alone one of the best results in the company’s history while maintaining a total float of roughly $171 billion.

Here’s the number that almost nobody focuses on: insurance investment income grew from $9.6 billion in 2023 to $13.7 billion in 2024, as Berkshire deployed its float into short-term Treasuries during the high-rate environment. That $13.7 billion generated entirely from investing other people’s money exceeds the annual revenue of most Fortune 500 companies. And Berkshire didn’t need to sell a product, hire a salesperson, or open a factory to earn it. They just... held the float and collected the coupon.

The underappreciated angle on Berkshire right now: with the float at $176 billion and growing, and with rates still elevated, the investment income line alone can generate $12-15 billion annually before Buffett’s successor invests a single dollar in equities. The floor on earnings is remarkably high.

Progressive (PGR) Float growing at machine speed

Progressive doesn’t get enough credit as a float story because people focus on the car insurance angle. But Progressive’s model is specifically engineered to grow float as fast as possible while keeping the underwriting machine profitable.

Progressive’s stated annual goal is explicitly to grow as fast as it can while achieving a 96 combined ratio, in other words, maximize float growth subject to a profitability constraint. In 2024, they ran an 88.8 combined ratio while growing premiums 21%. They were 7 points better than their own target. That means they were capturing float faster than planned, at a lower-than-expected cost.

What most people don’t realize: Progressive’s direct-to-consumer model is actually a float advantage. When you buy insurance directly rather than through an agent, the policy gets written faster, the premium clears faster, and the float sits on Progressive’s balance sheet rather than in some agency’s account for a few days. At $74 billion in annual premiums, even days of float acceleration are worth hundreds of millions of dollars annually.

Markel Group (MKL) The Berkshire blueprint for patient investors

Markel is the clearest case of someone explicitly reading Buffett’s manual and running the experiment at a smaller scale. Its model combines specialty insurance underwriting with a long-term investment portfolio and a growing roster of non-insurance businesses earning premiums from niche lines and deploying the float alongside retained earnings.

The specialty angle is the key differentiator. Markel doesn’t compete with Geico on personal auto. It insures summer camps, equine veterinarians, specialty contractors, and professional liability for architects markets so niche that the major carriers don’t bother, competition is thin, and pricing power is exceptional. This focus on specialty lines has historically produced underwritten profits, with a combined ratio averaging below 95% over the past decade.

Management estimated intrinsic value at $2,610 per share at year-end 2024 against a stock price of $1,726 a 34% discount to intrinsic value by the company’s own calculation. When management of a float business tells you the stock is 34% cheap and backs it up with a share repurchase program, that’s worth paying attention to. The float funds the buybacks. The buybacks reduce shares outstanding. The remaining shareholders own a larger slice of a growing float. That’s a compounding wheel that runs itself.

Why This Changes Everything About How You Build a Portfolio

The practical takeaway is this: when you’re comparing two businesses with identical growth rates, identical margins, and identical management quality, always pay a premium multiple for the one that generates float. Always. Because the float is an off-balance-sheet asset that the accounting doesn’t capture, a source of capital that the income statement doesn’t show, and a competitive advantage that compounds invisibly for decades.

The best float businesses tend to share three characteristics that make them genuinely different from the rest of the market. First, they get more valuable as they grow more float means more investment capital means more investment income means stronger earnings. Second, they’re remarkably durable in downturns premiums and memberships already collected don’t evaporate when the economy weakens. And third, management’s capital allocation skill matters enormously the same float, invested at 15% versus 5% over 30 years, produces wildly different outcomes for shareholders.

The market discounts float businesses incorrectly in two directions. Sometimes it ignores the float entirely, pricing an insurer purely on underwriting earnings without crediting the investment income a common mistake with small specialty insurers. And sometimes it prices the float correctly but underweights management’s ability to grow it missing the fact that float that compounds at 15% for 20 years doesn’t just add linearly, it explodes.

Spotting this before the market does is exactly the kind of edge that separates portfolios that beat benchmarks from portfolios that track them.

That is why I am launching the Waver Research Library: a permanent, structured archive of high-fidelity PDF research dossiers. Each one covers a single company end-to-end thesis check, P&L deep dive, KPI analysis, proprietary insight, scenario framework, and a full definitions appendix. The kind of document you actually save, reference again, and share with someone who asks “what do you think of this stock?”

The first 4 dossiers are live today.

Apollo Global Management (APO — NYSE) — FREE

8 pages. The Athene float, 16 origination platforms, 0.1% default rate, and why a 22% YTD selloff may be mispricing one of the most elegant business models in American finance.

Use code waver100 at checkout.

Vusion Group (VU — Euronext Paris) — $9.99

9 pages. FY2025 blowout analysis. The VAS flywheel, the ESL installed base model, the hidden ARR story inside non-recurring revenue, and a bottom-up 2030 valuation. The flywheel just went supercritical.

Lumin Group (LMN - TSXV) — $9.99

10 pages. Lumine trades at 19× FCFA2S, a multi-year low, while FY2025 delivered record revenue (+15%), operating income (+31%), and free cash flow available to shareholders (+153%). The discount is driven by organic growth anxiety and Synchronoss integration risk, not business deterioration.

What a Waver dossier looks like

Each report follows the same structure: a one-page cover that stands alone, an executive summary in four bullets, numbered investment pillars with embedded charts, a Bear/Bull scenario table, a proprietary insight section, and a definitions appendix. The goal is a document you could hand to someone who has never heard of the company and they would understand both the business and the risks within 15 minutes.

These are not slide decks. They are not Twitter threads reformatted into PDF. They are research documents with data, with citations, with honest risk flags, and without a buy/sell recommendation, because I am not your financial adviser.

FOR PREMIUM SUBSCRIBERS

You do not need to do anything. PDF dossiers are being integrated directly into your paid articles with a 100% discount code in each article, as a permanent upgrade to your membership. You already own the library. If you’re not a premium subscriber, this is the moment you grab -20%

Get 20% off forever

Why now

The standard newsletter format is good for keeping up. It is less good for building conviction. When a position is volatile when headlines are noisy and the market is panicking what you actually need is a structured document you can return to that reminds you why you own what you own. That is what the Library is for.

Apollo is a good test case. The stock is down 22% year-to-date on reputational noise while the business posts record results across every metric. Having an 8 pages document that walks through the Athene float, the default history, the FCF trajectory, and the risk matrix makes it considerably easier to hold or to decide it is not for you.

The Library is at waver.one/s/the-waver-research-library. Apollo is free for a month. Vusion is €9.99. Both are there right now.

More dossiers are in production. If there is a company you want to see covered in this format, reply to this article and tell me or send me a message. The queue is open.

Despite record-breaking FY2025 results including a 153% jump in Free Cash Flow Available to Shareholders (FCFA2S) the market remains distracted by short-term organic growth anxiety. With the “accounting fog” of preferred shares finally lifted, the real cash machine is now visible for the first time.

What’s inside this Institutional PDF:

• The WideOrbit Deep Dive: A detailed look at Lumine’s “hidden monopoly” that processes the dominant share of the $25B U.S. local broadcast TV advertising market.

• The Synchronoss Catalyst: Analysis of Lumine’s first-ever public company acquisition ($258.4M EV), its integration of 11 million subscribers, and what to watch for in the April 30 earnings call.

• The “Accounting Fog” Breakdown: How the March 2024 preferred share conversion removed an $87M annual distortion, revealing the business’s true 36% operating margins.

• Sum-of-the-Parts Valuation: Our proprietary model suggesting that at current prices, WideOrbit alone accounts for a massive chunk of the market cap, leaving the rest of the 30+ company portfolio at a deep discount.

• Waver Risk Scorecard: A clear-eyed assessment of organic growth trends and the unique cultural risks of scaling the Constellation playbook to Tier-1 public entities.

Get it for 9.99

Why this report matters right now

“Lumine is currently in a rare valuation disconnect. While the market focuses on 0-1% organic growth, it is ignoring a massive structural re-rating. The delta between the current 19x FCFA2S multiple and the historical 25-35x range for Constellation-style execution is where the alpha lies.”

Access for Waver Premium Members

Read more

The Asset: VusionGroup (formerly SES-imagotag)

VusionGroup is at the heart of the “Retail Renaissance.” Beyond simple digital price tags, the company is building the mandatory operating system for physical retail in a world of high labor costs and e-commerce competition.

With a massive backlog and the recent Walmart US rollout, VusionGroup is transitioning from a hardware manufacturer to a high-margin software & services platform.

What’s inside this Institutional PDF:

• The Walmart Alpha: A detailed breakdown of the US contract and why the market is still underestimating the “Halo Effect” on other global retailers.

• Margin Expansion Analysis: How the transition to VAS (Value Added Services) is decoupling revenue growth from COGS, shifting the profile toward a SaaS-like business.

• Proprietary Valuation Model: Our DCF and multiples-based analysis, including a breakdown of the 2027 “Vusion-27” strategic plan targets.

• Short-Seller Post-Mortem: A clear-eyed look at past allegations (Gotham City Research) and why our analysis suggests the fundamental story remains intact.

• Waver Risk Scorecard: Analysis of component supply chain risks and the competitive landscape from China.

Get the Standalone Report

If you prefer to own this specific analysis without a monthly subscription, you can purchase the standalone Institutional PDF via the link below.

Download the Vusion Report for 9.99

Why this report matters right now

“VusionGroup is currently in a ‘Sweet Spot’: the heavy R&D phase is behind them, and the hyper-growth phase in the US is just beginning. The delta between current sentiment and 2026 earnings power is where the alpha lies.”

Read more

Apollo Global Management ($APO)

This standalone dossier represents the complete institutional research compiled by Waver. Designed for capital allocators and deep-value investors, this report moves beyond standard editorial content to provide a structured, data-driven investment framework.

Apollo is no longer just an alternative asset manager; it is a high-yield, low-volatility “origination machine” powered by the Athene engine.

What’s inside the Research PDF:

• Institutional Layout: A 8-page structured dossier optimized for professional study and printing.

• The “Athene” Flywheel Analysis: A deep dive into Apollo’s unique permanent capital structure and why it creates a structural advantage over competitors like Blackstone.

• Proprietary Valuation Matrix: Detailed Bear, Base, and Bull case scenarios with explicit ANI (Adjusted Net Income) projections through 2026.

• Risk Scorecard: Analysis of the “Redemption Gating” noise vs. the fundamental credit reality.

• High-Resolution Data: All charts and financial tables in vector quality for clear analysis.

Get the Standalone Report

If you prefer to own this specific analysis without a monthly subscription, you can purchase the standalone Institutional PDF via the link below.

Download the Apollo Report for 9.99

Read more

Propel Holdings was born from a simple, uncomfortable truth: roughly 40% of Americans can’t get a loan from a bank. Not because they’re deadbeats but because a decades-old algorithm called FICO was never designed with them in mind. The typical Propel customer is an older millennial or younger Gen X, aged 35 to 54, with average credit limits between $2,000 and $3,000. These are nurses, warehouse workers, gig drivers people with jobs and income, invisible to the traditional financial system. Propel’s pitch: let AI figure out what FICO can’t.

The current drama is juicy. Q4 2025 net income dropped 49% year-over-year to $5.9 million, spooking the market. PRL stock fell roughly 42% from its six-month highs. But here’s the twist: the full year was a record revenue increased 31% to $589.8 million for fiscal 2025. The Q4 hit was mostly accounting timing: you provision for loan losses upfront (hit to earnings now), but collect the revenue over the life of the loan (profit later). It’s like a restaurant being penalized for buying ingredients before serving dinner.

Why analyze Propel right now? Because 8 out of 8 covering analysts rate it a Strong Buy, with an average 12-month price target of CAD $30.16 against a stock sitting near CAD $24. The market is pricing in a permanent deterioration; analysts think it’s temporary noise. One of them is right, and figuring out which one is worth your time.

1. THE MACHINE

The Simple Explanation

Think of Propel like a really smart pawn broker who figured out that the reason most pawn brokers go bust isn’t the customers it’s the bad scoring system. Banks use FICO like a bouncer with a single rule: “no entry if you’ve ever been broke.” Propel built a smarter bouncer. Its AI platform analyzes over 5,000 data points per applicant to get a holistic view of financial health, looking at cash flow patterns, employment stability, and pay-cycle timing rather than ancient credit history. It processes 60,000+ applications per day and delivers decisions in seconds. Then it lends money at high interest rates to borrowers who have few alternatives, and it keeps getting better at predicting who will actually pay back.

The Moat

The moat here is subtle but real it’s a data flywheel. With each loan screened and data inputted, the AI gets stronger. This is the core compounding asset: more loans → more repayment data → sharper models → lower charge-offs → higher profitability → ability to lend more. After 14 years of data accumulation and over $2 billion in credit facilitated, replicating that dataset is genuinely hard for a new entrant.

There are also switching costs on the borrower side Propel has “graduation programs” where good customers move to lower rates and higher limits over time, creating loyalty in a segment that normally has none. And then there’s the licensing moat: Propel received regulatory approval to launch Propel Bank in December 2025, which is not easy to replicate and opens up completely new product categories.

The ROIC Story

This is where it gets interesting for compounders. Annualized adjusted Return on Equity was 27% for the full year 2025, and management is targeting 28%+ adjusted ROE for 2026. That’s excellent it means the business can grow without constantly diluting shareholders. Crucially, the new FreshLine product launched with $210 million in forward-flow commitments from third-party investors, meaning Propel generates fee revenue while the credit risk sits largely off its own balance sheet. This Lending-as-a-Service pivot is the most important structural shift to understand: they’re moving from “lender who takes risk” to “platform that earns fees,” which is a dramatically higher-quality business model.

The Risks

Let’s be direct here because the risks are real and not to be waved away. Credit losses are very high around 50% of the loan book compared to banks’ average of 0.7–1%.That’s not a typo. This is the business model: you charge 100%+ APR on small loans to risky borrowers, and you lose a lot of them, but you win enough to make money. It works in a good economy. In a recession, delinquencies spike and the whole machine can seize up.

Regulatory risk is significant. The CFPB has historically targeted exactly this type of lender high-rate consumer credit to financially vulnerable people. One bad administration decision or a state-level rate cap could eliminate entire revenue streams overnight. And the AI moat, while real, is not impenetrable: low barriers to AI entry mean a well-funded competitor with access to similar data could theoretically catch up. Finally, the macro sensitivity is severe. This is not a business you want to hold through a deep recession.

2. THE NUMBERS

(All figures in USD unless noted; stock price in CAD)

Current Valuation

• Price: ~CAD $24.10

• Market Cap: ~CAD $950M (~USD $680M)

• Enterprise Value: ~USD $960M (adding ~USD $307M debt, subtracting ~USD $28M cash)

Profitability Snapshot (FY2025)

• Revenue (TTM): USD $589.8M (+31% YoY) record

• Net Income (TTM): USD $59.5M (+28% YoY) record

• Adjusted EBITDA: USD $130.3M

• Operating Margin: ~20% (on an adjusted basis)

• Note: For financial companies, we focus on earnings, not free cash flow

Valuation Metrics

• P/E (TTM): ~11.2x

• Forward P/E: ~6.3x (based on 2026 guidance)

• Historical P/E range: essentially not comparable pre-profitability; since becoming consistently profitable (2022 onward), the stock has traded between ~10x and ~25x earnings

• Earnings Yield (TTM): ~9%

• vs 10Y US Treasury (~4.4%): Propel offers a +4.6% spread over risk-free

• vs S&P 500 Earnings Yield (~4.1%): Propel offers a +4.9% premium substantial

• At a forward PE of 6.3x, forward earnings yield is ~16% almost 4x the S&P 500

Shareholder Returns

• Dividend yield: ~4.8% (following the 10th consecutive dividend hike, announced February 2026)

• Buyback: NCIB announced November 2025 buyback yield estimated ~1-2%

• Total Shareholder Yield: ~6–7%

Quality Indicators

• Debt/Equity: 117.9% elevated, typical for a lending business

• Interest Coverage: 3.5x (EBIT / Interest Expense) adequate but not comfortable

• The debt/equity ratio has fallen from 613% to 117.9% over the past 5 years dramatic deleveraging as the business matured

3. THE NAPKIN MATH

A. Growth Driver (EPS Growth)

Management guided for net income of $70–90 million in 2026, representing ~34% growth at the midpoint over 2025. That’s aggressive; let’s be conservative and cut it in half:

• Revenue Growth: ~20% annually (vs 31% in 2025, 44% in H1 2025)

• Margin slightly expanding as LaaS scales

• Minimal share count dilution (NCIB active)

• Conservative EPS Growth: ~18% per year

B. Shareholder Yield

• Dividend: ~4.8%

• Buybacks: ~1.5%

• Total: ~6.3%

C. Valuation Drag/Boost

Current forward P/E is ~6.3x. If you assume the market eventually rewards a business growing 18%+ with a more reasonable 12x P/E (still cheap by any standard):

• (12/6.3)^(1/5) - 1 = +14% per year tailwind from multiple re-rating alone

If the market stays permanently skeptical and P/E stays at 6.3x: +0% from multiple, but you still get the earnings growth and yield.

D. The Final Equation

Bear Case 5Y Return: ~24% annually Base Case 5Y Return: ~38% annually

Even in the bear case no multiple expansion, economy stays tough the math is hard to ignore. The S&P 500 historically delivers ~10% per year. You’re getting 2–4x that if the business just survives and executes.

4. MY PROPRIETARY INSIGHT

Here’s the thing nobody talks about: Propel is quietly transforming its business model in real time, and the market is still pricing it like the old version.

The old Propel: lend money to risky borrowers, collect high interest, write off the ones who don’t pay, repeat. Capital-intensive, cyclical, scary in downturns.

The new Propel: build an AI-powered credit intelligence platform, let other people’s capital fund the loans, collect platform fees. Management expects Lending-as-a-Service to deliver triple-digit growth and approach 10% of total revenue by Q4 2026. That’s a fundamentally different margin and risk profile and it trades at an entirely different multiple in every comparable fintech.

The market is valuing Propel at 6.3x forward earnings. Upstart Holdings which does a similar “AI underwrites loans, bank partners hold them” model has traded at 30–60x. Even discounting for Propel’s smaller scale and Canadian listing, the gap is absurd if the LaaS pivot succeeds.

Here’s my proprietary comparison: when you plot Propel against its nearest comps on the axis that actually matters growth rate vs. valuation multiple the picture becomes embarrassing.

The dashed line is the “fair value” reference a stock growing at 31% would reasonably deserve a P/E somewhere around 20x. Propel is at 6.3x. The entire peer group sits above the line. Propel is the one anomaly sitting far below it.

There’s a historical pattern worth flagging: every time Propel has traded below 10x trailing earnings since becoming profitable in 2022-2023, it has subsequently re-rated significantly higher. We’re currently at ~11x trailing, ~6x forward. The last time forward earnings were this cheap was early 2023, and the stock nearly tripled over the following 14 months before the latest selloff.

The contrarian take the market is missing: the Q4 miss wasn’t a sign of deterioration it was a sign of accelerating growth. You provision now for the loans you originate now. Late-quarter origination growth, particularly in December, required upfront provisioning under IFRS accounting, while the associated revenue will be recognized over future periods. A company that didn’t grow wouldn’t have this problem. The market punished Propel for investing aggressively in its own future.

5. MY TAKE

Sleep Well at Night Score: 5.5/10

This is not a 9/10 sleeper. It’s a high-conviction, higher-volatility situation. The business quality is genuinely improving, but the macro environment adds real uncertainty.

What Excites Me

• The LaaS pivot is a hidden re-rating catalyst. If Lending-as-a-Service hits 10% of revenue by year-end and continues scaling, Propel deserves to trade like a software platform, not a subprime lender. That multiple difference alone is worth 100%+ in the stock.

• Ten consecutive dividend increases, now yielding nearly 5% on a company still growing revenue at 30%+ that combination is genuinely rare. You’re being paid to wait.

• Propel Bank adds a completely new chapter to the story. Regulatory approval to launch Propel Bank came in December 2025, and the long-term implications cheaper cost of funds, deposit products, expanded lending capacity — are not priced in at all at current levels.

What Worries Me

• The macro timing is terrible. Tariffs, slowing consumer spending, and potential U.S. recession risk are exactly the conditions that cause near-prime borrowers to miss payments. The concern is that this stock works great in a good economy, but not so much when the economy turns and that turn may already be starting.

• Credit loss rates of ~50% of the loan book are a feature, not a bug until they’re not. There’s limited margin for error if charge-offs spike 10-15% above model assumptions.

• Institutional participation is thin. Very limited institutional participation means the stock can move violently on small order flow. That cuts both ways it’s why the discount exists, and why volatility can be stomach-churning.

The One-Liner

A rare fintech trading at utility-company multiples while quietly building a platform business the market is looking at last quarter’s loan losses and missing next year’s fee income.

In 2013, a software developer turned telecom entrepreneur named David Nyland walked into the offices of Constellation Software the most celebrated capital allocation machine in Canadian history with a pitch: give me a mandate and some capital, and I’ll build a vertical market software empire in the telecom and media niche. Constellation said yes. What followed was a decade-long apprenticeship inside one of the greatest compounders ever built, during which Nyland’s team quietly assembled a portfolio of 28 companies across 30+ countries, mostly businesses that telecom operators had been relying on for 20 years and couldn’t imagine living without. In early 2023, Constellation handed Nyland the keys to the car and spun the whole thing out as an independent public company. The Lumine Group era had officially begun.

Fast forward to today and the stock is down 58% from its peak. It was trading at $55 CAD a year ago. As of April 4, 2026, it sits at $23 CAD, with a 52-week low of $17.77. The underlying business, meanwhile, just delivered its best year ever revenue up 15%, operating income up 31%, and free cash flow available to shareholders up a staggering 153%. That kind of disconnect between the business and the stock price is usually noise. But it’s worth digging to understand whether the market has spotted something real, or whether this is one of those rare moments where a high-quality compounder gets mispriced because the fear is louder than the facts.

That’s exactly why we’re doing this now.

1. The Machine

A. The Simple Explanation

Let me paint you a picture. It’s 2003. A mid-size telecom operator in South Africa say, a regional mobile carrier needs software to handle subscriber billing. They hire a small boutique software shop that builds them a custom billing system. The integration takes two years. Every customer record, every payment workflow, every regulatory reporting module is wired into this platform. The software is ugly. The interface looks like it was designed by someone who peaked during the Windows XP era. But it works, perfectly, every time, for 20 years straight.

Now imagine someone offers to sell that telecom a shiny new cloud-native billing system. The demo is beautiful. The vendor promises 40% efficiency gains. The IT team is excited. And then the CFO runs the numbers: the migration would cost $12 million in engineering time, carry a 30% risk of data corruption, and require 18 months of parallel running. The answer is no. It will always be no. The new software never gets bought, not because it isn’t better, but because the old software is too embedded to remove.

This is what Lumine buys. The average business unit in Lumine’s portfolio was 21 years old when it was acquired, and customer retention rates run into decades. Lumine doesn’t buy great software. It buys infrastructure. And infrastructure doesn’t get replaced it gets maintained, managed, and milked for cash, forever.

The business model itself is almost offensively simple: find these sleeping giants of telecom and media software, buy them from retiring founders or distracted corporates at 3–5x EBITDA, plug them into the Constellation operating playbook, improve their margins, and use the resulting cash flow to buy the next one. The companies are never sold. The founders often stay on. The products keep running. Repeat for 30 years.

B. The Moat : Three Layers That Stack

Most businesses have one moat. Lumine has three that reinforce each other, and understanding all three is the real alpha here.

Moat #1: Switching costs so high they’re structurally permanent. The businesses Lumine owns don’t just have high switching costs they have switching costs of a kind that make the concept of churn almost theoretical. BSS (Business Support Systems) is software that manages telecom operators’ billing, customer service, order fulfillment, and product offerings. OSS (Operational Support Systems) manages the network itself real-time monitoring, traffic routing, fault detection. These aren’t productivity tools. These are the nervous system of a telecom. Replacing them is roughly equivalent to a hospital replacing its patient records system mid-operation. It is technically possible. Nobody does it voluntarily. Large-scale BSS renewals can cost hundreds of millions of dollars, making exit from existing vendors practically impossible.

The result: Lumine’s recurring revenue runs at 94%, which isn’t a metric, it’s a confession from customers that they’re not going anywhere.

Moat #2: The Constellation DNA. This is the one most people underappreciate. Lumine didn’t invent its operating model it inherited it from the most successful VMS acquirer in history. Mark Leonard built what one analyst calls “fractal decentralization” a system where capital allocation is itself decentralized to individual platforms, each running the same playbook at smaller scale. Substack Lumine is exactly that: a platform-within-a-platform, running the Constellation operating system with a defined niche mandate. The institutional knowledge that comes with this lineage is impossible to replicate. You can’t just decide to run this model. Constellation maintains a proprietary database of roughly 100,000 potential acquisition targets, where business managers with software backgrounds not finance professionals nurture relationships for years before approaching a seller. Lumine has been building the same kind of proprietary pipeline in its niche since 2013. That’s 12 years of relationship-building with telecom software founders who have never heard of most private equity firms. That pipeline is a moat in itself.

Moat #3: The Carve-Out Expertise. Here’s the one that nobody talks about enough. The spin-outs of Topicus and Lumine weren’t done for optics they were deliberate moves to preserve strategic focus, and critically, to create separate acquisition currencies for deals that Constellation’s own scale made uneconomical. What this means in practice is that Lumine has developed a very specific specialty: buying orphaned software divisions from large telecom vendors who no longer want to run software businesses. In 2024 alone, Lumine completed the carve-out of Nokia’s Device Management and Service Management Platform businesses for up to €185 million, then bought cloud-native 5G assets from Casa Systems out of Chapter 11 effectively acquiring distressed telecom software assets while their original owners were in financial distress. This is not a skill you develop overnight. Nokia trusted Lumine with 500 employees and thousands of global customers because Lumine had already done a previous carve-out with Nokia and earned that trust. This is a compounding institutional advantage, not a repeatable commodity deal.

C. The ROIC Story

Lumine targets a hurdle rate of 20–25% ROIC on acquisitions, based on NOPAT. Their historical portfolio of companies achieved a 27% average ROIC over two years before the spinoff. That’s not aspirational it’s historical. And there’s a reason this is possible when it’s nearly impossible for most acquirers.

The businesses Lumine buys are structurally beautiful from a working capital perspective. Software companies, especially older subscription-based ones, often collect annual licenses upfront. That means customers are effectively providing Lumine with free float cash-in-hand before a single dollar of service is delivered. This negative working capital profile (a concept Topicus’s analysts have flagged as one of the most underrated features of VMS businesses) means Lumine’s acquisitions are frequently self-financing within the first year. Add to that the margin expansion that comes from applying the Constellation playbook disciplined R&D spending, rationalized sales costs, shared back-office functions and you get acquisitions that generate their initial invested capital back in 3–5 years.

The catch, and this matters: ROIC tends to decrease as deal size increases. Lumine’s sweet spot was historically $10–15M deals. As they’ve moved into $185M (Nokia) and $258M (Synchronoss) territory, maintaining that 20-25% hurdle becomes harder. The small-deal machine still runs they’re acquiring at roughly 2–3 deals per year at small sizes alongside the larger carve-outs but the mix is shifting, and the math gets tighter at scale.

D. The WideOrbit Deep Dive The Hidden Crown Jewel

Most analysis of Lumine treats WideOrbit as a line item. It deserves its own section.

WideOrbit was founded in 1999 by Eric Mathewson after he realized the buying and selling of media advertising was shockingly manual and fragmented. By the time Lumine acquired it in early 2023, it had become the system of record for over 5,000 TV and radio stations, processing more than $35 billion in advertising revenue annually.

Put that number in perspective: $35 billion. The entire U.S. local TV advertising market generates roughly $23–24 billion per year. WideOrbit processes more than that because it also handles national networks, cable, and radio. WideOrbit’s WO Traffic platform is used by broadcast station groups to manage more than 90% of U.S. local TV ad revenue.

90% market share in the traffic management layer of U.S. local TV. This is a monopoly with a product name almost nobody has ever heard of. Every time a local TV station sells an ad slot for a car dealership, a law firm, a political campaign the technical infrastructure routing that transaction almost certainly runs through WideOrbit software. The broadcaster can’t function without it. Switching would mean rebuilding the entire revenue operations stack from scratch.

WideOrbit’s CTO has recently described how the company is now building AI agents to handle broadcast advertising’s most time-consuming workflow: makegoods the process of rescheduling commercials that didn’t air as planned, which traditionally consumed enormous amounts of manual labor. This isn’t AI risk. This is AI opportunity. WideOrbit is using AI to sell more productivity to the same captive customers who are already paying for the base platform. Think of it as the Jibbitz on the Croc the core platform is irreplaceable, and AI just lets you sell add-ons at high margin to customers who have nowhere else to go.

E. The Risks Don’t Skip This

Risk 1: Organic growth is almost zero, and occasionally negative. This is the real bear case and it deserves honest treatment. Q4 2025 showed just 1% organic growth after FX adjustments and prior quarters showed negative organic growth. The entire revenue growth story is acquisitive. If Lumine misses a year of deals because prices spike, credit markets tighten, or the Nokia-style carve-out pipeline dries up revenue growth stops. The business still generates cash, but the compounder thesis requires continuous reinvestment at high ROIC, and that requires deal flow. This is genuinely the Achilles heel.

Risk 2: The Synchronoss integration. Synchronoss was Lumine’s first-ever public company acquisition at an enterprise value of $258 million, it’s by far the largest deal in company history. The prior average deal was around $12–15 million. Integrating a public-company-scale business with global Tier-1 telecom operator customers, hundreds of employees, and the cultural complexity of a NASDAQ-listed entity is an entirely different animal. Lumine’s “autonomous operations” playbook works brilliantly for small, founder-led businesses. Whether it scales to a business this size is an open empirical question.

Risk 3: The telecom industry is structurally challenged. The telecom sector faces genuine disruption 5G network virtualization, cloud-native OSS/BSS stacks from hyperscalers, and AI-driven automation are all evolving, and Lumine’s customers are often the older, smaller, more financially constrained operators least equipped to navigate that disruption. If customers shrink, get acquired, or go bankrupt, Lumine’s revenue base erodes. The bull case says the software survives regardless of who owns the telecom. The bear case says a consolidating telecom industry means fewer customers paying fewer bills.

Risk 4: The governance structure. Constellation retains a single “super-voting share” in Lumine, giving CSU permanent control over strategic direction despite not owning a majority of common shares. This isn’t necessarily bad CSU’s track record is impeccable but minority shareholders are structurally reliant on the parent acting in everyone’s interest. Always worth remembering.

2. The Numbers

All figures in USD unless specified. Stock price and market cap in CAD.

Current valuation (April 4, 2026):

LMN trades at CAD $23.00, with a 52-week range of CAD $17.77–$55.00.

Market cap approximately CAD $5.9B (~USD $4.3B).

Enterprise value roughly ~USD $4.1B after netting out $232M in cash against modest debt (D/E at 21%).

Profitability snapshot (FY 2025, USD):

Revenue: $765.7M (+15% YoY). Operating income: $275.7M (+31% YoY). Net income: $118.8M (vs. a $258.9M net loss in 2024). Cash from operations: $236.5M (+106% YoY). FCFA2S: $217M (+153% YoY).

A word on that net income swing. The 2024 “loss” was entirely accounting noise: when Lumine spun out, it issued preferred and special shares to WideOrbit’s founders and Constellation. These converted mandatorily into common shares in March 2024, triggering an ~$87M accrued dividend obligation that was paid in shares, not cash, and showed up as an enormous non-cash GAAP charge in 2024. Post-conversion, the capital structure is now completely clean no preferred, no special shares, no non-cash distortions. Investors can now track the business purely through FCFA2S. This is not a small point. The entire reason FCFA2S exploded 153% in 2025 isn’t because the business went from bad to great in one year, it’s because a structural accounting fiction that was masking real cash generation got removed. The cash was always there. Now you can see it.

Valuation metrics:

• EV/FCFA2S: ~19x (USD $4.1B EV / $217M FCFA2S) the cleanest multiple for this machine

• Forward P/E: ~24x (analyst estimates)

• FCF yield on market cap: ~3.7% (USD ~$217M / ~$4.3B USD market cap adjusted)

• Operating margin: 36% in 2025 vs. ~26% in 2022 10 points of margin expansion in 3 years (see chart)

• EBITDA margin: ~39% (CAD terms, per TradingView)

Versus risk-free assets:

• 10-year Canadian government bond: ~3.3%

• S&P 500 earnings yield: ~4.5%

• LMN FCF yield: ~3.7% below the S&P, above the risk-free rate

At face value this looks uninspiring. But this comparison ignores the reinvestment engine. A business yielding 3.7% in FCF that then compounds that FCF at 20–25% ROIC through acquisitions is a fundamentally different animal than a bond yielding 3.3%. The bond returns 3.3% forever. LMN’s FCF pool grows with each deployment.

Shareholder return structure:

• Dividend yield: 0%

• Buyback yield: 0%

• Total shareholder yield: 0%

• This is intentional and correct. Every dollar returned to shareholders via dividends or buybacks is a dollar not compounding at 20–25% ROIC. The only way this structure makes sense to reject is if you believe management can’t find deals at acceptable returns which is the real bear thesis, stripped down to its core.

Quality indicators:

• Debt/Equity: 21% conservative, especially post-Synchronoss

• Operating margin trajectory: 22% (2020) → 36% (2025) relentless structural improvement

• Revenue CAGR since 2020: ~58% (from ~$80M to $766M) though heavily acquisition-driven

3. The Napkin Math

Let’s build the 5-year return scenario from scratch, showing the work.

Step A: FCF/earnings growth estimate.

The engine has two cylinders: organic and acquisitive. Organic is essentially 0–2% (being generous given recent history). Acquisitive growth depends on how much capital Lumine deploys and at what ROIC. With $217M of FCFA2S in 2025 and a $310M credit facility available, Lumine can deploy $300–400M per year into acquisitions. At a 20% after-tax ROIC, that’s $60–80M of incremental annual earnings power per year of deployment. Over five years, assuming deal flow stays consistent and Synchronoss integrates cleanly, you can build to a mid-teens FCFA2S CAGR without much heroism. Being conservative:

• Organic contribution: ~1%

• Acquisitive contribution: ~10% (roughly ~2 deals/year at $50M avg. enterprise value + one larger deal every other year)

• Operational margin improvement: ~2%

• Total FCF/earnings growth: ~12–13% per year

Step B: Shareholder yield.

0%. Full stop. Everything goes into M&A. No dividends, no buybacks, no yield at all.

Step C: Multiple expansion or contraction.

This is where the real uncertainty lives. At peak ($55 CAD), LMN was trading at roughly 40–50x FCFA2S premium “baby Constellation” pricing baked in. At $23 CAD, it sits at approximately 19–20x FCFA2S. The Constellation family has historically traded at 25–35x FCFA2S when operating well. What happens over 5 years depends entirely on whether the market regains its confidence in the model:

• Base case: Multiple stays flat around 20x as organic growth concerns persist → 0% annual impact

• Bull case: Synchronoss integrates cleanly, organic stabilizes at +2%, multiple re-rates to 28x → (28/20)^(1/5) − 1 ≈ +7% per year tailwind

• Bear case: Organic worsens, M&A deal quality deteriorates, multiple contracts to 14x → (14/20)^(1/5) − 1 ≈ −7% per year headwind

Step D: The Final Equation:

The base case at ~12% is slightly better than the S&P 500 historical average of ~10%. The asymmetry is interesting: the bull and bear cases are not equally probable. The mechanism for bull case realization (Synchronoss integration + organic stabilization + multiple re-rating) is a clear sequence of trackable events. The bear case requires a structural deterioration in deal flow that hasn’t appeared yet. This isn’t a 50/50 coin flip.

4. My Proprietary Insight

The Three Things The Market Is Getting Wrong Simultaneously

Insight #1: The “AI will kill VMS” panic is inverted for Lumine specifically.

The market spent 2025 pricing in the death of vertical market software via AI disruption. The logic is: ChatGPT-like tools will replace legacy enterprise software with flexible AI agents that are cheaper and better. This thesis is reasonable for horizontal enterprise software generic CRMs, basic ERP, productivity tools. It is nearly backwards for what Lumine owns.

David Nyland himself described in his semi-annual letter how AI is already a core capability leading to concrete product innovations, such as Openwave’s “AI Smart Assistant.” The Crocs-and-Jibbitz analogy from the CEO of Chapters Group (a similar VMS acquirer in Germany) is brilliant here: Lumine owns the “shoe” the robust, irreplaceable BSS/OSS core that 20-year customer relationships are built on and AI lets them produce the “Jibbitz” faster and cheaper. AI doesn’t replace the shoe. AI is new Jibbitz. Every AI automation module WideOrbit builds goes on top of the existing traffic management platform that 89% of U.S. local TV broadcasters are already paying for. The platform stickiness doesn’t decrease. The revenue per customer increases.

There’s also a counterintuitive second-order effect: BSS/OSS market consolidation is actually bad for telecom operators, who face fewer vendor choices and higher switching costs as the sector concentrates. Lumine operates in an environment where the structural forces are actively making its customers more captive over time. The market read this as a headwind. It is structurally a tailwind.

Insight #2: The 153% FCFA2S jump is real, not a fluke.

When a metric jumps 153% in one year, the first instinct is: “this is a distortion, find the accounting gimmick.” In this case, it’s actually three legitimate improvements compounding simultaneously, which is rarer and more durable than it looks:

First, the preferred share conversion in March 2024 removed ~$87M of annual cash obligations (dividends on the preferred and special shares) from the FCFA2S calculation. This is structural and permanent. Second, the Nokia Motive and Casa Systems Axyom.Core acquisitions in 2024 added revenue and operating income that only hit a full year of FCFA2S contribution in 2025. Acquisitions always look cheap in Year 2 vs. Year 1. Third, the operating leverage on a fixed-cost base: as Lumine’s revenue grew from $668M to $766M, operating income grew disproportionately from $210M to $276M a 31% operating income gain on a 15% revenue gain. That’s margin expansion compounding on a larger base. None of these three drivers is going away.

Insight #3: The WideOrbit valuation alone is probably worth more than the current stock price implies.

Here’s the napkin math the market seems to have forgotten. WideOrbit, when Lumine acquired it in early 2023, was generating approximately $167M in annual revenue. Two years of Constellation-playbook margin improvement later, assume it’s generating $180–190M in revenue at an operating margin of 35–40% (consistent with mature VMS businesses under this model). That’s ~$65–75M in operating income from WideOrbit alone. Apply a 20x multiple to that conservative for a business with 90% U.S. local TV market share and effectively 100% recurring revenue and you get $1.3–1.5B in implied value for WideOrbit alone.

The current market cap sits at CAD $5.9B (~USD $4.3B). So you’re paying roughly $2.8B for everything Lumine owns that isn’t WideOrbit — the 30+ other portfolio companies, the Nokia/Motive carve-out assets, the Axyom.Core 5G platform, the Synchronoss personal cloud business, and the institutional M&A infrastructure built over 12 years. At the current price, the market is valuing all of that at roughly $2.8B. That feels like a lot of value to get for free.

5. My Take

Sleep Well at Night Score: 6.5/10

The business is 9/10. The valuation overhang from organic growth anxiety, deal execution risk on Synchronoss, and the VMS-sentiment hangover knock it down to 6.5. I sleep fine most nights, but that Synchronoss integration sits in the back of my mind.

What Excites Me:

• The preferred share conversion is the most important event in Lumine’s public history that nobody talks about. You went from a business where $258M of net losses hid $217M of real cash generation to a clean, transparent FCF machine. The first year of “real” financials was 2025. The market hasn’t fully absorbed this.

• WideOrbit is a hidden monopoly. Managing more than 90% of U.S. local TV ad revenue through your software and processing over $35 billion in annual advertising transactions is the kind of market position that would attract a $5B+ valuation on its own in a bull market. Right now it’s being valued implicitly at something like $1.3B as part of a discounted conglomerate. At some point, someone notices.

• The spin-out structure gives Lumine a unique acquisition currency it can offer sellers Lumine stock as rollover equity, creating instant multiple arbitrage between what Lumine trades at and what it pays for acquisitions. As the stock recovers from its selloff, this currency becomes more powerful. A recovering stock price is an acquisition advantage for Lumine in a way it isn’t for most companies.

What Worries Me:

• The organic growth line needs to turn. Right now it’s 0–1%. If it goes negative on a sustained basis because telecom customers consolidate, cut budgets, or churn out of some of the smaller legacy businesses the acquisition machine is running to stand still rather than compound. The businesses age, and aging software businesses eventually lose customers to greener solutions. That clock is always ticking.

• Synchronoss is uncharted territory. The integration of a formerly public company is a different complexity class than Lumine’s typical $12M founder-led acquisition , and the “autonomous operations” playbook that works brilliantly for small businesses hasn’t been tested at this scale. If the first large public-company acquisition stumbles, it could reset the market’s confidence in the model for years.

• The valuation still prices in some optimism. At 19–20x FCFA2S, Lumine isn’t screamingly cheap it’s reasonably valued if the model continues to execute. There’s no margin of safety if execution falters.

The One-Liner:

Baby Constellation at its cheapest price in two years, the plumbing is world-class, the preferred-share accounting fog has finally lifted, and the market is confusing a structural re-rating with a value trap. This is one to watch very closely into the next earnings call on April 30th.

Nobody at a dinner party has ever said “I’m really excited about private credit origination platforms.” Nobody has ever posted an Apollo Global Management meme. No Reddit thread has gone viral about their annuity business. There is no Apollo fandom. The stock doesn’t trend on X. Marc Rowan, their CEO, looks like your accountant’s accountant a calm, methodical man in a navy suit who talks about spread compression with the quiet excitement most people reserve for describing a good sandwich.

And yet, quietly, in the background, while everyone was fighting over Nvidia shares and arguing about whether Tesla was worth $1 trillion, Apollo went from managing $100B to managing $938B in about fifteen years. They just posted record earnings across every single metric in FY25. Their insurance subsidiary wrote $82B in new annuity business in one year. Their default rate on $749B of credit investments over sixteen years, through two crashes, a pandemic, and a rate shock averaged 0.1%. Not 1%. Point one percent.

We’re looking at Apollo right now because the stock is down 22% year-to-date while the business has never been healthier, and the reasons for the disconnect have nothing to do with how they lend money or manage assets. They have to do with a dead billionaire, a gated retail fund, and an industry-wide liquidity problem that just blew up in BlackRock’s face too. Those are real issues. But they don’t touch the core machine. And the core machine is, quietly, one of the most elegant business models in American finance.

Sometimes unsexy is the sexiest thing in the room.

1. The Machine

The simple explanation. Most investment firms have a problem that nobody talks about openly: the money they manage doesn’t actually belong to them. When Blackstone raises a $20B private equity fund, they spend 18 months flying around the world begging pension funds and sovereign wealth funds to commit capital. If sentiment shifts, if returns disappoint, if a scandal emerges those LPs simply don’t re-up next cycle. The capital is episodic. The relationships are transactional. The business is structurally dependent on external goodwill.

Apollo figured out a different plumbing system. In 2022, they fully merged with Athene an insurance company they had built from scratch and the game changed entirely. Here’s how it works: Athene sells annuities to retirees. A retired teacher in Ohio hands over $200,000 in exchange for a guaranteed income stream for the rest of her life. Athene takes that $200,000, invests it into Apollo’s private credit products loans to data centers, aircraft lessors, mid-market companies and earns a spread of roughly 130 basis points above what it promised to pay the teacher. That spread, multiplied across $387B of invested assets, generated $3.4B in Spread Related Earnings in FY25 alone.

The magic is what this capital actually is. Insurance float isn’t a hedge fund. It doesn’t redeem on a bad quarter. It doesn’t get spooked by headlines. It doesn’t ask for its money back when a teacher’s union sends a letter to the SEC. It sits there, stable and growing, funding Apollo’s credit investments year after year. Warren Buffett built Berkshire Hathaway on exactly the same insight insurance float is the cheapest, most patient, most durable capital in finance. Apollo just applied it to private credit instead of public equities.

And then, on top of Athene, they built 16 proprietary origination platforms businesses that source the actual loans themselves, rather than buying them from someone else. MidCap Financial lends to mid-market companies. Atlas SP does asset-backed securities. Wheels does fleet financing. Haydock Finance does equipment lending in Europe. Each platform sits inside the Apollo ecosystem, originating assets at 100-200 basis points of excess spread relative to equivalent public market bonds. Apollo doesn’t just invest capital. It manufactures the assets it invests in. That’s a different business model entirely.

The moat. The moat here isn’t a brand or a patent or a regulatory license. It’s a flywheel that took 35 years to build and cannot be assembled from scratch. You need the origination platforms and those require relationships, sector expertise, and operational infrastructure that takes a decade to develop. You need the Athene float and building an insurance company from nothing requires capital, regulatory approval, actuarial credibility, and time. You need the track record because institutional investors don’t hand $50B mandates to managers without a multi-decade default history. Apollo has 0.1% annualized defaults since 2009 across their corporate credit portfolio. That number is a moat. It took 16 years of work to earn it.

Most competitors trying to replicate this model are buying the pieces separately acquiring an insurance company here, investing in an origination platform there. Apollo grew it organically. The integration is genuine, not bolted on.

The ROIC story. For asset managers, we skip the classic ROIC formula and look at the economics of the fee engine instead. Fee Related Earnings the recurring, management-fee-driven income that doesn’t depend on markets going up grew from $1.9B at Apollo’s 2024 Investor Day baseline to $2.5B in FY25. That’s 32% growth in the most boring, most predictable line item in the business. The target is $5B by 2029. Meanwhile, Adjusted Net Income per share went from $6.98 at the 2024 baseline to $8.38 in FY25 20% growth in one year, on a path to $15 by 2029. These are not hopeful projections from a startup. These are targets from a 36-year-old firm that has hit or exceeded guidance for the last several years running.

The risks. The honest version, not the marketing version.

The Epstein lawsuits are the current elephant. Two teachers’ unions representing over $27B in commitments publicly urged the SEC to investigate Apollo’s disclosures around executive ties to Jeffrey Epstein. Marc Rowan took a $158M pay cut last year partly as a response to the scrutiny. If institutional LPs start reducing commitments even quietly, at the margin the fee base takes a direct haircut that compounds over time. This is not a legal risk. This is a relationship risk, and relationship risk at this scale is genuinely hard to quantify.

The retail credit gating is real friction, but as we’ll explore below, it’s an industry-wide structural problem not an Apollo-specific failure. More on that shortly.

Macro risk: private credit defaults are near historic lows. A real recession without Fed backstops could push defaults to 1-2%, meaningfully compressing Spread Related Earnings for several years.

2. The Sector Nobody Wanted to Be In Until Now

Private credit spent the better part of a decade chasing the same trade: lend to software companies. The logic was seductive and, for a while, impeccable. SaaS businesses had recurring revenues, high gross margins, negative churn, and predictable cash flows. They were the ideal private credit borrower or so the story went. Blackstone’s BCRED built a 26% allocation to software. Blue Owl became one of the largest direct lenders to the SaaS sector in the world. The whole industry crowded into the same trade because the same characteristics that made software companies attractive to venture capitalists sticky revenue, asset-light balance sheets, scalable unit economics also made them look like perfect credit borrowers.

Then AI arrived and started eating those characteristics alive.

Analysts stress-testing software-heavy private credit portfolios are now warning that default rates in the sector could spike to 15% far exceeding the roughly 2% headline rates previously reported by industry indices. The problem isn’t that the loans were structurally weak, it’s that the business models underlying them are being disrupted faster than the loan maturities. A SaaS company with $100M in ARR and 90% gross margins looked like a fortress in 2022. In 2026, with AI tools replacing entire product workflows, that same company is fighting for its life. The equity cushion between the loan and the company’s enterprise value is compressing. Covenant-lite structures mean lenders find out late. And the marks, the valuations private credit managers assign to their portfolios are notoriously slow to reflect reality in illiquid markets.

Blackstone’s flagship BCRED posted its first monthly loss in three years in February 2026, marking down loans including debt linked to SaaS company Medallia. Investors then submitted redemption requests totaling a record 7.9% of BCRED’s assets approximately $3.8 billion in a single quarter. Blackstone honored them in full, which is a testament to their liquidity management, but the pressure exposed just how much the retail semi-liquid structure depends on new inflows covering outflows. The moment that dynamic reverses, the machinery strains visibly.

Blue Owl’s situation was starker. The firm ended regular quarterly liquidity payments in its OBDC II fund entirely, switching instead to periodic payouts funded by asset sales. This isn’t a fund failing, it’s a fund doing exactly what its legal documents allow. But it shattered the implicit promise of accessibility that made these products attractive to retail investors in the first place. When the “semi-liquid” part disappears and you’re left with just “private credit,” the product becomes much harder to sell to a 62-year-old planning for retirement.

Now here’s where Apollo’s positioning becomes genuinely interesting and where Marc Rowan’s comment on the Q4 2025 earnings call deserves to be read as something more than routine management commentary. While competitors were maximizing spread income by lending heavily into software, Rowan said explicitly that Apollo took the opposite approach: they built a $24 billion position in cash, Treasuries, and agencies inside Athene. Not because they couldn’t find yield. Because they chose not to reach for it.

That decision looks very different today than it did eighteen months ago when yield-hungry competitors were piling into SaaS loans at 11-12% and Rowan looked conservative by comparison. The $24B defensive reserve isn’t just a balance sheet number, it’s evidence of an underwriting philosophy that prioritizes not losing over maximizing returns in a hot market. It’s the credit manager’s equivalent of Buffett sitting on $300B in cash while everyone else is buying. It feels wrong until suddenly it doesn’t.

Apollo’s exposure to software in its retail BDC sits at roughly 12% less than half of Blackstone’s 26% in BCRED, and meaningfully below Blue Owl’s sector concentration. But the more important distinction isn’t the retail fund. It’s the $749B institutional credit book and the $387B Athene insurance portfolio, which are overwhelmingly weighted toward what Apollo calls the Global Industrial Renaissance, infrastructure debt, energy transition financing, aircraft and equipment lending, real estate credit, power and utilities. These aren’t the assets that AI is disrupting. A loan to finance a data center power grid or a portfolio of Boeing 737s doesn’t become impaired because ChatGPT got smarter. Physical assets with contracted cash flows and tangible collateral behave very differently in a downturn than loans against software ARR.

This is the part of the Apollo story that almost never gets told in the same breath as the Athene float and the origination flywheel: the firm made a deliberate strategic choice to stay grounded in the real economy at a time when the rest of private credit was floating into the cloud. Marc Rowan has been explicit that Apollo views infrastructure, energy, and physical asset lending as the defining opportunity of the next decade the $75T+ in capex needed for the energy transition, digital infrastructure buildout, and power grid modernization. That’s not a pivot. That’s a 10-year-old house view that is only now being validated by events.

The irony is rich. The unglamorous managers who stuck to bridges and pipelines and aircraft leases while their competitors chased the sexy software trade are now the ones sitting on a $24 billion defensive buffer while the SaaS-heavy portfolios start to crack. Private credit is not a monolithic asset class. The composition of what you lend against matters enormously. And right now, the composition gap between Apollo and several of its most prominent peers is starting to show up in the numbers.

3. The Numbers

Current valuation (March 2026)

Price: ~$110.

Market cap: ~$63B.

The interesting multiple sits on Adjusted Net Income per share $8.38 in FY25 giving you roughly 13x economic earnings on a business growing at 15%+ annually.

The S&P 500 trades at 21-22x earnings and grows at maybe 8-10%. On that comparison alone, something looks off. Either the market knows something about structural impairment that isn’t yet visible in the numbers, or you’re looking at one of the more obvious mispricings in a major financial stock in the past few years. The job of this analysis is to figure out which one it is.

Profitability snapshot (FY25)

Fee Related Earnings: $2.5B the recurring management fee engine, independent of market performance.

Spread Related Earnings: $3.4B the Athene insurance machine capturing the spread between investment returns and policy costs.

Adjusted Net Income: $5.195B total, or $8.38 per diluted share.

AUM: $938B. Total inflows: $228B. Origination volume: $309B. All records.

Valuation metrics

P/E on GAAP EPS (~$5.54): ~20x. P/E on ANI per share ($8.38): ~13x. Earnings yield on ANI: 7.6%. The 10-year Treasury sits at roughly 4.3%. The S&P 500 earnings yield is approximately 4.5%. Apollo’s economic earnings yield offers a 320 basis point premium over risk-free assets the highest relative premium Apollo has offered in several years.

Historical ANI multiple range: 15-25x, with the five-year average closer to 18-20x. You’re buying at 13x today. The last two times the multiple compressed this far March 2020 and October 2022 the stock roughly doubled within 18-24 months as the earnings engine kept printing. History doesn’t repeat, but it often rhymes.

Shareholder returns

Dividend: $2.25 annualized in 2026, up ~10% year-over-year, yielding approximately 2% at current prices. Net buyback yield: ~0.5-1%. Total shareholder yield: approximately 2.5-3%. Apollo targets dividend growth at roughly 50% of FRE growth meaning as the fee engine scales toward $5B by 2029, the dividend follows mechanically.

Balance sheet

Debt/equity ~0.33. Rated A2/A/A by all three major agencies.

Athene carries A1/A+/A+/A+ from Moody’s, S&P, Fitch, and A.M. Best.

This is not a leveraged bet on private credit. It’s a conservatively capitalized financial institution with more regulatory capital than most regional banks.

4. The Napkin Math

A. EPS growth: ~15% annually

Management’s own target takes ANI per share from $8.38 in FY25 to ~$15 by 2029 15.6% compound annual growth. The engine: FRE growing at 20% per year driven by origination volume and the global wealth expansion, SRE at 10% driven by Athene’s asset base growing at ~15% per year, and modest share count reduction from buybacks. I’ll use a conservative 15% given the litigation overhang on FRE.

B. Shareholder yield: ~2%

Dividend yield of ~2% plus net buybacks of ~0.5%, minus RSU dilution of ~0.5%. Net ~2% annually in direct capital return.

C. Valuation impact: flat

Current ANI multiple: ~13x. Historical average: ~18-20x. Full normalization would add ~3% per year in multiple expansion over five years. Given ongoing uncertainty, I’m assuming zero flat multiple for five years. If the lawsuits resolve cleanly and institutional LP relationships stabilize, that 3% per year is entirely optionality that you’re getting for free at current prices.

D. The equation

15% (earnings growth) + 2% (shareholder yield) + 0% (multiple) = ~17% annually in the base case. Bear case with FRE growth impaired to 12%: ~14% annually. Both beat the S&P 500 historical average by a meaningful margin. The downside scenario still outperforms the benchmark. That’s an interesting setup.

5. Proprietary insights : The Gating Drama Is an Industry Problem, Not an Apollo Problem

In early 2026, BlackRock’s newly acquired HPS Corporate Lending Fund (HLEND) restricted investor withdrawals after redemption requests exceeded the fund’s quarterly liquidity cap. BlackRock had just paid $12 billion to acquire HPS Investment Partners in late 2024, positioning it as their flagship entry into the private credit democratization trade. Within months of closing the deal, the crown jewel product was gating investors. The headlines were brutal. The optics were catastrophic the world’s largest asset manager, fresh off its most expensive acquisition ever, immediately running into the exact structural problem that critics of semi-liquid private credit had been warning about for years.

Apollo Debt Solutions had its own version of the same problem slightly earlier redemption requests hit 11.2% of assets in a single quarter, well above the 5% cap, with investors receiving roughly 45 cents on the dollar of what they requested.

Two different managers. Two different funds. The exact same mechanics. That’s not a coincidence it’s a structural feature of how these products are designed, and it was always going to play out this way.

Here’s what the breathless coverage missed: this was entirely predictable, and it was predicted. Semi-liquid private credit funds BDCs, interval funds, NAV facilities were built on a fundamental tension that every serious credit investor understood going in. The underlying assets are illiquid by definition. A direct loan to a mid-market company doesn’t have a bid-ask spread you can lean on. You can’t exit it in a day, or a week, or even a month. But the fund wrapper around it promised quarterly redemptions of up to 5% of assets. That worked fine during the inflow years, when new investor money covered outflows. The moment the cycle turned and net flows went negative, the structure had to gate. It had no choice.

This is not fraud. It is not mismanagement. It is the inherent plumbing of the product, and every manager running one of these vehicles Apollo, BlackRock/HPS, Blue Owl, Ares has the same contractual gate built into their documents. The gate is there precisely because the lawyers and product designers knew this moment would come. The only surprise is that some investors seemed surprised.

Here’s the critical nuance the market is missing: the gated products represent a small fraction of Apollo’s total economic engine. The institutional business, the Athene float, the 16 origination platforms none of that is touched. The $749B institutional credit book is fully locked up in 10-year structures with no redemption provisions. The $387B Athene insurance float is structurally incapable of redemption annuity policyholders don’t send gate notices. The retail product drama is happening in the 20% of the business that faces the public. The 80% that doesn’t is printing records.

The BlackRock/HPS situation, if anything, is mildly good news for Apollo. It confirms that the problem is industry-wide and structural, not specific to Apollo’s credit quality or fund management. It shifts regulatory and media attention partly toward the world’s largest asset manager. And it suggests that when the SEC eventually steps in to regulate semi-liquid private credit products which they will the rules will apply uniformly across all managers, not disproportionately to Apollo.

The deeper point is about what separates Apollo’s model from everyone else caught in this storm. BlackRock paid $12 billion to acquire HPS because they needed a private credit origination capability they couldn’t build organically. They bought the plumbing. Apollo spent 35 years building theirs from scratch Athene, MidCap, Atlas SP, Wheels, all of it grown internally and genuinely integrated. When the retail product hits turbulence, the foundation underneath is structurally different from a firm that bolted on a $12 billion acquisition and immediately ran into trouble. The house is built differently. The storm is the same.

6. Why the Float Changes Everything

Wall Street loves to compare Apollo to Blackstone. Same sector, similar AUM scale, both alternatives giants. The comparison makes intuitive sense. It also completely misses what makes Apollo structurally different and why the two businesses will behave very differently when the private credit cycle eventually turns.

Blackstone’s business model is built on episodic fundraising. They raise a massive real estate or private equity fund, deploy it over five years, generate returns, collect performance fees, and then go back out to raise the next fund. The whole model depends on LP sentiment staying positive and markets cooperating on exit timing. One bad cycle, one prolonged period of poor realizations, one major scandal and the fundraising machine slows. The capital is borrowed from external goodwill, renewed every five years.

Apollo has Athene. Athene wrote $82B in new annuity business in FY25 alone. That $82B doesn’t depend on LP sentiment. It doesn’t depend on institutional CIOs maintaining faith through a difficult quarter. It depends on whether retirees keep wanting guaranteed income which, with over one billion people worldwide approaching retirement age by 2030, seems like a fairly durable trend. The capital just keeps showing up. It is structurally compelled capital, not discretionary capital. An annuity holder in Ohio doesn’t email Marc Rowan to redeem her policy because she read a negative article about Epstein. She just keeps collecting her monthly check.

Here’s the number that makes this structural advantage concrete: Athene’s gross invested assets are $387B today, growing at roughly 15% per year organically through new annuity sales. At a 130 basis point net spread which has been remarkably stable across multiple interest rate cycles that trajectory implies approximately $5B in annual spread income by 2029, essentially regardless of what happens in fundraising markets, institutional LP sentiment, or the retail BDC channel. Apollo could theoretically raise zero new external capital for the next five years and Athene alone would still compound the earnings base at a meaningful rate.

No other alternative asset manager has this. Blackstone doesn’t. KKR doesn’t. Ares doesn’t. The closest analogue anywhere in finance is Berkshire Hathaway’s insurance float, which Buffett has described repeatedly as the single most important structural advantage the firm has ever had. The float funds investments at near-zero cost, never redeems, and grows as the insurance business writes new policies. Apollo built the exact same structure in private credit. The market is pricing it as if it were just another asset manager dependent on the kindness of institutional strangers every fundraising cycle.

And then consider what that float is actually invested in right now: not software loans with AI disruption risk, not covenant-lite SaaS debt, but a deliberately conservative $24B position in cash, Treasuries, and agencies that Rowan described as defensive positioning while competitors were reaching for yield. The float is patient. The float is permanent. And the float is sitting on $24B of dry powder at exactly the moment when distressed private credit assets are likely to start appearing at attractive prices.

That is the unsexy truth at the center of this investment thesis. While the headlines scream about Epstein and gated funds and BlackRock’s embarrassing HLEND situation, the actual engine of Apollo’s business the retirement savings of millions of Americans, steadily accumulating inside Athene, being deployed at 130 basis points of spread into real assets that AI cannot disrupt is quietly compounding. It has been doing exactly this, without fanfare, for over a decade.

Sometimes the most boring thing in the room is also the most powerful.

7. My Take

Sleep score: 7/10

The business deserves an 8. The current environment deserves a 5. I’m splitting the difference at 7. The machine is excellent. The moment is uncomfortable. Both things are true simultaneously.

What excites me:

• 13x economic earnings on a 15%+ growth compounder is not a normal valuation. The market is pricing short-term reputational noise on a long-duration asset. If your time horizon is five years or longer, the entry point looks genuinely compelling and the downside case still beats the S&P 500.

• The Athene float is the most underappreciated structural advantage in alternative asset management. $387B of permanent, stable, insurance-linked capital growing at 15% per year, generating 130 basis points of spread income, requiring zero LP relationship management. You cannot clone this in five years or even ten. It took 35 years to build.

• The BlackRock/HPS gating drama inadvertently clarifies the competitive landscape. When the world’s largest asset manager immediately runs into trouble with its freshly acquired private credit platform, it highlights just how difficult it is to replicate what Apollo built organically. You cannot buy your way into this business model. The incumbent advantage is structural, not circumstantial.

What worries me:

• The Epstein overhang is not going away on a predictable timeline. Institutional LP decisions happen slowly and quietly. You may not see the damage in reported AUM figures for 12-18 months, and by the time it’s visible in the numbers, the relationship deterioration has already compounded in the background.

• Semi-liquid retail private credit is going to attract serious regulatory attention now that multiple prominent managers have had visible gating events in quick succession. When Apollo, BlackRock/HPS, and Blue Owl all hit the same structural wall within months of each other, regulators feel compelled to act. If quarterly liquidity windows get restricted by rule, the fastest-growing distribution channel gets structurally capped at precisely the moment Apollo has invested the most in building it out.

• The 0.1% default rate is a beautiful number that cannot last forever. It reflects an extraordinary credit cycle, not permanent gravity. When conditions normalize and they will, Spread Related Earnings will compress, and a market currently pricing Athene’s spread income as near-riskless will need to reprice it as something considerably more complicated.

The one-liner:

Apollo is the toll booth on the highway of global capital formation unglamorous, indispensable, and quietly collecting money whether you notice it or not. The retail fund drama is noise; the Athene float is signal. The stock is on sale because a dead billionaire and an industry-wide liquidity hiccup that hit BlackRock even harder made the headlines. The machine doesn’t care about either.