Private Markets, Software Repricing and Capital Allocation | Marc Rowan on a16z

ELI5 / TLDR

Marc Rowan runs Apollo, a $1-trillion money manager most people still mislabel as a private equity shop. His actual argument: the stock market has become a bet on ten companies, the bond market is about to become a bet on five banks plus five tech giants, and the only place left to get real diversification is in things that don’t trade on an exchange. Meanwhile companies need to borrow staggering sums to build AI data centers, chips, and robots — and equity can’t fund all of it. Apollo’s whole game is to sit between retirees who need income and corporations that need to borrow, manufacturing the assets in the middle. As a bonus warning: he thinks the private equity industry overloaded on enterprise software is heading for a rough decade because AI just changed what that software is worth.

The Full Story

The firm everyone keeps mislabeling

Apollo is over a trillion dollars in assets, and Rowan spends real energy correcting the record on what it actually is.

Well, actually, Apollo is mostly an investment grade credit firm.

Eighty percent of the money is credit, most of it investment grade — the safe end, lending to large, solid companies. Of the remaining 20%, half is traditional private equity (the thing the firm is famous for) and half is something he calls “hybrid equity.” The private equity piece is roughly $100 billion of a $1.1 trillion whole. So the label “private equity firm” describes about a tenth of the business.

Two jobs underpin everything. One: Apollo is, by his telling, the largest provider of retirement income in the world — through insurance and annuities, it owes money to retirees for decades. Two: it’s a major financier of what he calls a “global industrial renaissance,” the build-out of energy, manufacturing, defense, and data centers. The model is to match one against the other. Retirees need steady long-term income; corporations need to borrow long-term money. Apollo stands in the middle, taking a spread.

Why private markets are the only diversification left

This is the headline argument. Public markets have quietly become extremely concentrated.

10 stocks right now in the US are nearly 50% of the S&P and they’re all levered to the same trend.

Think of it like a casino where most of the chips are riding on one number. That’s been wonderful while the number keeps coming up — but the whole retirement system is now tied to ten companies betting on the same thing (AI). And Rowan says the bond market is heading the same way: a market once spread across ten big banks is consolidating toward five banks plus five tech companies. So whether you hold stocks or bonds, you’re concentrated.

Where do you escape it? The genuinely large, valuable companies of this era — OpenAI, Anthropic, SpaceX, and so on — are private, worth trillions collectively, and most investors own none of them. Hence his pitch: diversification now lives outside the public exchanges.

Making private things behave like public things

There’s a catch. Private assets are traditionally illiquid and slow — priced quarterly, locked in funds for years. The old client was a big institution that tolerated that. The new clients Rowan wants — individuals, insurers, 401(k) plans, traditional asset managers — live in a public-market world and won’t tolerate it.

The notion that they are going to somehow conform to us is just hubris. We are going to have to conform to them.

So Apollo is bolting public-market machinery onto private assets: daily estimated valuations (live by June 30 for the investment-grade products, across all of credit by September), standardized identifiers so the assets can be tracked like public bonds, market-making, and regular price disclosure. The goal is a real ecosystem with price discovery. His reasoning is almost a law of nature:

I’ve never seen a market in the world where you have transparency and price discovery that is not 10 times its size.

Equity might get the same treatment eventually, but not this year.

The “in-between” is where the money is

Rowan’s mental model for finding opportunity: institutions sort their money into buckets — public equity, public fixed income, real assets, “alternatives.” Anything that doesn’t fit a bucket has no natural owner, so it’s mispriced. His best business is precisely the homeless stuff: private but investment-grade credit (too safe to be an “alternative,” but not public, so it falls through the cracks), and “hybrid” equity (safe private equity whose returns are too modest for the alternatives bucket).

In between is almost always the best asset class because there is poor capital formation… there’s no one who is assigned every day as their day job for the risk.

A concrete example: building a modern data center means marrying energy contracts, chips, and an offtake buyer into one financing. It’s creditworthy but it is not a plain ten-year bond, so the public market — which only does plain vanilla — can’t handle it. That complexity is exactly where Apollo earns its keep.

A capital-heavy bet, on purpose

There’s an industry debate over whether asset managers should be “capital light” (just collect fees) or “capital heavy” (put your own money in). Rowan is unapologetically heavy. His logic: Apollo can’t just buy what already exists; it can only invest as fast as it can create new assets. Since each created asset is scarce, he wants to own the upside of it, not merely earn a fee. And a big balance sheet lets him guarantee outcomes — to borrowers and to retirees — which he thinks is worth more in a fast-changing world.

The software reckoning

The sharpest, most actionable claim is a warning. Roughly 30% of the private equity industry over the past decade went into enterprise software. Rowan thinks AI has quietly broken that bet.

I personally expect the returns from private equity in the ground to be disastrous.

The mechanism is simple. Those companies aren’t dying — but they were bought at prices that assumed a future with no AI competitor. Now there’s AI in the picture, so reselling them to the public market or another buyer at the old price is far harder. The prices paid reflected a world that no longer exists. Credit problems in software have already surfaced in the press, and as he puts it, if the credit is troubled, the equity behind it is in far worse shape.

He pairs this with a framework for which jobs and businesses AI hits hardest: anything with a checkable right answer (coding, accounting, trade operations) changes vertically fast, because AI can verify its own work. Anything requiring judgment with no objectively correct answer (“what’s the best Shakespeare essay?”) improves slowly. He expects blue-collar work to rise and white-collar work to decline — politically awkward, especially for big coastal cities.

Lender’s instincts and culture

On lending against a future this uncertain, Rowan reaches for history: the market once lent against Yellow Pages, then TV and radio stations, then cable — each “unkillable” franchise diminished in turn. The discipline that survives all of it: stay diversified, stay senior (first in line to get paid), demand hard collateral, and never underwrite a 20- or 30-year bet — three to seven years is the horizon you can actually judge.

On culture, he spent the better part of a year answering one question — “what makes Apollo Apollo?” — because a firm of 6,000 people can’t run on the same osmosis a small firm did. The ethos he keeps returning to is “play to win,” and its enemy is success itself: companies decline when the desire to win gets overwhelmed by the fear of losing. So Apollo keeps a literal “wall of shame,” and the rule is you don’t get fired for a bad decision — you get fired for not owning it and fixing it.

Every senior professional here has lost money for the firm. If you haven’t, you’re just not doing enough.

Key Takeaways

Apollo is ~$1.1T AUM: ~80% credit (mostly investment-grade), ~10% hybrid equity, ~10% traditional private equity. The “private equity firm” label fits only a tenth of it.
Ten stocks are ~50% of the S&P 500, all levered to the same (AI) trend; the bond market is consolidating toward five banks + five tech firms. Public markets offer little diversification left.
The largest companies of the era (OpenAI, Anthropic, SpaceX) are private; most investors have zero exposure — Rowan’s core sales pitch for private markets.
Apollo’s business = matching long-term retirement liabilities (income owed to retirees) against long-term financing needs of corporations, earning the spread in the middle.
Banks are the best short-term lenders (they borrow short, lend short); public and private capital are the long-term lenders. Public markets only do “plain vanilla” — complex deals go private.
Apollo is rolling out daily estimated valuations on private assets (investment-grade by June 30, all credit by September), plus standardized IDs and market-making, to make private assets behave like public ones for new buyers (individuals, insurers, 401(k)s).
“In-between” assets — private-but-investment-grade credit, safe “hybrid” equity — are mispriced because they fit no institutional bucket and no one owns the risk full-time. That’s Apollo’s edge.
~30% of private equity went into enterprise software over the past decade; Rowan expects “disastrous” returns because those buyout prices assumed a pre-AI future that no longer exists.
AI changes fastest where answers are checkable (coding, accounting, ops); slowest where judgment rules. Rowan predicts blue-collar ascendancy, white-collar decline.
Financing the AI build-out (data centers, chips, robotics, defense) can’t be done with equity alone — it gets sliced into risk tranches, with hard-asset/infrastructure pieces pushed into credit markets.
Lending discipline that survives any disruption: diversify, stay senior, demand hard collateral, underwrite only 3–7 years out.

Claude’s Take

This is a sales pitch, and you should read it as one — Rowan’s job is to convince the world that private markets are where everyone needs to be, and conveniently, that’s what Apollo sells. The “diversification only lives in private markets” line is self-serving: private markets have their own correlated risks (a lot of that private credit is also funding the same AI trend), and “daily estimated value” is still an estimate, not a market-clearing price. Calling something marked-to-model “transparent” deserves a raised eyebrow.

That said, the man is sharp and unusually willing to say things against his own book. The software warning is the standout — “I personally expect the returns from private equity in the ground to be disastrous” is a remarkable admission from someone whose industry made that exact bet, and the reasoning (prices baked in a no-AI future) is clean and hard to argue with. The concentration stats are real and genuinely under-discussed: a retirement system implicitly riding on ten stocks is a legitimate concern, even if his proposed alternative is the thing he’s selling.

Eight out of ten. It’s a clear, structurally honest window into how the biggest private-credit machine actually thinks — the bucket-arbitrage framework and the bank-vs-public-vs-private financing breakdown are worth keeping. Docked points because it’s an a16z-hosted conversation between two people with aligned interests, so the interviewer never pushes back, and the moral-leadership detour, while sincere, is off-topic for the finance thesis. Take the diversification pitch with salt; take the software warning seriously.