Apollo's John Zito on $900B in private credit, financing AI infrastructure, and why OpenAI may go public sooner than expected

Summary

Apollo Global Management's co-president John Zito makes the case that private credit is the natural financing engine for the current infrastructure buildout — and that most of the tech world hasn't caught up to that reality yet.

Apollo manages over $900 billion in assets, with roughly half sitting on its own balance sheet through Athene, its retirement services business. That structure gives Apollo unusually long-duration capital — the kind that matches 15- or 20-year infrastructure projects that banks, constrained by short-dated deposits, can't comfortably hold. The other half is third-party capital split between a private equity business of over $100 billion and a credit book worth roughly $800 billion.

Private credit isn't what most people think it is

The persistent misconception, Zito argues, is that private credit means small, risky loans to obscure businesses. Apollo's actual deal sheet looks nothing like that. The firm recently closed an $11 billion loan for Intel, a $4.5 billion deal for RWE, a $6.5 billion deal for EDF, and a multi-billion dollar transaction for BP. These are investment-grade, secured, top-of-capital-structure credits for S&P 500 companies. Five years ago, Zito says, he wouldn't have believed those deals were possible in private markets.

The growth-equals-risk reflex is wrong in his view. Private credit is growing fast because tech and infrastructure companies are increasingly asset-heavy — data centers, nuclear plants, manufacturing facilities — and need long-duration bespoke financing that public bond markets and banks weren't designed to provide. The old model was to go public when you needed capital. Now companies stay private longer and are discovering they can optimize their entire capital structure through private markets.

GPU financing and the limits of conviction

Apollo has participated in GPU financing, including a multi-billion dollar deal for Valor and xAI. But Zito is deliberately staying short on duration there — maximum five years — and sitting in the senior tranche. The equity holders are the ones making the bet on what GPUs are worth in year seven or ten. He's blunt about why: anyone who claims to know GPU depreciation curves or release rates that far out isn't being credible. He calls it potentially "the most violent cycle we've ever had" with fat tails in both directions. Layering heavy leverage onto that kind of uncertainty is, in his words, "kind of scary."

OpenAI and Anthropic going public sooner than expected

The more pointed call is on AI lab IPO timing. Zito says scrutiny around off-balance-sheet debt at companies like Oracle and the Neoclouds will push OpenAI and Anthropic to access public capital markets earlier than the market currently assumes. The pressure will require them to tap convert markets, equity markets, and public securities rather than continuing to rely on private off-balance-sheet structures. He frames this not as a prediction but as a logical consequence of the transparency pressure building around how AI infrastructure is being financed.

2025 market outlook

Zito is more optimistic than his credit-investor reputation would suggest. He expects lower rates, a significant wave of M&A, and large technology transactions. He sees the trillions being committed to AI infrastructure ultimately benefiting the consumer. Apollo's private equity book avoided the 2021–22 vintage entirely — nothing met their low-multiple framework — which he implies positions them better than peers for exits over the next few years.

Apollo's thematic AI investing is run through an internal group called Lab 42, led by Rob Bittencourt, who covers the hyperscaler and broader AI ecosystem across both public and private markets.