The $35 Billion Bet on AI Infrastructure: What It Means for Markets and Investors

The Deal at a Glance

Apollo Global Management and Blackstone are structuring a financing arrangement tied to Broadcom that would total approximately $35 billion, according to The Wall Street Journal (published 9 June 2026). The capital is earmarked for AI infrastructure development — the physical networks, data centers, and chips that power artificial intelligence. If completed at that scale, the transaction would rank among the largest private credit deals ever assembled.

The mechanics are still being worked through, but the headline figure signals where the biggest pools of money outside the traditional banking system are heading as the AI infrastructure buildout accelerates.

Who Is Doing What

Apollo and Blackstone are among the world's largest alternative asset managers — firms that manage money for pension funds, insurance companies, endowments, and wealthy investors, operating outside the traditional banking system. Both have built substantial lending and infrastructure divisions over the past five years. Apollo's credit arm is one of the largest non-bank lenders globally. Blackstone's credit and infrastructure franchises give it similar firepower.

Broadcom is a semiconductor company that has recently become critical to AI infrastructure. The company supplies custom chips (called XPUs) to hyperscalers — the massive tech companies like Google and Meta that build AI data centers. Those relationships give Broadcom long-term revenue contracts, which is exactly what lenders need to underwrite a deal this large.

The WSJ report does not specify whether this will be a simple lending facility or a more complex arrangement, but the involvement of two firms with different strengths suggests a syndicated deal — multiple lenders combining capital to handle an amount no single lender could take on alone.

Why Private Credit, and Why Now

To understand this deal, it helps to know the difference between public and private debt. Public debt — bonds issued on the open market — comes with strict disclosure requirements, published pricing, and syndication timelines that can take months. Private credit arrangements, by contrast, are bilateral or club deals negotiated directly between borrower and lender. They offer flexibility: custom covenants (rules the borrower must follow), flexible drawdown mechanics (control over when money is deployed), and confidentiality around commercially sensitive information.

For an AI infrastructure company executing complex, multi-year buildouts, certainty and speed often matter more than saving a fraction of a percent on interest. When you are racing to commission new data center capacity, months matter.

The bigger picture here is one of market evolution. Private credit assets globally now exceed $2 trillion, and managers are actively hunting for places to deploy capital at the largest scale. AI infrastructure fits the bill perfectly: projects have long asset lives, revenue comes from investment-grade hyperscalers (the safest tech companies), and the capital requirements are enormous. This mirrors a pattern we saw in the mid-2010s, when telecom carriers sold tower and fiber assets to infrastructure investors seeking long-term, inflation-linked cash flows. The playbook is the same; the assets are just different — GPU clusters instead of steel towers.

The Scale Question

Thirty-five billion dollars is a genuinely outsized number. To put it in perspective: the largest syndicated loans in traditional markets in 2025 typically ranged from $5 billion to $15 billion. A single corporate bond issuance rarely exceeds $10 billion. A $35 billion private credit commitment — even if drawn over several years — compresses what would ordinarily take years of public market fundraising into one negotiated arrangement.

That concentration carries risks. Even for firms with hundreds of billions in assets under management, a $35 billion commitment represents a material chunk of capital. The deal will almost certainly be syndicated — distributed among insurance companies, pension funds, and sovereign wealth funds — rather than sitting entirely on Apollo's and Blackstone's own balance sheets.

Pricing has not been reported, but in the current interest-rate environment, large infrastructure credits of this type typically carry spreads of 150 to 250 basis points over SOFR (a short-term interest-rate benchmark). A basis point is one-hundredth of a percentage point. To translate: a 200 basis point spread on $35 billion equates to roughly $700 million in annual interest income for the lender group at full deployment. That is the economic incentive driving this deal.

Why Broadcom's Role Matters

Broadcom's presence is analytically significant. The company designs custom chips to hyperscaler specification under long-term supply agreements. Those contracts create the revenue visibility lenders require to underwrite deals this large. Unlike speculative AI software ventures, Broadcom's counterparties are investment-grade technology giants with multi-year capital spending commitments locked in.

This credit quality matters. The chain runs like this: hyperscaler makes a multi-year commitment to Broadcom, Broadcom's revenue stream goes to the infrastructure financing vehicle, and the lender gets paid from that stream. That structural credit enhancement — the guarantee built into contractual relationships — is precisely what allows private lenders to deploy at scale without requiring government backing or excessive equity subordination (equity is riskier than debt and sits below it).

Implications for Markets and Investors

If this transaction closes near its reported size, it will likely reset deal-size expectations across the private credit industry. Competition among large asset managers for mega-ticket AI infrastructure mandates has been intensifying since late 2024, with firms like Ares and Blue Owl publicly flagging AI infrastructure as a priority.

The broader implication is structural. Investment-grade companies with contractually locked-in cash flows are increasingly bypassing public bond markets for bespoke private arrangements — not because bond markets are closed to them, but because the flexibility and execution speed of private credit align better with how AI infrastructure gets deployed. That shift has real consequences: for public credit spreads (the additional interest rates companies must pay for public bonds), for traditional bank balance sheets, and for the pension funds and insurers whose capital is being directed into these vehicles.

The details of this specific deal remain sparse as of 9 June 2026. Structure, tenor (how long the loan lasts), pricing, and the identity of downstream co-lenders have not been publicly confirmed. But the direction is clear: the largest alternative asset managers are building the financial infrastructure for the AI era, and the ticket sizes involved are redefining what "large" means in private credit.