Wall Street Banks Begin Trading $35 Billion AI Chip Financing Package
N.R. Finch
Bank of America and Morgan Stanley have started trading initial portions of a $35 billion financing package backing Broadcom and Anthropic's AI infrastructure buildout — one of the largest private credit deals ever — signaling that AI compute financing is breaching the wall between private and public credit markets.
What is this $35 billion actually buying?
The borrower is a special-purpose vehicle — a shell entity set up solely for this deal. It uses the funds to purchase custom AI chips co-developed by Google and Broadcom, then leases them to Anthropic.
This means → Anthropic never owns the chips and carries no debt. It gets compute through a lease — the financing risk sits with the vehicle and the credit investors behind it.
In plain terms = think aircraft leasing: the airline doesn't buy the plane — a dedicated company buys it and rents it out.
Who is running this deal?
Apollo Global Management and Blackstone jointly arranged the full $35 billion package. Bank of America and Morgan Stanley serve as co-placement agents for a $240 billion tranche.
Apollo is also marketing the debt to institutional investors. The firm launched its own trading desk in late 2024, has been hiring from Wall Street banks, and had traded $15 billion in private assets by the end of Q1.
This means → Apollo is both the deal's arranger and a secondary-market maker — private-credit giants are moving onto traditional investment-bank turf.
How does the money flow — and how does it trade?
The deal uses a delayed-draw structure: the borrower can draw funds in roughly 16 tranches over about a year, with each draw tied to chip delivery milestones.
In plain terms = the full $35 billion doesn't land at once. Chips are delivered in batches; funding follows — lenders' risk tracks the hardware delivery schedule.
Once drawn, the debt can trade on the 144a market — a venue where qualified buyers such as insurers and mutual funds trade debt securities. Bloomberg-sourced dealer quotes put the bonds at roughly 100 to 100.5 cents on the dollar, near par.
What does this signal for the broader market?
This reflects the relentless capital appetite of tech companies locked in a compute arms race — a single financing now reaches the $35 billion mark.
Wall Street banks actively trading this paper on the secondary market marks a milestone: AI infrastructure financing is breaking through the barrier between private and public credit.
This means → AI compute deals that once belonged exclusively to large private-credit funds are opening up to a wider pool of institutional investors. Once a liquidity channel is established, follow-on deals could be even larger.
Content is for reference only, not financial advice.