Broadcom arranges a $63B deal with BofA, Blackstone, Citi, and Morgan Stanley to finance the purchase of chips for Anthropic via a new special purpose vehicle
Context & Ripple Effects
Broadcom’s financing effort progressed from August lender talks to an October report that a Wall Street syndicate was assembling roughly $60 billion for Anthropic and other AI-infrastructure buyers. The $63 billion arrangement formalizes that syndicated chip-finance plan through a dedicated vehicle rather than a conventional vendor sale.
It follows a separate $35 billion TPU leasing package for Anthropic in which Blackstone and Apollo financed capacity and Broadcom backstopped senior debt payments. Blackstone’s presence in both structures makes private credit a recurring part of Anthropic’s compute-access model.
First-order effects
- Anthropic gains a financed route to acquire Broadcom chips through the new special-purpose vehicle, reducing the need to fund the full purchase directly from its own balance sheet.
- BofA, Citi, Morgan Stanley and Blackstone become the financing counterparties for a $63 billion chip purchase, while Broadcom shifts from supplier to arranger of the capital structure supporting demand for its hardware.
Second-order effects
- The arrangement gives Broadcom a tool to support large customer orders when AI buyers’ infrastructure requirements exceed straightforward equipment-purchase budgets, tying chip sales more closely to lender underwriting.
- Blackstone’s participation alongside major banks extends the lender mix used in Anthropic-related compute deals, following its role in the earlier TPU leasing package.
Third-order effects
- If repeatable, special-purpose vehicles can make AI chips resemble financeable infrastructure: vendors originate demand, banks and private-credit firms supply capital, and model developers consume capacity over time.
- The structure increases the industry’s dependence on durable utilization and payment performance from a concentrated group of AI buyers, making credit conditions a more consequential constraint on hardware deployment.
The trend: AI compute is being financed through increasingly structured vendor–bank–private-credit arrangements rather than solely through direct purchases by model developers.