Apollo Wraps up $35 Billion Debt to Buy AI Chips for Anthropic

Why It Matters

The transaction provides Anthropic with low‑cost capital to scale its compute power, while showcasing how Wall Street is creating bespoke financing tools for the AI boom. It signals a broader shift toward debt‑driven funding for specialized chip and data‑center infrastructure.

Key Takeaways

• $35B debt package funds Anthropic's Google‑custom AI chips lease
• Broadcom backs senior tranches, lowering borrowing costs with its credit rating
• Deal introduces residual‑value support, protecting investors if lease payments default
• Highlights surge in private‑credit financing for AI data‑center hardware

Pulse Analysis

The $35 billion financing secured by Apollo and Blackstone marks one of the largest private‑credit deals ever recorded, specifically targeting the AI hardware supply chain. By structuring the loan across three tranches and securing Broadcom’s credit backing for the senior notes, the syndicate achieved coupons as low as 1 percentage point over Treasuries. This low‑cost debt enables Anthropic to lease Google‑designed tensor processing units, accelerating its compute capacity without diluting equity, a model increasingly favored by fast‑growing AI firms.

Broadcom’s involvement goes beyond mere credit enhancement; the company also provided a residual‑value support agreement. Should Anthropic miss lease payments, the special‑purpose vehicle can liquidate the chips, and Broadcom will cover any shortfall for senior investors. This protection mirrors similar structures seen in mega‑debt deals for data‑center construction, such as Meta’s Beignet bonds, and reduces risk for institutional lenders like Apollo’s Athene arm, which seeks stable, long‑duration assets.

The broader implication for the market is a clear signal that traditional equity financing is no longer sufficient for the capital‑intensive AI race. Investors are turning to sophisticated debt instruments to fund both the physical infrastructure and the specialized hardware that power generative models. As AI adoption expands, we can expect a cascade of similar financing vehicles, driving deeper integration between technology providers, chip manufacturers, and the private‑credit ecosystem, ultimately shaping the cost structure and speed of AI development worldwide.