Broadcom and Blackstone Line Up Sixty Billion Dollars to Finance AI Infrastructure

Anthropic’s filing did not disclose the loan’s interest rate or conversion price, and the reporting also found no details on a drawdown schedule, collateral or covenants—leaving important details about Broadcom’s credit exposure unclear.
Anthropic said in April that it had expanded its partnership with Google and Broadcom for multiple gigawatts of next-generation TPU capacity beginning in 2027; the compute is intended to power its Claude models.
The financing is being watched amid public backlash against data-center construction, as a sign of whether investors remain willing to fund AI’s infrastructure buildout. Separately, Nvidia announced a partnership with six major finance firms, including Blackstone, to mobilize more than $500 billion for AI and help customers finance chip purchases.
Broadcom is orchestrating a $60 billion financing package to help AI companies like Anthropic buy chips and data-center equipment, marking a dramatic shift in how the semiconductor industry funds its customers. Advisor Perspectives reported that banks are assembling a $42 billion senior-secured tranche and an $18 billion junior tranche led by Blackstone, which plans to commit $9 billion directly. The arrangement underscores how vendor financing has become crucial to winning AI infrastructure deals—and raises questions about credit risk when suppliers lend to their own customers.
Separately, Anthropic's IPO filing revealed that Broadcom agreed to lend the AI startup up to $42 billion via convertible debt to help cover roughly one-third of its $125.2 billion, five-year commitment to lease Google-designed TPU chips. Quartz noted the financing includes no disclosed details on interest rates, conversion prices, collateral, or payment schedules—leaving major gaps in how much credit risk Broadcom is actually taking on.
Broadcom is not just selling chips. It is also leasing computing capacity and now lending money to buy both. AOL reported that Broadcom agreed to lend Anthropic $42 billion in convertible notes as part of a broader deal covering chip supply and equipment. This triple role—vendor, lessor, and lender—is highly unusual and creates built-in conflicts of interest. If Anthropic defaults on the loan, Broadcom could restrict its access to hardware or pricing, trapping the customer.
Anthropic disclosed in its IPO prospectus that this structure presents "potential conflicts of interest" around pricing and compute availability. The company is betting it can afford to pay back Broadcom while also meeting its revenue targets. But if Claude AI adoption slows or competition intensifies, Anthropic could struggle—and Broadcom would face losses on both the loan and its chip sales.
Broadcom's financing package depends on AI companies generating enough revenue to repay billions in debt. But skeptics are watching closely. Seaport Research analyst Jay Goldberg warned that "Nvidia is putting in place a massive amount of its balance sheet, and Broadcom is having to follow suit." The concern: both semiconductor giants are betting their own capital that AI startups will succeed. If the bets go wrong, both firms absorb massive losses.
Robert Leitao, managing partner at Rothschild & Co., voiced similar doubts: "It feels that there's quite a concentrated bet right now on two companies being able to generate enough revenues to support all the financing that's happened." Anthropic itself is losing money—burning more than $8 billion annually while generating just $4.6 billion in projected 2025 revenue. The math does not add up unless Claude AI adoption explodes.
Broadcom is not alone in using vendor financing as a competitive weapon. AOL noted that Nvidia announced a partnership with six major finance firms—including Blackstone, BlackRock, and Apollo—to mobilize over $500 billion in capital for AI infrastructure. Both chip makers are essentially using their balance sheets to lock in customers and guarantee demand for their own products. The strategy works in the short term but creates systemic risk if demand collapses.
Broadcom projects AI chip revenue will hit $115 billion in fiscal 2027 and $230 billion by 2028. Those targets depend on deals like the Anthropic arrangement succeeding and spurring copy-cat financing across the industry. But if data-center expansion slows due to power grid strain, public opposition, or slowing AI adoption, the entire house of cards could crumble. Broadcom is betting its own money that none of that happens.
Anthropic's IPO filing did not disclose critical terms of the Broadcom loan. There is no stated interest rate, no conversion price for the debt-to-equity swap, and no details on how much Anthropic must draw down each year or what happens if it defaults. Quartz reported that collateral arrangements and financial covenants remain undisclosed. These gaps make it impossible for outsiders to assess how much risk Broadcom is actually carrying or how protected creditors really are.
The secrecy is unusual for a public company's financing arrangement. Typical bank loans spell out every detail: repayment schedules, triggers for default, what assets can be seized. But convertible debt between a vendor and customer often gets looser treatment, especially when both parties want to keep terms quiet. That opacity is precisely what worries Wall Street analysts watching this deal unfold. If Anthropic stumbles, no one outside the two companies will know how bad Broadcom's exposure really is.
Publishers
61
Articles
173
Reach
234