Blackstone Plans to Raise $36 Billion to Help Anthropic Lease AI Chips, Expanding the AI Financing Loop

Anthropic is continuously expanding its AI infrastructure. According to Bloomberg, citing sources familiar with the matter, Blackstone (BX-US) has begun engaging potential investors to lead the arrangement of at least $36 billion in debt financing, specifically to help Anthropic cover the costs of leasing AI chips and related computing resources from Google, a subsidiary of Alphabet (GOOGL-US). If the deal is successfully completed, it will surpass the earlier $35 billion AI financing deal and set a new record in the private credit market.

Sources indicate that this new financing will extend the $35 billion private credit transaction arranged about two months ago by Apollo Global Management (APO-US) and Blackstone. That deal was considered one of the largest private credit deals in history, primarily funding Anthropic’s rental of Google’s custom AI chips (TPUs) across five data centers.

The report notes that Google is not only a key strategic investor in Anthropic but also its primary computing power supplier. Google has invested billions of dollars into Anthropic in recent years, and Anthropic, in turn, uses these funds to lease AI chips and cloud infrastructure from Google, creating a highly watched 'capital loop' where Google’s investment partially flows back as lease revenue.

This $36 billion financing is seen as a further extension of that model. Noted analyst and financial podcast host Ed Elson described it as Google investing billions into Anthropic, which then pays Google to lease chips, but still requires additional funding through financial institutions like Blackstone to support its ever-expanding AI compute needs.

In addition to Google and Blackstone, Broadcom (AVGO-US) is also a key player in the overall financing structure. Earlier this year, Broadcom, Blackstone, and Apollo jointly established a platform called 'AI XPV Platform,' specifically designed to provide infrastructure financing for large AI companies. Market sources indicated that the previous $35 billion deal was the platform’s first major financing transaction.

According to previously disclosed transaction structures, Google provides credit support for the senior-most portion of the loan, Broadcom offers partial repayment guarantees, and Morgan Stanley (MS-US) acts as financial advisor and helps arrange the transaction—highlighting how Wall Street is actively designing new financial instruments to meet the massive capital demands of the AI industry.

Anthropic has been accelerating its expansion of compute sources in recent months, not relying solely on Google. Industry research firm SemiAnalysis recently revealed that Anthropic has signed a $10 billion, six-year compute partnership agreement with Bitdeer Technologies (BTDR-US) and startup cloud infrastructure company Volta Infra. This agreement will deploy NVIDIA’s (NVDA-US) latest Vera Rubin AI platform through data centers in Norway, providing 133MW of computing capacity to support the training and inference needs of the Claude series models.

SemiAnalysis stated that this round of compute expansion is primarily driven by the rapid rise in demand for Claude Code, as enterprise customers continue to adopt AI programming tools, forcing Anthropic to continuously increase its GPU and AI chip supply capacity.

Additionally, Anthropic has recently secretly filed initial public offering (IPO) application documents with U.S. regulators. The market expects the company is seeking to go public before OpenAI. Analysts believe that continuously raising large-scale debt financing not only strengthens compute reserves but also helps build a more comprehensive infrastructure layout ahead of an IPO.

From a broader perspective, AI infrastructure construction is rapidly reshaping the global credit market. To meet the hundreds of billions of dollars in demand for AI data centers, chips, and power infrastructure, Wall Street has been continuously launching large-scale structured financing, project financing, and private credit products, attracting investments from asset managers, banks, and insurance funds.

However, the market still has doubts about the return on AI investments. Some recent AI-related financing deals required higher interest rates to close, reflecting that while investors are optimistic about long-term AI demand, they remain cautious about when massive capital expenditures will translate into stable cash flows.

Analysts point out that if this $36 billion financing ultimately closes, it will not only set a new record for AI infrastructure financing scale but also signify a deeper capital collaboration model between Google, Anthropic, and Wall Street financial institutions—potentially becoming a key financing blueprint for large-scale AI compute construction in the AI era.