AI's $235 Billion Funding Test Comes for Wall Street

By Nathan Brooks

Reuters Breakingviews counted three AI-related funding announcements on October 26 with a combined value of up to $235 billion, as banks and specialist lenders took on more AI infrastructure risk. The total mixes one signed equipment lease with two capital-raising plans, not $235 billion of completed financing.

• SpaceX: An Apollo Global Management affiliate will buy AI chips and servers worth approximately $40 billion, then lease the equipment back to SpaceX. The structure gives SpaceX access to computing power without funding the full initial purchase.
• Meta: The company plans to raise up to $95 billion in equity, including about $35 billion through common and convertible preferred stock issuance and up to $60 billion in a non-voting stock offering.
• Anthropic: An initial public offering filing seeks up to $100 billion.

How much risk do banks carry?

A $63 billion financing agreement on October 26 covered Broadcom AI chips and networking equipment for Anthropic. Bank of America, Citi, Morgan Stanley and investment firm Blackstone were involved. Reuters described it as "one of the largest loan deals ever to finance AI infrastructure development."

Who carries the downside?

A specialized legal entity will raise capital from financial institutions, purchase the Broadcom equipment and lease it to Anthropic. Equipment delivery is expected to begin in 2027. The structure separates equipment ownership from Anthropic's AI development and commercial use; lease revenue becomes a crucial repayment source, while the equipment can serve as collateral depending on contract terms.

Approximately $43.5 billion, equal to 69% of the total financing, sits in the highest-priority loan group. Bank of America, Citi and Morgan Stanley arrange that senior portion, while Blackstone provides committed, lower-priority secondary financing.

Broadcom will provide collateral for the remaining value of the financed assets, with coverage equal to the senior financing amount. That support gives lenders asset protection, but it does not remove repayment or depreciation risk.

How does credit access now differ?

Large technology companies still enjoy advantaged access to capital. Smaller AI computing-infrastructure providers face tightening credit conditions and declining stock prices.

For SpaceX's chip-and-server lease, Reuters, citing Semafor, reported that lenders were expected to charge interest at 7.75% to 8.3%. The roughly 6% yield of the BBB-rated corporate bond index served as the benchmark, making those projected rates significantly higher. Those rates are projections, not final announced terms.

Financiers now demand higher returns against uncertain payback periods, technological volatility and uncertain future equipment values. AI funding no longer rests on growth expectations alone; the extra spread compensates banks for financing risk.

How large could AI investment become?

Goldman Sachs analysts estimate that total AI capital spending could reach $7.6 trillion between 2026 and 2031. Meta, Amazon, Alphabet and Microsoft could also issue $420 billion in bonds in 2027 to meet investment needs, according to the forecast. These are estimates and issuance scenarios, not confirmed expenditures or completed bond sales.

What can break the structure?

AI chip values could fall quickly as new generations arrive. Lease repayment depends on Anthropic's business prospects, while creditor protection depends on contract terms and Broadcom's ability to honor its commitments. Broadcom's credit strength does not prevent chip depreciation.

Moody's regards Broadcom as one of the strongest technology companies. The rating agency nevertheless warns that risk will increase if many similar transactions are assembled quickly to meet AI companies' enormous capital needs.

The arrangements draw banks, investment funds and equipment manufacturers into the same financial chain. They share expected financial benefits and contractual obligations, leaving each build-out exposed to the cash generated by AI services.

The decisive variable is AI cash flow. Future pricing and contract structures will show whether leases can repay multibillion-dollar capital outlays at commercially sustainable rates or whether lenders must demand more collateral and higher returns.