Broadcom, Oracle, and SpaceX Turn to Wall Street for Billions to Fund AI Chip-Buying Spree

Broadcom, Oracle, and SpaceX are turning to Wall Street for tens of billions of dollars to finance the purchase of artificial intelligence (AI) chips, highlighting the enormous capital requirements behind the intensifying global AI race.

The companies are pursuing separate financing arrangements, but the rationale is similar: demand for AI computing is growing so rapidly that securing enough chips and data-centre capacity moving forward could increasingly require outside capital rather than relying solely on operating cash flow.

The deals also show how AI infrastructure is becoming a major credit market, with private-equity firms, asset managers, and banks financing hardware that can cost billions before it generates revenue.

Broadcom Seeks Financing to Scale OpenAI's Custom Chips

Broadcom is in discussions with Apollo Global Management and Blackstone over more than $50 billion in financing linked to custom AI chips it is developing with OpenAI, according to The Wall Street Journal. The talks remain in preliminary stages, and the eventual size of the debt deal could change.

The financing would support several gigawatts of OpenAI computing capacity. That follows the companies' 2025 agreement to deploy 10 GW of OpenAI-designed AI accelerators, with deployment targeted to begin in H2 2026 and run through 2029.

OpenAI seeks greater control over the hardware powering its models as demand for AI inference and training continues to rise. Custom chips can also be designed around the company's specific workloads, potentially improving cost and performance compared with relying entirely on general-purpose accelerators.

For Broadcom, arranging financing helps turn that demand into a large infrastructure programme without requiring OpenAI to fund the entire build-out upfront.

Oracle Plans Debt Deal Amid Surging AI Cloud Demand

Oracle is also reportedly in discussions with Apollo and Goldman Sachs about financing a major chip purchase. The proposed structure could involve an independent entity purchasing the chips and leasing them to Oracle, allowing the company to expand computing capacity while managing the effect on its balance sheet.

Oracle has already demonstrated why it needs so much capital. Its cloud infrastructure business is expanding rapidly as customers, including OpenAI, Meta, and Nvidia, demand more AI computing capacity.

Oracle reported $638 billion in remaining performance obligations at the end of fiscal 2026, with much of the increase coming from large AI contracts. It also said $75 billion of those contracts involved customers prepaying for GPUs or supplying the hardware themselves.

That means Oracle is effectively racing to build capacity quickly enough to fulfil contracted demand. Financing chips externally can help it do that while preserving cash for other infrastructure requirements.

SpaceX Wants $40 Billion for Nvidia Chips

SpaceX is pursuing an even more direct financing route, seeking about $40 billion to purchase Nvidia AI chips. The proposed package is expected to comprise roughly $10 billion of bank loans and $30 billion of investment-grade debt, with Apollo expected to lead the transaction and Pimco among potential participants.

The motivation is to secure computing power for SpaceX's expanding AI ambitions and related data-centre operations. Elon Musk has also said xAI's Colossus 2 data centre could roughly double its deployment of advanced Nvidia chips by the end of the year.

The scale of the proposed borrowing shows the central problem facing AI companies: advanced computing capacity is becoming an enormous upfront expense, but delaying purchases risks leaving companies without enough chips to compete.

Oracle is seeking to complete its financing deal as early as this year, while SpaceX's proposed $40 billion financing is expected to close in 2027.

The deals illustrate how companies are increasingly using debt, leasing, and other financing structures to fund AI infrastructure. For lenders and investors, that creates greater exposure to the economics of AI infrastructure, including the ability of borrowers to generate sufficient returns from increasingly expensive computing assets.