Ai Engineering 3 min read

SpaceX Seeks $40 Billion in Debt to Buy Nvidia Chips

The Financial Times reported October 7 that SpaceX is seeking roughly $10 billion in bank loans plus $30 billion in investment-grade debt, led by Apollo and Pimco, to fund an Nvidia chip purchase, one of the largest corporate debt raises ever attempted.

SpaceX is seeking about $40 billion in debt, roughly $10 billion in bank loans plus $30 billion in investment-grade bonds led by Apollo and Pimco, to fund an Nvidia chip purchase, per the Financial Times’ report on October 7, corroborated the same day by Axios, Bloomberg, Semafor, and Reuters. The scale deserves a moment: a $40 billion debt raise would rank among the largest corporate borrowings ever attempted, in service of buying accelerators, and the rocket company is the borrower. This is the compute-as-financial-asset era arriving at its logical extreme: chips are now collateral-grade assets that investment-grade debt markets will finance.

Why a Rocket Company Is Buying GPUs

The strategic logic has been assembling all year. SpaceX has positioned itself as compute infrastructure: Gemini enterprise demand drove a $30 billion SpaceX GPU contract, Cursor was absorbed into a SpaceX AI division, and Reflection AI secured a SpaceX GB300 cluster as a tenant. Against that revenue base, borrowing $40 billion to buy the chips that generate the rental income is a data-center REIT playbook executed at launch-company speed. The unusual part is the asset: GPUs depreciate on a two-to-three-year cadence as Nvidia ships new generations, which means the debt must be serviced from compute contracts that themselves depend on the chips staying competitive. It is a leveraged bet that Nvidia’s upgrade treadmill does not strand the collateral.

The Credit Market’s Answer Is the Real Signal

That Apollo and Pimco are leading the investment-grade tranche tells you how the traditional credit market now prices AI infrastructure: as financeable, at investment grade, at this scale, for a company whose core business is rockets. Twelve months ago, GPU purchases were cash expenditures on hyperscaler balance sheets. This month they became: exchange-traded futures for the rental-price exposure, venture debt and securitized structures for the purchase exposure, and now investment-grade bonds for the largest single purchase. Each layer exists because someone believes future compute revenue is predictable enough to underwrite. The Apollo-Pimco endorsement of a rocket company’s GPU fleet is the strongest statement yet of that belief, and its terms, when disclosed, will become the market’s reference for what GPU-backed leverage costs.

The Concentration Risk Nobody Prices

The trade has a visible failure mode that none of the coverage dwelt on: every layer of the AI-financial stack, futures, debt, and equity, is now correlated to the same two variables, Nvidia’s roadmap and AI revenue growth. A $40 billion chip purchase financed with investment-grade debt against two-to-three-year chip lifecycles concentrates refinancing risk at a specific future date, and the H100 and B200 rental indexes will make any softness in GPU pricing public and immediate rather than absorbed quietly in private contracts. The system built to de-risk AI infrastructure has also built the instrument that will price its stress first. Historians of 2008 will recognize the architecture; whether the analogy holds depends on whether AI compute demand has the durability that housing-backed securities did not.

What to Watch

Three things. First, the deal’s final terms: the coupon on a $30 billion investment-grade tranche for a GPU purchase will be the market’s cleanest price on AI-infrastructure credit risk. Second, Nvidia’s side: a $40 billion single-customer purchase changes Nvidia’s own revenue-quality profile, and how it is disclosed will matter. Third, the second-order effect on Musk’s other company: Tesla’s AI compute needs now compete with SpaceX’s for the same capital allocation, a governance question the overlap makes unavoidable. The week’s theme holds: compute has become finance, and finance is now the industry’s fastest-growing product.

Get Insanely Good at AI

Get Insanely Good at AI

The book for developers who want to understand how AI actually works. LLMs, prompt engineering, RAG, AI agents, and production systems.

Keep Reading

Ai Engineering

How to Run NVIDIA Nemotron 3 Nano 4B Locally on Jetson and RTX

Learn to deploy NVIDIA's Nemotron 3 Nano 4B locally with BF16, FP8, or GGUF on Jetson, RTX, vLLM, TensorRT-LLM, and llama.cpp.

Ai Engineering

Compute Futures Launch Today: GPU Rentals Become a Traded Commodity

CME Group and Silicon Data launched Compute futures on October 5, exchange-traded contracts tracking hourly H100 and B200 GPU rental prices, giving AI companies a way to hedge compute costs and giving the market a public reference price for the first time.

Ai Engineering

US closes the cloud loophole: new BIS rule licenses AI compute for China-linked customers and five diversion hubs

BIS's November 16 interim final rule requires a license for advanced AI chips going to any data center that will give China- or Macau-headquartered firms more than 12 million aggregate TPP, adds license requirements for Brazil, Indonesia, Malaysia, Singapore and Thailand, and formally buries the AI Diffusion Rule.

Ai Engineering

Jensen Huang resigns as Nvidia CEO: Colette Kress takes over, and the AI industry loses its backstop of last resort

Nvidia said on November 12 that Huang had resigned as CEO and director, effective immediately, with CFO Colette Kress as interim CEO. The exit puts more than $100 billion in lease guarantees, a $12.9 billion Hugging Face deal and US chip policy in new hands days before earnings.

Ai Engineering

SpaceX Closes $40 Billion in GPU Debt and Predicts Death for Most Neoclouds

Musk said October 25 that SpaceX closed $40 billion in debt financing, mostly callable, for GPUs arriving in the first half of 2027, while predicting a massive shakeout in which most of the 200-plus neoclouds die as credit tightens.