SpaceX is trying to raise $40 billion from Apollo Global Management to buy Nvidia chips, according to The Information. That’s one of the largest single financing efforts aimed purely at AI compute. It also shows how far Elon Musk’s rocket company has moved into the AI infrastructure race. The Information’s report doesn’t include final terms, so treat this as a deal still in progress.
🎯 Situation Assessment
This is about capital, not code. Frontier AI now runs on whoever can secure the most GPUs fastest. A $40 billion check from one private credit giant would put SpaceX in the same compute bracket as the biggest hyperscalers.
Key points:
- Who: SpaceX, the borrower, and Apollo, one of the world’s largest alternative asset managers and a major player in private credit.
- What: A $40 billion raise earmarked for Nvidia hardware.
- Why it stands out: The money is tied to chips, not rockets or Starlink satellites. That points squarely at AI training and inference capacity.
- Status: SpaceX is still seeking the money. It isn’t a closed deal yet.
🧭 How We Got Here
For most of the last two years, the AI build-out ran on two kinds of money: hyperscaler cash flow (Microsoft, Google, Amazon, Meta) and venture equity poured into labs like OpenAI, Anthropic and xAI. That’s changed. Labs and their backers increasingly fund GPU purchases with debt and structured deals, often through special-purpose vehicles that hold the chips as collateral.
Apollo isn’t new to this. The firm has already backed chip financing tied to Musk’s AI ambitions, and private credit firms broadly have become major lenders to data center and GPU projects. Lending against Nvidia hardware has turned into an asset class of its own.
What’s different here is the scale. $40 billion from a single partner is a statement about how much compute Musk’s companies think they’ll need, and how fast.
⚠️ Why It Matters
This is significant for three reasons.
- Compute keeps concentrating. Every massive GPU order shrinks the pool of top-tier chips for everyone else. Smaller labs and startups already pay more and wait longer for Nvidia capacity, and deals like this make that worse.
- Debt is now the default way to fund AI. When a company raises tens of billions specifically to buy chips, it’s betting those chips will earn their keep before they go obsolete. Nvidia ships a new generation roughly every year, so that bet comes with real depreciation risk.
- Nvidia’s grip holds. The money goes to Nvidia hardware, not custom silicon or rival accelerators. Whatever the talk about alternatives, the biggest buyers keep writing checks to Nvidia.
What stands out to me is the borrower. A rocket and satellite company raising this kind of money for GPUs blurs the line between space infrastructure and AI infrastructure. Musk has tied his companies’ futures to compute, and SpaceX’s balance sheet and Starlink revenue make it a credible borrower for a deal this size.
🔍 What Practitioners Should Watch
- GPU availability and pricing. Huge orders from a few buyers can tighten supply and push up cloud GPU rental prices. If you depend on rented compute, plan for that.
- Deal structure. Look for whether this becomes a chip-backed SPV, straight debt or something hybrid. The structure tells you how much risk Apollo thinks it’s taking on.
- Copycat financing. If this closes, expect other AI players to chase similar private credit deals. The playbook spreads quickly once one big name proves it works.
- Model releases. Compute bought today shows up as bigger models 12 to 18 months from now. A buildout this size points to an aggressive release schedule from Musk’s AI efforts.
📡 Outlook
The AI race keeps turning into a financing race. The players who can tap private credit at this scale will set the pace, and everyone else will compete for what’s left. If SpaceX and Apollo close this deal, $40 billion could become the new benchmark for a single compute raise.
The Information has the full report, including more detail on the talks.