SpaceX seeks $40B Apollo led debt to buy Nvidia AI chips

If accurate, a single company borrowing $40 billion largely to buy Nvidia accelerators would be one of the largest discrete bets on AI compute in history. It would confirm that SpaceX intends to be more than a launch and connectivity company, building a vertically integrated compute layer that can power Starlink, internal models and possibly commercial AI services. Strategically, this pulls SpaceX into the same capital intensive arms race as hyperscalers, but with a space infrastructure twist and tighter alignment with Nvidia’s financing ecosystem.

For the AGI race, scale of this magnitude matters because it concentrates even more cutting edge chips in the hands of a small set of actors. A SpaceX data center fleet on top of xAI and existing hyperscaler deployments would further tighten the market for high end GPUs and make it harder for second tier labs and public institutions to access comparable hardware. It also shows how private credit and alternative asset managers like Apollo are becoming core enablers of AI infrastructure, deciding who gets to scale and on what terms. Even if the deal closes at a smaller size, the direction is clear: AI compute is being financed like major energy or telecom projects, not like typical software capex.