SpaceX has announced a landmark $40 billion GPU financing agreement with Apollo Global Management, marking the largest committed GPU lease in the industry. The deal will supply Nvidia’s next‑generation Vera Rubin chips and related server hardware to power SpaceX’s expanding AI ecosystem, including the Grok chatbot and its cloud compute services.
Deal Highlights
- Financing amount: ~$40 billion for AI chips and data‑center equipment.
- Lease structure: Five‑to‑five‑and‑a‑quarter‑year terms, beginning upon delivery of each equipment tranche.
- Equipment provider: Apollo‑managed entity purchases hardware and leases it to SpaceX.
- Closing timeline: Expected by year‑end, with leases commencing in the first half of 2027.
- Funding diversification: Provides an alternative to traditional debt markets, enabling rapid compute scaling.
Structured Overview
| Aspect | Details |
|---|---|
| Financing Source | Apollo Global Management (private equity) |
| Hardware | Nvidia Vera Rubin GPUs + server infrastructure |
| Lease Term | 5–5.25 years per tranche |
| Start Date | First half of 2027 |
| Capital Impact | Supports $35‑$40 billion spend in Q4 2026 |
| Strategic Benefit | Less than one‑year payback on AI infrastructure investments |
Market Implications
The agreement underscores SpaceX’s strategy to leverage its robust cash flow from Starlink and launch operations to fund AI ambitions. By opting for a lease rather than outright purchase or debt issuance, SpaceX mitigates exposure to volatile capital markets that have tightened for pure‑play AI compute firms. The deal also signals a shift toward diversified funding models for large‑scale AI infrastructure, potentially prompting other neoclouds to explore similar arrangements. For Nvidia, the partnership secures a major customer for its Vera Rubin chips, reinforcing its position in the high‑performance GPU market. Overall, the transaction positions SpaceX to accelerate its AI services while maintaining financial flexibility in an increasingly competitive landscape.