SpaceX Secures $40B GPU Financing Deal
SpaceX arranges approximately $40 billion in chip and server financing through an Apollo-managed entity, with equipment to be leased over several years.
Building AI data centres no longer depends only on purchasing chips with cash. SpaceX has arranged a financing structure that treats advanced GPUs as long-lived equipment available through leases.
What happened
SpaceX disclosed a binding commitment for approximately $40 billion of financing for AI chips and related server hardware. Under the arrangement, an entity managed by Apollo Global Management will purchase the equipment and lease it to SpaceX.
Each lease is expected to run for approximately five to five-and-a-quarter years, beginning as equipment is delivered. The transaction is expected to close by the end of 2026, subject to conditions, with initial deliveries and leases anticipated in the first half of 2027. The planned equipment includes Nvidia's Vera Rubin generation of chips.
This is a financing commitment, not a completed $40 billion equipment delivery, and it does not mean SpaceX has received an unrestricted cash investment of that amount.
Why it matters
AI infrastructure operators are committing extraordinary sums to equipment that can lose economic value quickly as newer accelerators arrive. Leasing can spread the cash burden over time, but it also creates fixed obligations that must be supported by demand for compute.
For the financing provider, the underlying chips and servers are assets with resale and residual-value risk. Their future value depends on hardware performance, utilisation, energy costs and the speed of technological replacement.
The transaction also helps explain why infrastructure finance specialists are entering AI: GPU deployments increasingly resemble large industrial assets with structured payment streams.
The bigger picture
The next phase of AI competition may be shaped as much by financing access as by model capabilities. Large operators with established revenue, relationships and balance sheets could negotiate structures unavailable to smaller compute providers. The key question is whether future AI demand produces enough cash to cover the commitments being signed today.