SpaceX inks compute deal with Reflection AI, an open-source AI lab

The business move

SpaceX has signed a substantial compute deal with Reflection AI, an open-source AI lab. Starting July 1, 2026, Reflection AI will pay $150 million each month through 2029 for direct access to Nvidia’s latest GB300 AI chips and supporting hardware. This compute power will run across SpaceX’s Colossus 2 data center located near Memphis, Tennessee.

Why it matters

This multi-year agreement locks in a heavy monthly spend by an AI lab on premium Nvidia hardware housed in SpaceX infrastructure. For Reflection AI, it guarantees priority access to cutting-edge GPU resources crucial for training large AI models and developing open-source AI technologies. On SpaceX’s side, it diversifies revenue beyond its aerospace operations and leverages underutilized data center capacity. The deal signals growing demand for raw compute from new AI players willing to pay upfront for scale. It also pressures cloud providers and other data center operators to compete on price and performance.

Who gains and who gets squeezed

Reflection AI gains a reliable, high-capacity compute pipeline that should accelerate its AI development velocity without building its own data centers. Nvidia bags a billion-dollar-plus chip sale with an embedded multi-year reseller, reinforcing its hardware dominance. SpaceX secures recurring revenue to offset capital expenses on Colossus 2 while locking in a marquee customer that could attract similar deals.

Traditional cloud vendors face added pressure. Deep-pocketed labs and startups with direct deals reduce dependency on commercial clouds, squeezing their margins. Smaller AI players without direct access to low-cost, high-volume GPUs may face higher compute costs or wait times.

What to watch next

Monitor how other AI labs respond—will they seek similar direct chip access or new partnerships? Watch SpaceX’s data center business evolve as it balances its aerospace ambitions with becoming an AI compute vendor. Nvidia’s pricing and supply strategies will be key for competition in the GPU space through 2029. It will also be important to track Reflection AI’s output—whether this deal translates to meaningful advances in open-source AI or just a shift in who controls large-scale AI infrastructure.

AI Quick Briefs Editorial Desk