SpaceX Inks $6.3 Billion Compute Deal with Open-Source Lab Reflection AI

SpaceX is expanding its role as a massive provider of AI infrastructure by signing a landmark multi-year agreement with Reflection AI. This strategic partnership secures high-performance compute resources for one of the most ambitious open-weight AI projects currently in development.

A Multi-Billion Dollar Bet on High-Performance Compute

In a move that underscores the growing demand for specialized AI hardware, Reflection AI has entered into a massive compute contract with SpaceX. Starting July 1, 2026, and running through 2029, Reflection will pay $150 million per month to access cutting-edge hardware. The total value of the deal is projected to reach up to $6.3 billion.

The agreement grants Reflection immediate access to Nvidia’s latest GB300 AI chips and supporting hardware. These resources will be hosted at SpaceX’s "Colossus 2" data center located near Memphis, Tennessee. While the contract is substantial, it is smaller than SpaceX’s existing compute deals with industry giants Anthropic ($1.25 billion/month) and Google ($920 million/month). Notably, both parties retain the flexibility to terminate the contract with 90 days’ notice after an initial three-month period.

Fueling the Open-Weight AI Movement

Founded in 2024 by two former Google DeepMind researchers, Reflection AI is positioning itself as the premier open-source alternative to closed-frontier labs like OpenAI. By focusing on "open-weight" models—where trained parameters are publicly released—Reflection aims to provide a transparent and accessible alternative to proprietary systems.

This strategic shift is gaining momentum following the U.S. government’s ban on Anthropic’s closed models, Fable and Mythos. Reflection views this deal as a validation of its mission, claiming it is one of the largest open AI infrastructure commitments announced to date. The company argues that increased compute runway will allow them to build the world's most capable open models at scale, reducing the global reliance on closed-loop AI ecosystems.

SpaceX’s Pivot to AI Infrastructure Provider

The Colossus 2 data center represents a significant evolution in Elon Musk’s infrastructure strategy. Originally constructed by xAI (now part of SpaceX) to power its internal AI initiatives, the facility is now being leveraged as a high-margin rental service.

As internal xAI pursuits have faced fluctuations, SpaceX has successfully pivoted to renting its massive stockpile of AI chips to the world's leading AI labs. This transition transforms SpaceX from a purely aerospace and transport entity into a critical backbone of the global AI compute supply chain, providing the raw horsepower necessary for the next generation of frontier models.

Key Takeaways

  • Massive Financial Commitment: Reflection AI will pay up to $6.3 billion for access to Nvidia GB300 chips at SpaceX’s Colossus 2 data center.
  • Open-Weight Strategy: The deal provides the necessary compute runway for Reflection to challenge closed-model labs by developing high-scale open-weight AI.
  • Infrastructure as a Service: SpaceX is successfully monetizing its massive AI chip holdings by renting them to top-tier labs like Anthropic, Google, and Reflection.