Tech

SpaceX secures $6.3 billion compute deal with open-source lab Reflection AI

The agreement marks the third major infrastructure arrangement for the aerospace giant, following contracts with Anthropic and Google, as it leverages its newly public market valuation to monetise AI hardware.

Author
Owen Mercer
Markets and Finance Editor
Published
Draft
Source: TechCrunch · original
SpaceX inks compute deal with Reflection AI, an open-source AI lab
Monthly payments of $150 million to begin July 2026 for access to Nvidia GB300 chips at Colossus 2 facility

SpaceX has entered into a compute agreement with open-source artificial intelligence laboratory Reflection AI, granting the latter immediate access to Nvidia’s GB300 AI chips and supporting hardware at the Colossus 2 data centre near Memphis, Tennessee. The contract, valued at up to $6.3 billion, requires monthly payments of $150 million from 1 July 2026 through 2029.

The deal allows either party to terminate the agreement with 90 days’ notice after an initial three-month period. This flexibility mirrors previous arrangements SpaceX has made with other major technology firms, although CEO Elon Musk has previously downplayed the rigidity of such three-year terms, emphasising their cancellable nature.

Reflection AI, founded in 2024 by two former Google DeepMind researchers, cited the agreement as a strategic endorsement of its open-weight AI model approach. The company stated that the deal represents one of the largest announced open AI infrastructure commitments to date, providing additional runway to develop open models at scale.

The monthly fee of $150 million is significantly lower than the $1.25 billion paid by Anthropic and the $920 million paid by Google for similar capacity. These earlier contracts also run through July 2029, establishing a tiered pricing structure for SpaceX’s external compute leasing.

The Colossus data centre was originally constructed by xAI, a company now integrated into SpaceX, for its internal AI development. As SpaceX’s internal pursuits have shifted, the company has moved from using the facility exclusively for its own purposes to leasing capacity to external AI laboratories, capitalising on its substantial hardware holdings.

This agreement follows SpaceX’s initial public offering on the Nasdaq on 11 June 2026, which valued the company at approximately $1.77 trillion. The IPO, priced at $135 per share, raised approximately $75 billion and coincided with gains in US equity markets.

The timing of the deal comes amid increased regulatory scrutiny of closed models, with the US government recently banning Anthropic’s Fable and Mythos systems. This has heightened attention on open-source alternatives, with a SpaceX spokesperson noting that recent events highlight the risks and costs associated with depending exclusively on closed models.

Continue reading

More from Tech

Read next: The Walrus warns of collapsing digital memory as AI erodes search reliability
Read next: Open-source tool claims 97 per cent token savings for AI agents
Read next: Valvoline Unveils August 2026 Promotional Offers for Service and Retail Buyers