Editorial illustration for SpaceX signs compute deal with Reflection AI, founded by ex‑DeepMind researchers
SpaceX signs compute deal with Reflection AI, founded by...
SpaceX signs compute deal with Reflection AI, founded by ex‑DeepMind researchers
Why does this matter? SpaceX is now supplying a fledgling AI lab with the compute power that once fed industry giants. Starting July 1, 2026, Reflection AI will pay $150 million a month for access to Nvidia’s latest GB300 chips and supporting hardware housed in SpaceX’s Colossus 2 data centre near Memphis, Tennessee. The three‑year agreement, running through July 2029, could total $6.3 billion, though either side can pull the plug with 90 days’ notice after the first quarter.
While the figure dwarfs SpaceX’s earlier deals—Anthropic at $1.25 billion per month and Google at $920 million—the contract still marks a sizable commitment for an open‑source‑focused startup. Founded in 2024 by two former DeepMind researchers, Reflection AI pitches its “open‑weight” models as a public‑parameter alternative to closed‑lab offerings such as Anthropic and OpenAI. Recent U.S. restrictions on Anthropic’s proprietary models, Fable and Mythos, have sharpened interest in that approach.
Here’s the thing: Musk has repeatedly stressed that the three‑year term is flexible, noting the contracts can be cancelled at any time. The deal, Reflection says, is among the largest announced for open AI infrastructure to date.
The startup, which was founded in 2024 by two former Google DeepMind researchers, said the compute deal is one of the largest announced open AI infrastructure commitments to date. "Recent events highlight how important open source is to the AI ecosystem, with more nations and enterprises recognizing the risks and costs associated with exclusively depending on closed models," a spokesperson said in an emailed statement. "Our deal with SpaceXAI signals Reflection's strategic importance within the frontier AI ecosystem, and more compute means more runway to build the world's best open models at scale." The Colossus data center was originally built by xAI, a company founded by Elon Musk that is now part pf SpaceX, for its own AI efforts.
Why this matters We see a concrete shift in how open‑source AI projects secure compute. Reflection AI, a startup founded by former DeepMind researchers, has locked in a $150 million‑per‑month agreement with SpaceX for Nvidia’s GB300 chips at the Colossus 2 facility near Memphis, a commitment that could total $6.3 billion through 2029. For developers, the deal signals that large‑scale hardware access is no longer the sole domain of the biggest cloud providers; a launch‑vehicle company is now a viable partner.
Yet the contract also raises questions. The terms allow either side to walk away with 90 days’ notice, and the actual workload that Reflection will run on those chips remains undisclosed. Founders should watch how the open‑source model scales when tied to such massive infrastructure spending, while researchers might wonder whether the promised “abundant source of AI chips” will translate into broader community benefit.
In short, the partnership underscores the growing relevance of open‑source AI, but whether it will reshape development practices is still unclear.
Further Reading
- SpaceX signs compute deal with open-source AI startup Reflection - CNBC
- SpaceX Inks Multibillion-Dollar Computing Deal With Reflection - Bloomberg Law
- Nvidia-backed Reflection lands SpaceX compute deal - Yahoo Finance
- SpaceX has signed a new compute lease deal with open-source AI startup Reflection - X / Sawyer Merritt
- NEWS: SpaceX has signed another major compute deal, this time with open source AI startup Reflection - X / Deirdre Bosa
Related Reading
Browse all Google news →
Business & Startups OpenAI, a Series F San Francisco startup founded in 2015 by eight pioneers
Business & Startups Terminal-Bench 2.0 launches with Harbor, testing any container-installable agent
Business & Startups Zuckerberg Unveils Meta Compute to Build Global AI Infrastructure
Related Topic India proposes licensing and royalty rules for AI training by Google, OpenAI
Open Source
Related Topic Google AI Advisors Let Users Probe Performance with Conversational “Why” Queries
Research & Benchmarks
From Archives AWS launches Continuum and another service to add context, security to AI agents
From Archives Amazon pulls OpenAI drama starring Andrew Garfield after USD 50 B deal
From Archives Nobel laureate John Jumper departs DeepMind for Anthropic after AlphaFold win
Research & Benchmarks
From Archives Google DeepMind uses MITRE ATT&CK to monitor AI agents as rogue employees
Research & Benchmarks