SpaceX continues to solidify its position as a major provider of artificial intelligence compute power, this time by striking a significant deal with Reflection AI. Following previous agreements with industry giants like Anthropic and Google, the open-source AI startup is now leveraging SpaceX’s vast reserves of high-performance AI chips.
This partnership marks a pivotal moment for Reflection AI, as it secures crucial infrastructure to scale its open-weight models. The move underscores a growing trend in the AI landscape where access to top-tier compute resources is as vital as the innovation itself.
A Multi-Billion Dollar Compute Powerhouse
Under the terms of the agreement, Reflection AI will pay $150 million per month to SpaceX, commencing July 1, 2026, and running through 2029. This substantial commitment grants Reflection AI immediate access to Nvidia’s cutting-edge GB300 AI chips and their supporting hardware.
The compute resources will be housed within SpaceX’s advanced Colossus 2 data center near Memphis, Tennessee. Valued at up to $6.3 billion over its lifespan, this deal is a testament to the immense cost and demand for next-generation AI processing power.
While impressive, this contract is proportionally smaller than SpaceX’s earlier agreements with Anthropic and Google, which were reported at $1.25 billion and $920 million per month, respectively. All these contracts share a common end date of July 2029, though each includes a provision allowing either party to terminate the agreement with 90 days’ notice after the initial three months.
Championing Open-Source AI
For Reflection AI, this compute deal is not just about raw power; it’s a strategic move to validate and expand its open-weight AI strategy. The company positions itself as an open-source alternative to the more closed “frontier labs” like Anthropic and OpenAI, aiming to foster greater transparency and accessibility in AI development.
The emphasis on open-weight AI models, which publicly release their trained parameters, has gained considerable traction recently. This shift in focus has been particularly highlighted following the U.S. government’s ban on certain closed models, such as Anthropic’s Fable and Mythos, underscoring the perceived risks associated with proprietary systems.
Founded in 2024 by two former Google DeepMind researchers, Reflection AI considers this their first compute deal, and a landmark one at that. A company spokesperson articulated its significance, stating, “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.”
The spokesperson further added, “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.” This commitment solidifies Reflection AI’s position as a significant player in the evolving landscape of open AI infrastructure.
SpaceX’s Evolving Role in AI Infrastructure
The origins of the Colossus data center, where Reflection AI will run its operations, are rooted in Elon Musk’s ventures. It was initially constructed by xAI, a company founded by Musk and now integrated into SpaceX, specifically for its own ambitious AI development endeavors.
However, as xAI’s internal pursuits faced certain challenges, SpaceX strategically pivoted. Recognizing the immense value of its vast holdings in advanced AI chips, the company transitioned into a crucial provider of compute infrastructure, renting out its resources to leading AI laboratories globally. This adaptability has allowed SpaceX to leverage its technological assets in a rapidly expanding market, supporting diverse AI initiatives across the industry.
Source: TechCrunch – AI
