Key Takeaways:
In a surprising alliance, Anthropic will take over the entire compute capacity of a data center from Elon Musk's SpaceX, a direct competitor in the AI race.
Anthropic is leasing the entire Colossus 1 data center from rival Elon Musk’s SpaceX, gaining immediate access to over 220,000 Nvidia GPUs to ease capacity constraints on its popular Claude series of AI models. The deal, announced Wednesday, addresses surging user demand that has often left paid subscribers facing temporary usage limits.
"We’re going to need to move a lot of atoms in order to keep up with AI demand, and there’s nobody better at quickly moving atoms (on or off planet Earth),” Anthropic cofounder Tom Brown said on X.
The agreement provides Anthropic with more than 300 megawatts of capacity and a powerful mix of Nvidia's H100, H200, and next-generation GB200 accelerators. This allows the company to immediately double usage limits for its Claude Code subscribers and significantly raise API limits for its high-end Opus models, directly addressing customer complaints about rate limits during peak hours.
The partnership validates the intense demand for AI compute that underpins Nvidia's soaring valuation, while also suggesting Musk's own xAI is struggling for traction. The fact that a world-class data center was available for a major competitor ahead of SpaceX's planned IPO raises questions about the usage of its own AI assistant, Grok.
The deal is a pragmatic solution for two companies in a fiercely competitive landscape. Anthropic, backed by Google and Amazon, gets the immediate, massive-scale compute it needs to compete with OpenAI, sidestepping the long lead times on its other capacity deals with Amazon and Google that won't come online until late 2026 or 2027. For SpaceX, which acquired Musk's xAI venture in February, it represents a way to monetize underutilized assets. Musk confirmed that xAI has already transitioned to its newer Colossus 2 data center for training future models.
Musk, who has previously been critical of Anthropic, endorsed the deal after meeting with its senior team. "No one set off my evil detector," he wrote on X, adding that he was impressed by their competence and commitment to safety. However, he also stated that SpaceX would "reserve the right to reclaim the compute if their AI engages in actions that harm humanity."
Perhaps the most ambitious part of the announcement is a shared interest in developing "multiple gigawatts of orbital AI compute capacity." This aligns with Musk's long-term vision for SpaceX, which he believes can make space-based data centers cost-effective within years, powered by solar energy and launched by its reusable Starship rocket. "The compute required to train and operate the next generation of these systems is outpacing what terrestrial power, land, and cooling can deliver on the timelines that matter,” xAI said in a statement.
This vision of data centers in space is a key component of SpaceX's lofty $1.25 trillion valuation as it heads towards a much-anticipated IPO. The deal with a marquee AI customer like Anthropic could help strengthen the investment case for this highly capital-intensive endeavor.
This article is for informational purposes only and does not constitute investment advice.
