Anthropic, the artificial intelligence company behind the Claude chatbot, has signed a deal to use the full computing capacity of SpaceX's Colossus 1 data centre in Memphis, Tennessee, in a partnership that pairs one of the leading AI startups with Elon Musk's space and infrastructure empire.
The agreement gives Anthropic access to more than 220,000 Nvidia GPUs and over 300 megawatts of compute capacity within the month, a substantial and immediate boost to the company's ability to serve paying subscribers.
Anthropic said the additional capacity will directly benefit Claude Pro and Claude Max users, with the company doubling rate limits on Claude Code for paid plans, removing peak-hour usage caps, and significantly increasing the volume of requests developers can make to its Claude Opus models through the API.
The deal is striking given the public hostility between the two camps.
Musk has previously called Anthropic "misanthropic" and said the company was "doomed to become the opposite of its name," while the Trump administration, which Musk has closely supported, ordered federal agencies to phase out Anthropic technology following a dispute over military use of Claude.
Anthropic had resisted demands for unrestricted military deployment of its AI, citing its safeguards against mass domestic surveillance and autonomous weapons.
Despite that history, Musk said on X that senior Anthropic staff had convinced him the company was handling AI responsibly, writing that everyone he met was "highly competent and cared a great deal about doing the right thing."
He added that xAI's own training workloads have already moved to Colossus 2, a larger cluster being built to surpass Colossus 1, and that the xAI entity would be folded into a new division called SpaceXAI.
Musk noted he would reserve the right to reclaim the compute if Anthropic's AI "engages in actions that harm humanity."
For Anthropic, the timing is significant.
The company has faced mounting complaints from paid subscribers about rate limits and service degradation during peak hours, with demand outstripping its available infrastructure.
Other compute agreements, including a five-gigawatt deal with Amazon, a five-gigawatt partnership with Google and Broadcom, a $30 billion arrangement with Microsoft and Nvidia, and a $50 billion domestic infrastructure investment with Fluidstack, are not expected to deliver capacity until late 2026 or 2027.
The SpaceX deal fills the gap immediately.
Anthropic also disclosed that it has expressed interest in partnering with SpaceX to develop multiple gigawatts of orbital AI compute capacity, a concept that would use space-based infrastructure to overcome the power, cooling and land constraints facing terrestrial data centres.
Related reading
- OpenAI explains why ChatGPT developed a goblin obsession and why it took six months to fix
- Five things you didn't know about ChatGPT 5.5, including the one stat OpenAI doesn't want you to see
- DeepMind scientist argues no AI system will ever become conscious, calling the assumption a 'fundamental fall…
SpaceX said space-based compute could offer near-limitless sustainable power if engineering challenges are resolved, though the idea remains at the exploratory stage.
The announcement was made during Anthropic's second annual Code with Claude developer event, where the company also unveiled a new feature called "dreaming" that allows Claude to review work between sessions and update stored context.
The recap
- SpaceX will provide Colossus 1 compute access to Anthropic.
- The deal adds more than 300 megawatts of capacity.
- Anthropic will ramp up Claude inference on Colossus within days.
