- Anthropic SpaceXAI Colossus 1 deal adds 300MW and 220,000 GPUs for Claude.
- xAI folds into SpaceXAI as Anthropic expands compute.
Anthropic has signed an agreement with SpaceXAI to use Colossus 1, the AI supercomputer cluster originally built by xAI, as part of its infrastructure for Claude . The companies announced the deal on Wednesday. Anthropic said the capacity will support Claude Pro and Claude Max users.
Anthropic said it will use all compute capacity at Colossus 1. The agreement gives it more than 300 megawatts of new capacity and more than 220,000 Nvidia GPUs within the month.
The company said the added capacity is tied to higher Claude Code limits for paid users and higher API rate limits for Claude Opus models. Anthropic said the SpaceX agreement and other recent compute deals allowed it to double Claude Code’s five-hour rate limits.
The change applies to Pro, Max, Team, and seat-based Enterprise users. Anthropic also removed peak-hour limit reductions for Claude Code on Pro and Max plans.
Anthropic co-founder Tom Brown said on X that Claude inference would begin moving onto Colossus in the coming days. He said the company needed more physical infrastructure to keep up with AI workloads and thanked SpaceX for the partnership.
Elon Musk said in a separate post that discussions with senior Anthropic staff had shaped his view of the company’s approach to AI development. He said the people he met were competent and focused on responsible development.
Musk said SpaceXAI had moved training to Colossus 2, the cluster being used for future Grok models.
xAI folds into SpaceXAI
Musk said on X that xAI would be dissolved as a separate company and that its AI products would operate under SpaceXAI. SpaceX acquired xAI earlier this year in an all-stock deal that valued xAI at about US$250 billion.
The deal brought the Grok developer and its Colossus infrastructure under the rocket company.
Anthropic has faced scrutiny over its restrictions on some military uses of Claude. The dispute centred on Anthropic’s restrictions against uses such as mass domestic surveillance and fully autonomous weapons. Pentagon officials said AI systems used by the government must allow any lawful use.
The compute deal comes after Musk folded xAI into SpaceX earlier this year. He has said that power supply and cooling are constraints for large AI data centres . He has also pointed to land availability as another limit.
Colossus 1 was built for xAI’s AI operations. Anthropic will now use the site’s compute capacity for Claude, while SpaceXAI trains future Grok models on Colossus 2.
Musk had previously criticised Anthropic, calling it “misanthropic” after the company raised US$30 billion at a US$380 billion post-money valuation.
Compute deals expand
Anthropic said the SpaceXAI deal adds to its existing infrastructure relationships. Those partners include Amazon, Google, Broadcom, Microsoft, and Nvidia. The partnerships support computing resources for training and inference workloads.
Anthropic has also announced several other infrastructure agreements. These include up to 5 GW of capacity with Amazon and a 5 GW agreement with Google and Broadcom.
The company has also announced US$30 billion of Azure capacity through Microsoft and Nvidia, along with a US$50 billion investment in US AI infrastructure with Fluidstack.
The International Energy Agency expects data centre electricity consumption to roughly double from 485 TWh in 2025 to 950 TWh in 2030. It said electricity consumption from AI-focused data centres is growing faster than the broader data centre sector and is projected to triple over the same period.
Orbital compute enters the discussion
The agreement also includes Anthropic’s stated interest in working with SpaceX on multiple gigawatts of orbital AI compute capacity.
SpaceXAI said terrestrial power and cooling infrastructure may not scale fast enough for future AI systems. It also cited available land as a constraint.
Space-based data centres would require solutions for launch costs, hardware maintenance, and thermal management. They would also need reliable networking and system maintenance.
Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events , click here for more information.
Tech Wire Asia is powered by TechForge Media . Explore other upcoming enterprise technology events and webinars here .