SpaceX to power Anthropic’s AI models with Colossus supercomputer

The agreement gives Anthropic access to SpaceX’s Colossus 1 supercomputer infrastructure, strengthening its AI computing capacity as competition intensifies among leading artificial intelligence firms including OpenAI, xAI and other enterprise-focused model developers.

SpaceX has entered into an agreement with Anthropic to provide computing infrastructure support for the company’s AI models, marking a notable development in the rapidly intensifying artificial intelligence race.

Under the arrangement, Anthropic will gain access to Colossus 1, a large-scale AI supercomputer platform developed by SpaceX to support advanced AI training and high-performance computing workloads. The move comes amid growing demand for computing resources among AI developers as companies compete to build more powerful foundation models.

The partnership has drawn attention because Anthropic competes directly with both OpenAI and xAI, Elon Musk’s AI venture. Musk has also been involved in an ongoing legal dispute with OpenAI over the company’s direction and governance structure.

Responding to discussions around the agreement on social media platform X, Elon Musk said he recently met Anthropic executives to better understand the company’s approach to AI safety and governance. According to Musk, those interactions helped convince him that the company was focused on responsible AI development.

Anthropic expands computing capacity

SpaceX describes Colossus 1 as one of the world’s largest AI-focused supercomputing systems. Reports indicate the infrastructure includes more than 220,000 NVIDIA GPUs, featuring advanced accelerators such as H100, H200 and GB200 chips designed for AI model training and inference.

Musk stated that Anthropic’s access to the system became possible after internal training operations shifted toward a newer Colossus 2 platform.

The additional computing power appears to have already benefited Anthropic’s services. The company recently announced expanded usage limits for Claude Code users across several subscription tiers while also easing peak-hour restrictions for some accounts. API limits for certain Opus AI models were also reportedly increased.

Anthropic has emerged as a strong competitor in the enterprise AI market, particularly through its Claude family of models. However, like several AI firms, the company has faced challenges linked to limited compute infrastructure amid rising demand for AI applications.

AI rivalry and future space-based computing ambitions

The partnership is also significant because of Musk’s ongoing criticism of OpenAI, which he has accused of deviating from its original mission of developing open and safe artificial intelligence technologies.

Online reactions to the agreement largely focused on the competitive dynamics between Anthropic and OpenAI, with many users speculating that the deal could strengthen Anthropic’s position in the broader AI market.

Beyond current infrastructure collaboration, Anthropic has reportedly expressed interest in exploring future projects with SpaceX involving orbital AI computing systems. Musk has previously advocated for data centres in space powered by sustainable energy sources, arguing that such infrastructure could eventually support large-scale AI operations with lower environmental impact on Earth.