Anthropic is gaining access to one of Elon Musk’s largest AI computing facilities in a surprising partnership between competitors in the AI race.
The company will reportedly use the full capacity of Colossus 1, a massive data centre operated by SpaceX and connected to Musk’s AI company xAI. The facility contains more than 220,000 Nvidia AI chips, giving Anthropic a major boost in computing power for its Claude models.
Anthropic CEO Dario Amodei said the company is racing to expand compute capacity as demand for AI grows.
The partnership is especially notable because xAI’s chatbot Grok directly competes with Anthropic’s Claude. Musk had previously criticized Anthropic earlier this year, calling its technology “misanthropic and evil.”
At the same time, Musk announced that xAI would no longer operate as a standalone company. Future AI development will instead move under a new structure inside SpaceX called “SpaceXAI.”
Despite their rivalry, Musk recently praised Anthropic after meeting with members of its team, saying he was impressed by the company’s approach to AI safety. He added that xAI is transitioning its own workloads toward a newer system known as Colossus 2.
SpaceX also revealed that Anthropic has shown interest in collaborating on future AI data centres in space — an idea Musk believes could eventually solve energy and computing limitations on Earth.
Meanwhile, Anthropic remains in conflict with the US government after refusing to allow its AI systems to be used for autonomous weapons or domestic mass surveillance. The Pentagon reportedly labeled the company a supply chain risk, while Anthropic is fighting the decision legally.
The company recently unveiled a powerful internal model called Claude Mythos Preview, designed to detect hidden software vulnerabilities — including flaws that had remained undiscovered for decades. Anthropic said the system will not be released publicly due to its potential cyberwarfare implications.