Anthropic Secures 220,000 NVIDIA GPUs Through SpaceX Partnership

• Anthropic gains access to over 300 MW of compute power and 220,000 NVIDIA GPUs.
• Through SpaceX’s Colossus 1 supercomputer deal, Anthropic aims to expand Claude’s limits.
• Claude Pro and Max users receive doubled limits and removal of peak-hour restrictions.

Anthropic has signed a deal with Elon Musk’s SpaceX to access Colossus 1, a massive AI supercomputer built in Memphis, Tennessee. The agreement gives Anthropic immediate access to more than 300 megawatts of compute power backed by over 220,000 NVIDIA GPUs.

The partnership was announced on May 6 during Anthropic’s developer event in San Francisco. Anthropic plans to use the added compute capacity to expand Claude AI usage limits and stabilize services that have faced growing pressure from heavy demand.

Claude Usage Limits Expanded

Anthropic confirmed that Claude Pro and Claude Max subscribers will immediately receive higher limits. The company doubled five-hour usage caps for Claude Code across Pro, Max, Team, and Enterprise plans.

Peak-hour restrictions for Pro and Max users are also being removed. Anthropic additionally stated that API rate limits for Opus models are being raised substantially.

The changes come after developers repeatedly complained about rate limits and disruptions as coding agents consumed more compute resources. Anthropic said the average developer now spends at least 20 hours per week using Claude Code.

Colossus 1 Becomes Key Compute Backbone

SpaceX describes Colossus 1 as one of the world’s largest and fastest-built AI supercomputers. The facility was completed in 122 days and runs on roughly 220,000 NVIDIA GPUs, including H100, H200, GB200, and newer GB300 systems.

The infrastructure was originally built for Musk’s AI company xAI before its merger into SpaceXAI earlier this year. Musk stated that SpaceXAI had already shifted its own training workloads to Colossus 2, allowing Anthropic to access Colossus 1 capacity.

NVIDIA also highlighted the agreement, calling it a major step for AI infrastructure expansion. Musk later described the GB300 systems as “the best AI computer.”

Musk Changes Tone on Anthropic

The deal marks a major transformation in Musk’s public stance toward Anthropic. Earlier this year, he criticized the company and accused its models of political and social bias.

Now Musk says meetings with Anthropic’s leadership changed his view. He stated that senior staff convinced him they care deeply about AI safety and responsible development.

Musk added that no one at Anthropic triggered his “evil detector” and said Claude would likely remain safe if the company continued critical self-examination.

The timing matters because Musk is simultaneously locked in a legal battle with OpenAI, accusing the company of abandoning its original mission around safe AI development.

Related: Google’s Anthropic Stake Could Become Its Next Massive Win: Analysts