DEVOURED
SpaceXAI signs agreement with Anthropic for massive AI supercomputer access
Anthropic gets exclusive access to SpaceX's entire Colossus 1 data center—220,000+ Nvidia GPUs and 300+ megawatts—doubling Claude Code rate limits as AI labs hit the ceiling of terrestrial compute.
What: SpaceX signed a deal giving Anthropic 100% access to its Colossus 1 facility in Memphis—over 300 megawatts and 220,000+ Nvidia GPUs starting within the month. Anthropic immediately doubled Claude Code's 5-hour rate limits for Pro, Max, and Team plans, removed peak-hour throttling, and raised Opus API limits. Elon Musk met with Anthropic leadership last week and said he was impressed by their focus on ensuring Claude is good for humanity.
Why it matters: AI labs are exhausting terrestrial power grids, land, and cooling capacity, forcing them into unconventional partnerships. Anthropic chaining together multiple multi-gigawatt deals (Amazon, Google, Microsoft, now SpaceX) shows compute scarcity is the defining constraint in the race to scale AI. SpaceX positioning this as a stepping stone to orbital compute via Starlink satellites reveals how space infrastructure is converging with AI infrastructure.
Takeaway: If you're a Claude Pro or Max subscriber, expect noticeably higher usage limits starting this month. API users on Opus models should check for updated rate limit documentation.
Deep dive
- SpaceX agreed to provide Anthropic with 100% access to its Colossus 1 data center in Memphis starting within the month—over 300 megawatts of power and 220,000+ Nvidia GPUs
- Anthropic doubled Claude Code's 5-hour rate limits for Pro, Max, and Team plans; removed peak-hour throttling on Pro and Max; and substantially raised Opus API rate limits
- Deal announced May 6, 2026, with Elon Musk noting he spent time meeting Anthropic leadership last week and was impressed by their commitment to ensuring Claude is beneficial for humanity
- Part of Anthropic's aggressive multi-deal compute strategy including up to 5GW with Amazon, 5GW with Google/Broadcom, $30B Azure deal with Microsoft/NVIDIA, and $50B U.S. infrastructure investment with Fluidstack
- SpaceX framed the deal as validation of its larger vision: orbital AI compute via Starlink V3 satellites capable of hosting workloads in space to bypass terrestrial power and cooling limits
- Immediate user impact: fewer usage restrictions, faster Claude responses, and support for heavier workloads on Pro/Max subscriptions and API
- AI scaling is fundamentally bottlenecked by terrestrial power grids, land availability, and cooling infrastructure—driving labs toward data center partnerships and alternative compute architectures
- For SpaceX, the deal converts its rapid data center buildout into revenue while demonstrating capability to deliver infrastructure at world-leading speed and scale
Decoder
- Colossus 1: SpaceX's Memphis data center, one of the largest GPU clusters globally with 220,000+ Nvidia GPUs and 300+ megawatts of power capacity
- Orbital compute: SpaceX's long-term vision to run AI workloads on satellites in orbit via Starlink constellation, bypassing terrestrial power grid and cooling constraints
Original article
SpaceX has agreed to provide Anthropic access to its Colossus 1 data center in Memphis, Tennessee. The plant has more than 300 megawatts of power and over 220,000 Nvidia GPUs - Anthropic has access to all of the compute. Anthropic is doubling Claude Code's 5-hour rate limits for Pro, Max, and Team plans, removing the peak hours limit reduction on Claude Code and Max plans, and substantially raising its API limits for Opus models.
Explore all 73 articles in today's digest
Fresh dev news, AI-digested daily. Never miss a trend at devoured.fyi