Immediate Capacity Boost Addresses Surging AI Demand and User Frustration
Anthropic has finalized a compute deal with Elon Musk's SpaceX, immediately doubling Claude Code rate limits and removing peak-hour restrictions for paid users, leveraging SpaceX's Colossus 1 data center. This strategic partnership addresses critical infrastructure demands for Anthropic's rapidly growing AI services.
The agreement with SpaceX provides Anthropic with immediate access to substantial compute resources, directly impacting user experience. Paid subscribers to Claude Code, including Pro, Max, Team, and Enterprise plans, now benefit from doubled five-hour rate limits. Furthermore, peak-hour usage reductions have been eliminated for Pro and Max users, while API rate limits for Claude Opus models have seen considerable increases.
This influx of capacity directly responds to a surge in Claude usage that outpaced Anthropic's projections. CEO Dario Amodei previously noted an 80x revenue growth against a projected 10x, leading to paid subscribers frequently encountering frustrating rate limits. The SpaceX deal acts as a crucial "compute bridge," providing immediate relief until Anthropic's larger 5 GW capacity contracts become operational.
The deal grants Anthropic access to all compute capacity at SpaceX's Colossus 1 data center in Tennessee. This facility, equipped with 220,000 GPUs, will add over 300 megawatts of new capacity within the month. This expansion is vital for maintaining reliability and performance, which had been strained by high demand, particularly during peak hours.
For SpaceX, this collaboration marks a significant step into the AI infrastructure market. The Colossus 1 data center, previously underutilized with xAI reportedly running at 11% flops utilization, now monetizes a valuable asset. xAI, also owned by Elon Musk, moved its training operations to Colossus 2, freeing up the capacity at Colossus 1.
This move converts an asset that was previously destroying enterprise value into a revenue-generating component. Despite past public criticisms from Musk regarding Anthropic, he expressed being "impressed" with the team, noting Claude "will probably be good." He also included a caveat about reclaiming compute if Anthropic's AI engages in actions harmful to humanity.
Beyond immediate capacity, the deal includes forward-looking plans for space-based compute development. Anthropic has expressed interest in partnering with SpaceX to develop multiple gigawatts of orbital AI compute capacity. This aligns with SpaceX's recent FCC filing, which indicated plans to launch a million satellites to create an orbital data center.
This long-term vision underscores Anthropic's aggressive strategy to secure diverse compute resources. The company has also recently forged multibillion-dollar agreements with other major providers like Amazon and Google. These partnerships collectively aim to ensure Anthropic can meet the escalating demand for its advanced AI models.
