Anthropic Gains 300 MW from SpaceX, Doubles Claude Code Limits

Context At the Code with Claude developer conference, Anthropic announced a partnership with SpaceX to use the full capacity of SpaceX’s Memphis data center. The deal aligns with Anthropic’s push to expand AI services for paying subscribers and SpaceX’s ambition to explore orbital compute.

Key Facts - Anthropic will access more than 300 MW of additional compute power, a scale sufficient to run large AI models continuously. - The agreement covers SpaceX’s Colossus 1 supercomputer, which houses over 220,000 NVIDIA GPUs, including H100, H200 and the newer GB200 accelerators. - Usage limits for Claude Code Pro and Max users are now doubled, extending the five‑hour window and removing peak‑hour reductions that previously throttled performance. - Opus API limits, the interface for Anthropic’s flagship model, have been raised, allowing higher request volumes. - Anthropic signaled interest in collaborating on “multiple gigawatts” of orbital compute, a concept that could bypass terrestrial constraints on power, land and cooling.

What It Means The influx of 300 MW positions Anthropic to handle more concurrent workloads and offer developers longer, uninterrupted coding sessions with Claude Code. Removing peak‑hour caps eliminates a common bottleneck for enterprise users, potentially boosting subscription upgrades to Pro and Max tiers.

SpaceX’s disclosure of Colossus 1’s GPU count underscores the scale of its hardware investment, suggesting the company can support not only its own launch‑related AI needs but also external partners. If the orbital compute vision materializes, both firms could pioneer a new class of space‑based data centers, reducing latency for satellite‑linked services.

Watch for Anthropic’s next performance benchmarks and any formal steps toward building the proposed gigawatt‑scale orbital infrastructure.