Published: 2026-05-07
Anthropic Doubles Claude Code Rate Limits via SpaceX Compute Deal
Anthropic announced a major compute partnership with SpaceX, securing 300 megawatts of capacity and over 220,000 Nvidia GPUs. The immediate effect: Claude Code's 5-hour session limits are doubled across all Pro, Max, and Team plans, peak-hour throttling is eliminated entirely, and API output rate limits jumped from 8,000 to 80,000 tokens per minute on tier 1. Builders who abandoned Opus-based workflows due to rate limits now have a compelling reason to retry them.
Source video
"Claude Just Solved Session Limits" by Nate Herk — Watch on YouTube →
Key Takeaways
- Claude Code 5-hour rate limits are doubled for all plan tiers effective immediately — no action needed on the user side.
- Peak-hour throttling (previously cutting limits during weekday mornings) has been removed for Pro and Max accounts.
- API output rate limits increased from 8K to 80K tokens/minute on tier 1; higher tiers also got significant bumps, making parallel Opus agent setups practical.
- The SpaceX deal adds 300MW of compute and 220K+ Nvidia GPUs; Anthropic also has agreements with Amazon, Google, Broadcom, Microsoft, Nvidia, and Fluid Stack.
- Practical recommendation: retest Opus agent workflows that previously failed due to rate limits, and consider reducing the fraction of work delegated to Haiku/Sonnet for cost reasons.
Commands & Code Mentioned
/opus plan
/context