Anthropic Doubles Claude Code Usage Limits After Striking Compute Deal with SpaceX
Anthropic has abruptly reversed a planned downgrade to its cheapest Claude subscription tier, instead doubling Claude Code usage limits for all paying subscribers starting today. The dramatic policy shift follows an unexpected alliance with Elon Musk's SpaceX that grants Anthropic access to one of the world's largest AI supercomputers.
The announcement came Wednesday at Anthropic's inaugural Code with Claude developer conference in San Francisco. Company executives revealed a multi-year deal securing compute capacity at SpaceX's Colossus 1 data center in Memphis, Tennessee—home to more than 220,000 Nvidia GPUs.
Key Policy Changes
Under the new terms, Claude Pro, Max, Team, and seat-based Enterprise subscribers will see their five-hour Claude Code usage limits doubled. Anthropic is also raising per-token API rate limits for its powerful Opus model tier. The company further pledged to eliminate peak-hour usage caps on Pro and Max accounts—restrictions that were imposed just weeks ago in late March.
Anthropic did not address weekly limits for Claude Code. A company spokesperson declined to comment on whether further restrictions could return.
Background: Capacity Crisis Averted?
Just last month, Anthropic was weighing the elimination of Claude Code access from its $20-per-month Pro plan entirely. An executive admitted that the company's consumer plans were "not built for token-hungry agentic tools like Claude Code and Cowork." The subscription tiers promised more compute than Anthropic's own data centers could deliver.
The SpaceX deal plugs that gap almost overnight. In addition to ground-based compute, Anthropic said it is interested in leveraging SpaceX's orbital AI computing capacity—a futuristic solution that could further ease terrestrial bottlenecks.
An Unlikely Partnership
The alliance is remarkable given Elon Musk's history of public attacks on Anthropic. The SpaceX CEO, who owns a commanding 42% stake in the rocket company, previously called Anthropic "misanthropic and evil." Yet on Wednesday, Musk posted on X: "No one set off my evil detector." He added that after spending time with senior Anthropic staff, he was "impressed" by their commitment to building AI that is "good for humanity."
"I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed. Everyone I met was highly competent and cared a great deal about…"
— Elon Musk, May 6, 2026
Alongside SpaceX, Anthropic has already struck compute deals with Amazon, Google, and Microsoft—all of which now compete for capacity alongside Musk's AI ventures.
What This Means
For the hundreds of thousands of developers and enterprises relying on Claude Code, the immediate effect is higher daily usage allowances and removal of punitive peak-hour throttling. The deal gives Anthropic the breathing room it desperately needed as Claude Code's popularity surged alongside agentic coding tools.
However, the long-term picture remains uncertain. Anthropic has not committed to making these limit increases permanent. The company's dependency on third-party compute—especially from a rival like SpaceX—raises questions about pricing and reliability down the road. Users should expect usage caps to remain a moving target until Anthropic builds its own data centers or secures more predictable capacity.
For now, though, the crisis is averted—thanks to an unlikely partnership that turned a public enemy into a critical supplier.