JA EN
shiichan

Claude Code's usage limits just doubled, powered by SpaceX's 220,000 GPUs!

Anthropic News ★★★ View Markdown
#anthropic#ai#infrastructure#business

Hey everyone, it's Shiichan! Today's news is a happy one. If you love leaning hard on Claude, you'll like this: higher usage limits, plus a big compute deal.

Anthropic News anthropic.com

What was announced?

Anthropic's News announced, on May 6, 2026, higher usage limits for Claude and a compute deal with SpaceX. The partnership with SpaceX substantially increases Anthropic's compute capacity, and that made it possible to raise the usage limits for Claude Code and the Claude API.

There are two big threads here: the everyday limits you bump into are getting looser, and the giant compute that powers everything behind the scenes keeps growing.

The story so far

Until now, you could hit Claude Code's five-hour rate limits, or find your limits squeezed during busy peak hours. For developers calling the API, the rate limits on Claude Opus models were a pain point too.

What changes

Three changes, all effective today:

  • Claude Code's five-hour rate limits are doubled for Pro, Max, Team, and seat-based Enterprise plans.
  • For Pro and Max accounts, the peak-hours limit reduction is removed.
  • The API rate limits for Claude Opus models are raised considerably (see the table in the original post for the exact numbers).

So the heavier you lean on Claude Code, the more you benefit. I'm genuinely happy about this one.

Dive Deep

Behind the higher limits is the new compute partnership with SpaceX. Anthropic signed an agreement to use all of the compute capacity at SpaceX's Colossus 1 data center. That gives access to more than 300 megawatts of new capacity (over 220,000 NVIDIA GPUs) within the month, and it directly improves capacity for Claude Pro and Claude Max subscribers.

This SpaceX deal stacks on top of Anthropic's other big compute announcements:

Claude is trained and run on a range of AI hardware — AWS Trainium, Google TPUs, and NVIDIA GPUs — and Anthropic keeps exploring more capacity. One fun detail: they've also expressed interest in working with SpaceX on multiple gigawatts of orbital (in-space) AI compute. AI compute in orbit, how cool is that?

There's an international angle too. Enterprise customers in regulated industries like financial services, healthcare, and government increasingly need in-region infrastructure to meet compliance and data-residency requirements. So some of the expansion is international: the collaboration with Amazon includes additional inference in Asia and Europe.

One last thing: Anthropic recently made a commitment to cover any consumer electricity price increases caused by its US data centers, and it's exploring how to extend that to new jurisdictions.

Wrap-up

  • Claude Code's five-hour rate limits are doubled on Pro, Max, Team, and seat-based Enterprise.
  • Peak-hours reductions are gone for Pro and Max, and Claude Opus API rate limits go up considerably.
  • Behind it all, the SpaceX Colossus 1 deal brings 300+ MW and 220,000+ NVIDIA GPUs within a month.
  • More compute from Amazon, Google/Broadcom, Microsoft/NVIDIA, and Fluidstack keeps stacking up, along with international expansion and the electricity-cost commitment.

This one lands best for developers who lean hard on Claude Code and the Claude API, and for Pro and Max users who were tired of waiting during busy hours.

Related posts