Anthropic Taps SpaceX Colossus 1 to Double Claude Code Limits

Anthropic just solved its rate limit problem with a SpaceX deal. On May 6, 2026, the company announced it locked up all compute capacity at SpaceX’s Colossus 1 data center in Memphis, Tennessee, and started doubling Claude Code limits the same day.

For anyone who’s been blowing through their five-hour Claude Code window in 90 minutes, this is the announcement you’ve been waiting on.

The Colossus 1 facility hands Anthropic more than 300 megawatts of additional compute and access to over 220,000 NVIDIA GPUs. The capacity goes live within one month of the announcement. That’s the part that matters. Most of Anthropic’s other big infrastructure deals don’t come online until 2027 or later.

What changed for Claude users today

Three things shifted immediately. Claude Code’s five-hour rate limits doubled for Pro, Max, Team, and seat-based Enterprise plans. The peak hours limit reduction got removed entirely on Claude Code for Pro and Max accounts, so you’re not getting throttled during the workday anymore. And Claude Opus API Tier 1 limits jumped hard, maximum input tokens per minute went up 1,500% and maximum output tokens per minute went up 900%.

Those API numbers aren’t a typo. A 1,500% increase on input tokens means workflows that were rate-limit-bound an hour ago are basically unblocked.

The catch. These bumps apply to current paid tiers, not free users, and the Opus API jump specifically targets Tier 1. Higher tiers were already further up the curve, so the relative gain is biggest at the entry level where developers were hitting walls fastest.

Why Anthropic needed this now

Developer frustration with Claude Code limits has been building for months. Pro and Max subscribers reported torching their five-hour rolling windows in under two hours during heavy agentic coding sessions. In March 2026, Anthropic ran a temporary off-peak doubling promotion as a patch after a bug surfaced in limit enforcement. That promotion was a tell. The bottleneck wasn’t policy. It was raw compute supply.

Every major lab is locking in GPU capacity ahead of the agentic AI demand wave. Anthropic already has a 5 GW Amazon deal with nearly 1 GW landing by end of 2026, a 5 GW Google and Broadcom partnership starting in 2027, a $30 billion Azure capacity commitment with Microsoft and NVIDIA, and a $50 billion Fluidstack infrastructure investment in American AI infrastructure. Big numbers. Mostly future numbers.

Colossus 1 is different because it’s already built.

The Musk angle

This is also a strange pairing on paper. SpaceX and xAI are tightly linked, and Anthropic is one of xAI’s direct competitors. But Elon Musk posted on X that he was “impressed” by Anthropic’s team after meeting them the week before the announcement. Compute is compute. When you’ve got 220,000 GPUs sitting in Memphis and a buyer ready to take all of it, business gets done.

Anthropic also said it’s interested in partnering with SpaceX on something wilder, multiple gigawatts of orbital AI compute capacity. That’s satellites running inference in low Earth orbit, basically. Years out, if it happens at all, but the fact that it’s on the table tells you where Anthropic thinks the ceiling is on terrestrial power and cooling.

What you can do with it right now

If you’re already paying for Claude Pro, Max, Team, or Enterprise, you don’t need to do anything. The doubled Claude Code limits and removed peak hours throttling are live in your account. API users on Opus Tier 1 should check their dashboards because the new ceilings let you push much heavier batched workloads without hitting 429s.

The full official announcement is on Anthropic’s site at anthropic.com.

Next question worth watching. Whether the new capacity holds up once the entire Claude Code user base actually starts using it.

Share

Frequently Asked Questions

Anthropic needed usable capacity right now, not in 2027. The SpaceX Colossus 1 deal delivers more than 300 megawatts and 220,000 plus NVIDIA GPUs that go live within a month, which lets Anthropic immediately raise rate limits that developers have been hitting for months.

Pro, Max, Team, and seat-based Enterprise users get their five-hour Claude Code rate limits doubled. Pro and Max accounts also lose the peak hours limit reduction, so they get the full allowance any time of day. Claude Opus API Tier 1 customers get a 1,500% bump on input tokens per minute and a 900% bump on output.

Anthropic announced the deal on May 6, 2026, and said the Colossus 1 capacity will come online within one month of the announcement. The Claude Code limit increases and Opus API changes took effect immediately.

Harish Prajapat (Author)

Hi, I’m Harish! I write about AI content, digital trends, and the latest innovations in technology.

Related news

Get the latest news, tips & tricks, and industry insights on the MagicShot.ai news.