← Blog
explainx / blog

Anthropic secures SpaceX Colossus 1 supercomputer: rate limits doubled, peak-hour cuts removed

Anthropic announced a compute partnership with SpaceX for exclusive access to the Colossus 1 supercomputer in Memphis—300+ MW, 220,000 NVIDIA GPUs—immediately doubling Claude Code rate limits and raising API limits.

8 min readYash Thakker
AnthropicSpaceXColossus 1Compute partnershipClaude CodeRate limitsNVIDIA GPUs

MDX restores the committed source plus an HTML comment attribution; plain text bundles the rendered markdown body with the explainx.ai attribution footer.

Anthropic secures SpaceX Colossus 1 supercomputer: rate limits doubled, peak-hour cuts removed

Anthropic announced a compute partnership with SpaceX on May 6, 2026, securing exclusive access to the Colossus 1 supercomputer in Memphis, Tennessee—a facility with over 300 megawatts of power and 220,000 NVIDIA GPUs. The deal immediately doubles rate limits for Claude Code on Pro, Max, Team, and Enterprise plans, eliminates peak-hour restrictions, and substantially raises API rate limits for Opus models.

The partnership comes alongside Anthropic's other recent compute deals and represents a massive capacity increase that enables more autonomous, sustained AI workflows for developers. Both companies hinted at future orbital AI compute powered by SpaceX launches, though no timeline has been announced.

Primary sources


TL;DR

TopicDetails
PartnershipAnthropic gets exclusive access to Colossus 1 supercomputer via SpaceX
Compute specs300+ MW power, 220,000 NVIDIA GPUs in Memphis, Tennessee
Claude Code limitsDoubled 5-hour rate limits for Pro, Max, Team plans
Peak-hour policyPeak-hour cuts removed for Pro and Max plans
API limitsSubstantially raised for Opus models
Effective dateMay 6, 2026 (immediate)
FutureOrbital AI compute hinted; SpaceXAI integration mentioned by Musk

What changed for Claude users: rate limits doubled, peak-hour cuts removed

Anthropic's official announcement listed three immediate changes:

1. Doubled Claude Code 5-hour rate limits

Pro, Max, Team, and seat-based Enterprise plans now have 2x the previous 5-hour rate limits for Claude Code.

What this means:

  • Longer coding sessions before hitting limits
  • More complex refactors in a single session
  • Less downtime waiting for rate limit resets

Example: If you previously hit limits after 2.5 hours of sustained Claude Code use, you can now work for 5 hours before hitting the same wall.

2. Peak-hour limit reduction removed

Pro and Max plans previously had reduced rate limits during peak hours (roughly US business hours, 9 AM–5 PM Pacific). This restriction is now gone.

What this means:

  • No more slowdowns during US daytime
  • Consistent performance 24/7
  • Better experience for global teams (no penalty for working in US time zones)

Developer reaction:

"Usage limits are up, effective today we're: 1) Doubling Claude Code's 5-hour limits for Pro, Max, Team and seat-based Enterprise plans 2) Removing peak hours limit reduction on Claude Code for Pro and Max plans 3) Substantially raising our API rate limits for Opus models" — @ClaudeDevs

3. API rate limits substantially raised for Opus models

Claude API users (developers integrating Claude into their own apps) now have substantially higher rate limits for Opus 4.7 and Opus 4.6 models.

What this means:

  • Fewer 429 rate limit errors in production
  • More parallel requests for high-traffic apps
  • Faster response times during bursts

Pricing unchanged: Rate limit increases are capacity improvements, not price hikes. Token costs remain the same ($5/MTok input, $25/MTok output for Opus 4.7).


What is Colossus 1?

Colossus 1 is described as one of the world's largest and fastest-deployed AI supercomputers, located in Memphis, Tennessee. Key specs from public announcements:

  • Power: Over 300 megawatts (enough to power a small city)
  • GPUs: 220,000 NVIDIA GPUs (likely H100 or H200 generation, though not confirmed)
  • Deployment speed: Built and deployed faster than most comparable facilities (SpaceX/xAI operational efficiency)

xAI announcement:

"SpaceXAI will provide @AnthropicAI with access to Colossus 1, one of the world's largest and fastest-deployed AI supercomputers, to provide additional capacity for Claude → http://x.ai/news/anthropic-compute-partnership" — @xai

Note: xAI (Elon Musk's AI company) originally built and operated Colossus 1. The announcement suggests xAI infrastructure is now being made available to Anthropic via SpaceX.


Elon Musk's involvement: safety review and SpaceXAI comments

Elon Musk publicly discussed his direct involvement in evaluating Anthropic before approving the partnership:

"Same here. By way of background for those who care, 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 [AI safety]." — @elonmusk

Context: Musk has historically been critical of OpenAI (which he co-founded but later left) and cautious about AGI safety. His public endorsement of Anthropic's safety approach is significant.

SpaceXAI: xAI dissolving into SpaceX?

In a separate thread, Musk hinted at organizational changes:

"xAI will be dissolved as a separate company, so it will just be SpaceXAI, the AI products from SpaceX" — @elonmusk

Reading the tea leaves:

  • xAI's compute infrastructure (Colossus 1) is being integrated into SpaceX
  • Anthropic gets access to this infrastructure via SpaceX partnership, not xAI partnership
  • Future AI products from Musk's ecosystem may be branded SpaceXAI rather than xAI

Caution: Musk tweets are not official press releases. Wait for formal announcements from SpaceX and xAI before treating this as confirmed.


Orbital AI compute: hinted but not confirmed

Multiple sources mentioned future orbital AI compute powered by SpaceX launches, but no timeline or technical details have been released.

Speculation:

  • Starlink-style satellite clusters running AI inference in orbit
  • Low-latency global access to Claude via satellite downlinks
  • Edge AI for SpaceX missions (Starship, Mars colonization)

Reality check: Launching GPUs into orbit is expensive, thermally challenging, and limited by power constraints. Orbital AI is more likely a PR vision than a near-term product.

Focus on what shipped: The Colossus 1 partnership is real and live today. Orbital compute is aspirational.


Why this matters: capacity was a bottleneck

Before this deal, Claude Code users frequently complained about:

  • Hitting rate limits during long coding sessions
  • Slowdowns during peak hours (US business hours)
  • 429 errors on the API during traffic bursts

Anthropic's previous compute was sufficient for chat but not for sustained agent workflows—where users expect to run multi-hour sessions without interruption.

SpaceX Colossus 1 solves this bottleneck by 2-3x'ing Anthropic's available compute (based on the "other recent compute deals" phrasing in the announcement).

Developer reaction:

"We've agreed to a partnership with @SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we've been able to increase our usage limits for Claude Code and the Claude API." — @claudeai


How this compares to competitors

OpenAI (via Microsoft Azure) and Google (via Vertex AI) have massive compute from their parent companies' cloud infrastructure. Anthropic has historically been smaller and more compute-constrained.

This partnership changes the equation:

  • Anthropic now has dedicated supercomputer access (not shared cloud instances)
  • 220,000 GPUs puts Anthropic in the same league as OpenAI and Google for training and inference
  • SpaceX operational efficiency may give Anthropic faster deployment cycles than traditional cloud providers

Trade-off: Anthropic is now dependent on SpaceX infrastructure. If the partnership ends or Colossus 1 has downtime, Claude's capacity could drop.


Pricing: no changes (yet)

Rate limits doubled, but pricing did not change. This is a capacity upgrade, not a price hike.

Current Claude Code pricing:

  • Pro: $20/month
  • Max: $200/month
  • Team: Custom (per seat)
  • Enterprise: Custom (usage-based or seat-based)

Current API pricing (Opus 4.7):

  • Input: $5 per million tokens
  • Output: $25 per million tokens

If Anthropic's compute costs are now lower (due to SpaceX's deployment efficiency), we might see price cuts in the future—but don't bet on it. More likely: additional features at the same price.


Availability: effective immediately

All changes went live on May 6, 2026. If you're a Claude Code Pro, Max, Team, or Enterprise user, you already have the doubled limits and no peak-hour cuts.

To verify:

  1. Open Claude Code in your terminal
  2. Run a long coding session during US peak hours (9 AM–5 PM Pacific)
  3. Check if you hit rate limits later than before

API users: Check your rate limit headers in API responses. Opus limits should be substantially higher.


Related on ExplainX


Bottom line

Anthropic's SpaceX Colossus 1 partnership is a compute infrastructure deal that immediately improves the Claude Code and API experience: doubled rate limits, no peak-hour cuts, and substantially higher API capacity. The 220,000 NVIDIA GPUs and 300+ MW power in Memphis put Anthropic in the same compute league as OpenAI and Google.

For developers, this means longer coding sessions, fewer rate limit errors, and better performance during US business hours—all at no additional cost. Combined with the Dreaming and multiagent orchestration announcements (covered in our separate article), Anthropic is positioning Claude as the most capable and scalable agent platform for autonomous developer workflows.

Orbital AI compute is aspirational. Colossus 1 is real and live today. Enjoy the doubled limits.


This article is an independent summary for developers on explainx.ai and is not sponsored by Anthropic or SpaceX. Details are based on public announcements and community reports as of May 7, 2026; verify rate limits in your own usage before making business decisions.

Related posts