Anthropic doubles Claude Code 5-hour limits after SpaceX capacity deal

• xAI's partnership post said SpaceXAI will give Anthropic access to Colossus 1 to add Claude capacity, and ClaudeDevs' product update landed minutes later with doubled Claude Code 5-hour limits.
• Effective immediately, ClaudeDevs said Anthropic is doubling Claude Code's 5-hour caps for Pro, Max, Team, and seat-based Enterprise, removing peak-hour limit reductions for Pro and Max, and substantially raising Opus API rate limits.
• om_patel5's complaint roundup and AIandDesign's repost of a limit screen captured the backdrop: users had spent the previous day complaining about fast-burning sessions and hard Claude Code caps.
• The weird extra clause came from xAI's thread follow-up, which said both companies are also interested in developing multiple gigawatts of orbital AI compute capacity.

xAI's announcement put Colossus 1 at the center of the deal, while ClaudeDevs' post turned that capacity story into an immediate product change. bcherny's follow-up confirmed the limit reset was live the same day, petergyang's Dario quote tied the scramble to 80x growth in usage and revenue earlier this year, and xAI's orbital-compute thread slipped in the kind of sentence that sounds half press release, half sci-fi roadmap.

xAI said SpaceXAI will provide Anthropic with access to Colossus 1, which it described as one of the world's largest and fastest-deployed AI supercomputers, to provide additional capacity for Claude. The phrasing matters because the announcement framed this as extra capacity for an already hot product line, not a future research pact.

That made the competitive optics instantly strange. LLMJunky's reaction post summed up the mood in one line: the cluster many people assumed was reserved for Grok was suddenly being rented to Anthropic instead.

Anthropic tied the capacity increase to three concrete changes:

• Claude Code 5-hour limits are doubled for Pro, Max, Team, and seat-based Enterprise, per ClaudeDevs' post.
• Peak-hours limit reduction is gone for Pro and Max, according to the same update and trq212's follow-up.
• Opus API rate limits are being raised substantially, per ClaudeDevs.

Boris Cherny, Claude Code lead at Anthropic, said the changes were effective that day in bcherny's thread. That is the useful part for heavy Claude Code users: this was not a vague promise that more GPUs might eventually reduce throttling.

Before the reset, the loudest complaints were about limits burning faster than expected. om_patel5's screenshot-backed roundup cited Reddit posts from Max 20x subscribers who said sessions started draining far more quickly after late-April changes, while Everlier's reply joked that the new announcement mostly meant users could reach their cap twice as fast.

A short clip that AIandDesign reposted from RoundtableSpace showed the blunt failure mode: "YOUR CLAUDE CODE SESSION LIMIT HAS BEEN REACHED."

Anthropic's update does not answer every complaint in that thread. It does answer the most immediate one, which was plain shortage.

The most direct explanation for the deal came from Dario Amodei's public remarks rather than the product posts. In petergyang's quote card, Dario said Anthropic saw 80x growth earlier this year in usage and revenue, then added, "That's why we are going to keep acquiring as much compute as we can."

A second clip from trq212's event post carried the same line in plainer language: Anthropic is trying every day to obtain more compute and pass it on to users. That makes the SpaceX partnership look less like a one-off stunt and more like emergency infrastructure procurement for Claude's coding surge.

The strangest sentence in the entire news cycle was not about rate limits. In the thread under the announcement, xAI said SpaceXAI and Anthropic had also "expressed interest in partnering to develop multiple gigawatts of orbital AI compute capacity."

That line sits well beyond the concrete changes Anthropic shipped on Tuesday. trq212's "Claude in Space" post treated it as a joke, but xAI put it in the official thread anyway, which means the partnership announcement contained both a very practical near-term outcome, more Claude capacity now, and a very speculative long-term one, orbital compute later.