On May 6, 2026 (US time), Anthropic announced both higher Claude Code usage limits and a major compute partnership with SpaceX. The five-hour rate caps for Claude Code Pro, Max, Team, and Enterprise plans are doubled effective today, and the peak-hours throttling for Pro and Max is being removed outright. For heavy users, this is the kind of news that has been a long time coming[1].

Claude Code 5-hour limits doubled, Opus API rate caps raised

The new usage limits are effective today and come in three pieces. First, the five-hour rate cap on Claude Code is being doubled, applying to Pro, Max, Team, and seat-based Enterprise plans alike[1]. Second, the peak-hours limit reduction on Claude Code is being eliminated for Pro and Max accounts, which directly addresses the complaint that day-time coding sessions were burning through quota faster than expected[1].

The third change is on the API side: API rate limits for Claude Opus models are being raised considerably, with the exact updated numbers laid out in a table on Anthropic's announcement page[1]. For teams running Opus in production, those caps map straight into throughput, so it is worth checking the latest values in the official rate-limits documentation[1].

The backstory makes sense too. Since March, multiple outlets have reported that Claude Code quotas were running out faster than users expected and that Anthropic was tweaking its limits in line with available compute[5][6]. Today's expansion looks like the answer to those reports — and a clear sign that the supply side has finally caught up.

SpaceX's Colossus 1 fully booked, 220,000+ GPUs online within a month

The headline news is the new compute partnership with SpaceX. Anthropic says it has signed an agreement to use all of the compute capacity at SpaceX's Colossus 1 data center, giving it access to more than 300 megawatts of new capacity and over 220,000 NVIDIA GPUs within the month[1][2].

Colossus 1 is located in Memphis, Tennessee, and is known for densely packed deployments of H100, H200, and next-generation GB200 accelerators[2]. Folding that entire site into Anthropic's inventory is no small thing, and Anthropic itself notes that the new capacity will directly improve capacity for Claude Pro and Claude Max subscribers[1]. Whether users feel a noticeable bump in responsiveness over the coming weeks will be the real test.

On top of that, Anthropic also says it has expressed interest in working with SpaceX on multi-gigawatt "orbital AI compute capacity" as part of the same agreement[1][3]. With the competition for terrestrial data-center sites and power only intensifying, looking up to space is the kind of move that is genuinely fun to watch. CNBC and Axios both lead with this space-development element of the deal[2][3].

5 GW × 2 + $30B + $50B: a compute build-out that just keeps growing

The SpaceX deal is not a one-off — it sits in line with several other major compute agreements from the past few months. In the same blog post, Anthropic lists four of them: an agreement with Amazon for up to 5 GW (with nearly 1 GW of new capacity coming online by the end of 2026), a 5 GW agreement with Google and Broadcom (set to begin coming online in 2027), a strategic partnership with Microsoft and NVIDIA that includes $30 billion of Azure capacity (about 4.68 trillion yen)※at 1 USD = 156 JPY, and a $50 billion (about 7.8 trillion yen) investment in American AI infrastructure with Fluidstack※at 1 USD = 156 JPY[1].

The hardware mix is also worth noting: Anthropic explicitly says it is training and running Claude across AWS Trainium, Google TPUs, and NVIDIA GPUs[1]. By avoiding lock-in to a single vendor, the company can stack new capacity wherever it appears — and SpaceX's Colossus 1 fits that strategy cleanly.

The post also touches on international expansion. Customers in regulated industries — finance, healthcare, government — increasingly need in-region infrastructure for compliance and data-residency reasons, so some of the capacity expansion will land in Asia and Europe[1]. Anthropic also says it is exploring how to extend its earlier commitment to cover any consumer electricity-price increases caused by its US data centers to new jurisdictions, alongside investing back into communities that host its facilities[1]. Continuing that "if we draw on local power, we give back to local residents" stance across borders is a stance fans can comfortably get behind.

Summary

Anthropic has locked in SpaceX's Colossus 1 entirely, brings 300+ MW and 220,000+ GPUs online within a month, and immediately turns that supply into doubled Claude Code limits and higher Opus API caps for users. Coming on top of the deals with Amazon, Google, Microsoft, and Fluidstack, it is a clear signal of how serious the company is about scaling up its compute base.

出典:[1] Anthropic「Higher usage limits for Claude and a compute deal with SpaceX」 https://www.anthropic.com/news/higher-limits-spacex

出典:[2] CNBC「Anthropic, SpaceX announce compute deal that includes space development」 https://www.cnbc.com/2026/05/06/anthropic-spacex-data-center-capacity.html

出典:[3] Axios「Anthropic will get compute capacity from Elon Musk's SpaceX」 https://www.axios.com/2026/05/06/anthropic-spacex-elon-musk-compute

出典:[4] Bloomberg「Anthropic, SpaceX Sign Deal to Boost AI Computing Power for Claude Software」 https://www.bloomberg.com/news/articles/2026-05-06/anthropic-inks-computing-deal-with-spacex-to-meet-ai-demand

出典:[5] The Register「Anthropic admits Claude Code quotas running out too fast」 https://www.theregister.com/2026/03/31/anthropic_claude_code_limits/

出典:[6] The Register「Anthropic tweaks Claude usage limits to manage capacity」 https://www.theregister.com/2026/03/26/anthropic_tweaks_usage_limits/