Darhost

2026-05-07 04:37:41

Anthropic's Surprising SpaceX Deal Doubles Claude Code Limits

Anthropic doubles Claude Code limits via SpaceX deal, using 220K GPUs from Colossus 1. Elon Musk reverses criticism. Lifts peak-hour limits. Long-term solution uncertain.

In a move that has surprised many, Anthropic announced a significant expansion of Claude Code usage limits for its subscribers, starting today. This change came about through an unexpected partnership with SpaceX, which will provide Anthropic with massive computing capacity from a data center in Memphis, Tennessee. Here are the key details in a Q&A format.

What exactly did Anthropic announce regarding Claude Code limits?

Anthropic has doubled the five-hour usage limits for Claude Code across all its subscription plans, including Claude Pro, Max, Team, and seat-based Enterprise tiers. This decision came just as the company was considering dropping Claude Code access from its $20-per-month Pro plan due to capacity constraints. Instead, the deal with SpaceX allowed Anthropic to increase limits without downgrading any plan. Additionally, the company raised pay-per-token API rate limits for its powerful Opus models. No changes were made to weekly limits, and Anthropic has been contacted for further details.

Anthropic's Surprising SpaceX Deal Doubles Claude Code Limits
Source: www.pcworld.com

How did SpaceX help Anthropic increase computing capacity?

SpaceX is providing Anthropic with access to the full computing capacity of its Colossus 1 data center in Memphis, Tennessee. This facility houses over 220,000 Nvidia GPUs, which will be used to handle the heavy computational demands of Claude Code and other AI tools. The partnership significantly boosts Anthropic's available compute power, allowing the company to meet growing subscriber usage. Anthropic also has existing deals with Amazon, Google, and Microsoft for computing capacity, but this SpaceX arrangement brings a substantial new resource online.

Why is the Anthropic–SpaceX partnership considered surprising?

The partnership is unexpected because Elon Musk, who owns a 42% stake in SpaceX, has been a vocal critic of Anthropic in the past. He previously called the company “misanthropic and evil.” However, in a social media post on Wednesday, Musk reversed his stance, saying he was “impressed” after recent meetings with Anthropic staff. He noted that “no one set off my evil detector.” This about-face has raised eyebrows, given Musk's history of public feuds with AI companies. The deal underscores how business needs can overcome personal disagreements, especially when both parties stand to gain from increased AI computing power.

What other changes did Anthropic make to its subscription plans?

Beyond doubling Claude Code limits, Anthropic also lifted peak-hour usage limits for Claude Pro and Max accounts. These limits had been imposed in late March to manage network strain during high-demand periods. Now, users on these plans can access Claude during peak times without throttling. This change, combined with the increased code limits, gives subscribers more flexibility. The company also increased API rate limits for the Opus model, which is used by developers for more intensive tasks. These adjustments aim to balance user expectations with the company's actual computing capacity.

Anthropic's Surprising SpaceX Deal Doubles Claude Code Limits
Source: www.pcworld.com

What capacity challenges was Anthropic facing before the SpaceX deal?

Anthropic had been struggling to meet demand for its AI services, particularly for token-hungry tools like Claude Code and Cowork. The company's consumer plans, such as the $20-per-month Claude Pro, were not designed for such heavy usage. An Anthropic executive admitted that the plans were not built for agentic tools that consume large amounts of computational tokens. As a result, the company considered dropping Claude Code from the Pro plan entirely. The deal with SpaceX provides much-needed breathing room, bridging the gap between what subscribers expect and what infrastructure can currently deliver.

Is the SpaceX deal a long-term solution for Anthropic's capacity issues?

It remains unclear whether this partnership will be a lasting fix. While the addition of Colossus 1's 220,000 GPUs is a massive boost, Anthropic's capacity woes may resurface as usage grows. The company has not specified the duration or terms of the SpaceX agreement. Moreover, the ever-shifting usage limits that Pro and Max users have experienced could continue if demand outpaces supply. Anthropic is also exploring orbital AI compute capacity with SpaceX, which could open new frontiers. For now, the deal offers temporary relief, but long-term stability will depend on further infrastructure investments and possibly more partnerships.

What is the Colossus 1 data center and why is it important?

The Colossus 1 data center is a SpaceX facility located in Memphis, Tennessee. It houses over 220,000 Nvidia GPUs, making it one of the largest concentrated computing clusters in the world. For Anthropic, this means access to immense parallel processing power, which is essential for training and running large language models like Claude. The GPUs enable faster inference and code generation for Claude Code users. This compute capacity is critical as Anthropic rolls out more agentic AI tools that require significant tokens. The partnership also highlights SpaceX's growing role as a cloud computing provider, beyond its traditional space ventures.