Key Changes for Users

Effective Tuesday, Claude Code’s five-hour usage windows have been doubled for Pro, Max, Team, and seat-based Enterprise plans. Pro and Max subscribers, in particular, benefit from the removal of peak-hours throttling, allowing for smoother, uninterrupted access during high-demand periods. Additionally, rate limits for the Opus API have been increased, enabling developers to scale their applications more effectively.

“These updates directly improve service for Claude Pro and Claude Max subscribers,” Anthropic stated in its announcement. The changes address growing user frustrations over recent tightening of limits, which had been nerfed in prior months due to surging demand for compute-intensive tasks like agent-style coding workflows.

The SpaceX Colossus 1 Powerhouse

At the heart of these improvements is Anthropic’s new agreement to utilize all compute capacity from SpaceX’s Colossus 1 data center. This facility boasts over 300 megawatts of power and more than 220,000 Nvidia GPUs, set to come online within the month. Located in facilities in Tennessee and Mississippi, Colossus 1 represents a colossal leap in AI infrastructure, tailored to fuel the next generation of large language models.

SpaceX’s entry into AI compute partnerships underscores Elon Musk’s broader ambitions beyond rocketry and satellites. While details on pricing remain undisclosed, the deal positions Anthropic at the forefront of scalable AI deployment, potentially subsidizing inference costs that have strained even premium $20 Pro plans.

Bolstering a Robust Compute Portfolio

This SpaceX pact joins an impressive lineup of compute commitments for Anthropic. The company recently expanded its collaboration with Amazon to secure up to 5 gigawatts (GW) of capacity, including nearly 1 GW of new Trainium2 and Trainium3 capacity online by the end of 2026. Trainium2 clusters are already ramping up in the coming months, promising rapid scalability.

Further afield, Anthropic maintains a 5 GW deal with Google and Broadcom, slated for 2027 rollout; a strategic alliance with Microsoft and Nvidia encompassing $30 billion in Azure capacity; and a whopping $50 billion investment in American AI infrastructure alongside Fluidstack.

“Today’s agreement with Amazon will quickly expand our available capacity, delivering meaningful compute in the next three months,” Anthropic noted in an April update, highlighting the multi-pronged strategy to outpace AI demand.

Industry Context and User Impact

The timing is poignant amid industry-wide shifts. As noted on forums like Hacker News, “the era of subsidized inference is truly ending,” with providers like Anthropic tightening limits to manage costs from power users dominating compute resources. Claude Code, in particular, proves resource-hungry for multi-step tasks, prompting tiered access—Max plans, five times the cost of Pro, were briefly considered exclusive before reverting amid backlash.

Separate buzz surrounds SpaceX’s “try now, decide later” deal with coding tool Cursor, potentially valuing a full acquisition at $60 billion or a $10 billion partnership payout. While unrelated directly, it signals SpaceX’s deepening AI footprint, ripe for synergies like the Anthropic tie-up.

Looking Ahead

For developers, the doubled limits and SpaceX infusion mean fewer interruptions and greater reliability for Claude’s coding prowess. As AI models grow ever-larger, such hyperscale compute deals will define competitive edges. Anthropic’s aggressive infrastructure buildup positions it to challenge rivals like OpenAI and Google DeepMind, ensuring Claude remains a developer favorite.

Users are already reporting smoother experiences, with social media abuzz over the changes. Whether this heralds sustained expansions or merely a temporary reprieve from compute crunches remains to be seen—but for now, Anthropic and SpaceX have ignited a new chapter in AI accessibility.