Anthropic said it dramatically outpaced its own growth forecasts in the first quarter, forcing the AI startup into an extraordinary infrastructure deal with Elon Musk’s xAI to keep its flagship model running at scale.
CEO Dario Amodei told developers in San Francisco that Anthropic had planned for 10-fold growth, but usage and revenue surged roughly 80-fold on an annualized basis in the quarter — a pace he called “just crazy” and “too hard to handle.” The company said annualized revenue has climbed to about $30 billion, roughly three times last year’s run rate, driven in part by corporate customers such as Uber and Netflix embedding Claude’s capabilities via the Claude API and adopting Claude Code for coding and agentic workflows.
That breakneck expansion strained Anthropic’s compute capacity and precipitated a string of performance problems. In a postmortem published in late April, the company acknowledged that three bugs had affected Claude Code since March 4, producing several weeks of degraded service. Anthropic said its internal tests did not catch the issues; when users first reported problems in early March the company said it could not reproduce them and did not immediately address the faults. On social platforms, some customers said they had cancelled subscriptions after experiencing regressions in recent months.
To address the capacity shortfall, Anthropic announced earlier this week that it will rent the entire compute capacity of Colossus 1, the Memphis data centre built by Musk’s xAI. The arrangement is designed to relieve the immediate shortage of GPU compute and allow Anthropic to raise usage limits for paying customers. The company is doubling Claude Code’s five-hour rate limits for its Pro, Max, Team and seat-based Enterprise plans and removing peak-hour limit reductions on Pro and Max accounts, changes Anthropic said will take effect across paid plans and the Claude API.
The move — leasing an entire rival-built data centre — is a sign of how acute infrastructure pressures have become as commercial demand for large language models grows. “Software engineers are the ones who are fastest to adopt new technology,” Amodei said at the conference, pitching the surge in developer uptake as evidence of how AI will reshape broader parts of the economy. Companies have turned to models like Claude to automate customer service, streamline internal workflows and power new developer tools; in some cases customers prefer Claude Code’s agent-like behavior over alternatives.
Anthropic’s postmortem and recent user complaints underscore the challenges of scaling complex models while maintaining reliability. The company acknowledged the bugs but did not provide a detailed timeline for full remediation or say whether additional capacity from Colossus 1 would immediately eliminate residual performance gaps. Some customers have continued to voice frustration publicly over what they described as a “significant regression” in recent months.
Trending Now
Uncertain AMOC weakening highlights need for intensified ocean monitoring and emissions cuts after 2020 acidification risk finding
D.C. Circuit panel questions Trump administration argument that projects underway are untouchable in White House East Wing ballroom dispute
Paige Bueckers’ 2025 WNBA debut jersey sells for $64,720, a record for women’s game-worn memorabilia
Dua Lipa and Callum Turner Hold Three-Day Sicily Wedding Celebration in Palermo After London Civil Ceremony
Industry observers say the episode highlights a broader trend: demand for GPU-backed compute is surging across cloud providers, model builders and enterprises, and capacity constraints can rapidly translate into product disruptions for fast-growing AI firms. For Anthropic, the rental agreement with xAI buys time to stabilize Claude’s performance and to expand paid usage — but it also signals that even well-funded startups must sometimes rely on unconventional partnerships to keep pace with explosive commercial demand.
