On Monday morning, Claude was offline. Worldwide. For nearly three hours, users couldn’t access claude.ai, the API, or Claude Code. About 2,000 users reported issues at the peak of the outage around 6:40 AM New York time on Downdetector.
The official reason? ‘Unprecedented demand.‘
The numbers behind it
And the numbers really are impressive. Anthropic confirms: free users have increased by more than 60 percent since January. Paid subscribers have more than doubled since October. Claude topped the Apple App Store for days.
What’s happening here isn’t normal growth. It’s a wave — triggered by the Pentagon crisis and the #CancelChatGPT movement. Users switching from OpenAI to Claude aren’t coming one by one. They’re coming in droves.
What went wrong technically
The outage didn’t affect the AI models themselves. The Claude API was apparently working fine. The problem was in the authentication and web infrastructure — exactly where user load hits most directly. Users who were already logged in got server errors. Those trying to start new sessions were blocked entirely.
Companies using Claude via the API were unaffected. But for the millions of end users on claude.ai and Claude Code, the service was simply unreachable.
Growing pains
A three-hour outage sounds like a problem. But honestly: when your product breaks because too many people want to use it, that’s the most enviable kind of problem a startup can have.
Anthropic is growing faster than its own infrastructure can keep up. That needs fixing — no question. But the cause is enviable: people trust Claude, and they’re coming in record numbers.
Sources: