On March 2, 2026, Anthropic's Claude AI service suddenly went down, triggering widespread dissatisfaction among global users. According to Anthropic's official status page, the interruption began around 5 AM Pacific Standard Time and lasted over two hours, primarily affecting the Claude.ai web version, Claude Code coding tool, and login/logout functions.
Thousands of users reported inability to access the service, with error rates spiking dramatically. Downdetector data showed peak reports exceeding 2,000 issues from US users.
Anthropic attributed the outage in a statement to
"unprecedented demand surge". Since January, Claude's free users have grown over 60%, paid users have doubled, and recent days saw traffic surges that even caused ChatGPT uninstall rates to spike 295%, with many users switching to Claude.
The service topped Apple App Store's free app rankings last weekend, surpassing ChatGPT, further amplifying load pressure.
Notably, enterprise-level Claude API remained largely unaffected, with only consumer services experiencing interruptions.
On X.com (formerly Twitter), user reactions were intense. Many shared screenshots showing login failures or HTTP errors, with one user stating
"Claude has been down for 50 minutes, everyone is wailing", reflecting how Claude has become central to many people's workflows in a short time.
@MetaBaz7
Another developer pointed out this wasn't an AI model failure, but rather
"login and UI collapsing under load—infrastructure matters more than hype".
@AKirtesh
Some users linked the outage to global events, but officials confirmed it was purely a technical issue that has been fixed.
@EchoWireDai
As of March 3, service has returned to normal, though sporadic errors persist.
This incident isn't isolated. Claude also experienced a brief interruption on February 28, affecting the Opus 4.6 model.
Experts analyze this exposes pain points for AI startups amid rapid expansion: despite Anthropic's partnerships with agencies like the US Department of Defense, explosive demand growth tests its infrastructure.
In the future, such
"success-induced outages"may become commonplace, calling for AI companies to strengthen resilient design to ensure reliability.
© 2026 Winzheng.com 赢政天下 | 转载请注明来源并附原文链接