Claude Suffers Two Major Outages in Two Days, Anthropic Cites “Unprecedented Demand”

According to 1M AI News, on March 3rd, Anthropic's AI assistant Claude experienced a major service interruption for two consecutive days on March 2nd and 3rd. On March 2nd at 6:49 AM Eastern Time, Anthropic announced that they were investigating an issue of "elevated error rates" affecting claude.ai, the mobile app, Claude Opus 4.6, Claude Console, and Claude Code. Downdetector data showed that at its peak, over 2000 users reported the outage. Upon service restoration, Anthropic stated that the team was "working hard to match Claude's tremendous recent demand."

On March 3rd at UTC 03:15, the official status page again reported elevated error rates for claude.ai, cowork, platform, and Claude Code, with the investigation still ongoing as of UTC 04:43. During the two outages, the Claude API (api.anthropic.com) and Claude for Government remained unaffected, with the consumer-facing services experiencing the most impact.

Anthropic attributed the outage to "unprecedented demand," disclosing that since January this year, free user growth has exceeded 60%, and paid subscription users have more than doubled since October last year.

Bloomberg reported that the surge in demand was precipitated by OpenAI's AI partnership with the US Department of Defense, leading to user backlash—ChatGPT saw a 295% daily uninstallation spike in the US, 775% surge in one-star ratings, while Claude's US downloads increased by 37%-88% during the same period, exceeding ChatGPT's total daily downloads at one point. Users can check real-time status updates for each service component at status.claude.com.