Simultaneous ChatGPT, Claude and Grok outages

ChatGPT, Claude, and Grok all went down within a 90-minute window on September 3, 2026. One source traces the disruptions to a Microsoft Azure East US regional infrastructure failure, noting that Google's Gemini, hosted on Google Cloud rather than Azure, remained largely operational at about 500 peak Downdetector reports while ChatGPT drew over 37,000 reports, Claude over 1,300, and Grok roughly 1,365. A separate analysis frames the Azure link as a hypothesis rather than a confirmed cause and characterizes the event as a confirmed overlap of separate provider-side incidents. Claude and Grok went down within four minutes of each other, with ChatGPT and Codex following 73 minutes later. Anthropic's partial outage affected Claude Mythos 5.1, Claude Fable 5.1, and Claude Opus 5, starting at 9:23 am Eastern and resolving by 12:16 pm, with a separate elevated-error incident on Claude Sonnet 5 briefly after noon. OpenAI's degraded performance across ChatGPT and Codex ran from 10:43 am to 12:55 pm, with OpenAI attributing it to a routing error. SpaceX AI said its Memphis compute center going offline caused Grok's outage. Anthropic resolved its incident but has not disclosed the root cause.