ChatGPT, Claude, and Grok were all degraded or unavailable within the same few-hour window on the morning of September 3 -- a rare simultaneous stumble by three of the industry's largest AI products. Each company opened its own, separate incident report, and each gave a different explanation -- and none of the three has said its outage was connected to either of the others'.
- Sept. 3, 6:26 a.m. PT — Anthropic's status page reports investigating elevated errors across Claude Mythos 5.1, Fable 5.1, and Opus 5
- Sept. 3, ~6:30 a.m. PT — Grok goes down; xAI later says a failure at its Memphis compute center was the cause
- Sept. 3, 7:43 a.m. PT — OpenAI logs a routing error, taking down ChatGPT and Codex for some users
- Sept. 3, 9:16 a.m. PT — Anthropic marks Claude's roughly 3-hour outage resolved
- Sept. 3, later that day — Grok and ChatGPT are both restored; none of the three companies states a shared cause
xAI said the outage traced to a failure at its Memphis compute center -- the same facility that trains Grok -- and apologized not just to users but to unnamed "impacted compute partners," without naming them. Anthropic's status page never named a root cause beyond "an infrastructure issue," though it confirmed elevated "529 overloaded" errors across five separate Claude model versions before resolving them. OpenAI attributed its own incident to a routing error that affected ChatGPT, Codex, logins, file uploads, voice mode, and several other components at once.
Cloudflare, AWS, and Google Cloud all reported no relevant incidents on their own status pages that morning, and Microsoft Azure -- which hosts OpenAI's infrastructure -- showed nothing that lines up either. (The likelier read, per outlets that checked each shared-infrastructure provider's status page independently, is three unrelated failures landing in the same few hours -- not one bigger event wearing three names.) xAI's own apology to unnamed "compute partners" is the one loose thread still hanging: it implies at least one other company depends on xAI's Memphis facility, but xAI hasn't said who, and nothing ties that dependency to OpenAI's or Anthropic's outage specifically.
- ChatGPT, Claude, and Grok each had separate outages on September 3, overlapping within hours.
- xAI said a failure at its Memphis compute center caused Grok's roughly 3.5-hour outage.
- Anthropic cited an unspecified infrastructure issue for a 3-hour Claude outage.
- OpenAI logged a routing error that took down ChatGPT and Codex the same morning.
- Caveat: major cloud providers reported no incidents, and no shared cause has been confirmed.