OpenAI, Anthropic, and xAI All Went Down Simultaneously on September 3 — No Shared Cause Confirmed
Three of the five largest frontier AI providers went offline within minutes of each other on the morning of September 3, 2026. None of them confirmed a shared cause.
Timeline (All Times PT, September 3)
6:23 AM — Anthropic begins alerting on “elevated errors” for Claude Mythos 5.1, Claude Fable 5.1, and Claude Opus 5. “Identified the cause. A fix has been deployed.” Resolved by 9:16 AM. Claude Sonnet 5 briefly showed similar issues just after 9 AM.
6:30 AM — xAI posts “investigating outage” to status.x.ai. “Grok is experiencing issues across all platforms and services.” Resolved 10:05 AM.
7:43 AM — OpenAI’s routing error hits. ChatGPT and Codex unavailable for some users. “A solution was successfully implemented” by 8:17 AM.
Total overlap window: roughly 6:23 AM to 10:05 AM, with all three incidents active simultaneously between 7:43 and 8:17 AM.
What Each Lab Said
xAI gave the most specific statement: “An outage at our Memphis compute center this morning.” The company also added: “We’d also like to apologize to our impacted compute partners” — notably plural.
OpenAI described “a routing error” as the cause. No external dependency mentioned.
Anthropic declined to comment publicly and provided no cause in response to media requests. Its status updates said only that a cause was identified and a fix deployed.
The Compute Partnership Angle
Anthropic and xAI announced a compute partnership with SpaceX in May 2026. xAI’s Memphis facility is part of SpaceX’s Colossus supercomputing cluster. xAI’s apology to “compute partners” is consistent with shared infrastructure at that site — though neither Anthropic nor xAI confirmed any operational link to Thursday’s incidents.
Major infrastructure providers — AWS, Microsoft Azure, Cloudflare — reported no outages on September 3. The absence of a reported shared third-party failure makes the coincidence harder to explain by the usual suspects: CDN failure, cloud provider event, DNS outage.
What Was Not Said
There were unconfirmed reports of a brief Google Gemini disruption on the same morning. Google did not confirm any incident and its status dashboard showed no entries.
The Anthropic-xAI compute partnership creates a plausible path to a shared dependency, but neither company has confirmed it as a factor. Until one of them does, the simultaneous outage remains officially unexplained: three separate providers, three separate stated causes, no confirmed common root.