Glonce

OpenAI, Anthropic and xAI go down together, no shared cause

infrastructureThe Glonce editorsWired🔥61Glonce rating61Impact45Novelty35Relevance95Surprise55Players90Credibility88
OpenAI, Anthropic and xAI go down together, no shared cause

Frontier AI services from Anthropic, OpenAI and xAI all ran into trouble on Thursday, September 3, an occurrence WIRED described as rare precisely because it hit three leading labs within the same morning. The trouble reached OpenAI's ChatGPT and Codex, several Claude models, and xAI's Grok, disrupting each chatbot for some users. Each company gave a different account of its own incident, and none of them said the outages were connected to each other or to an outside cause.

Anthropic's own status page was first to show trouble, opening a partial outage alert at 6:23 am PT for elevated errors on requests to Claude Mythos 5.1, Claude Fable 5.1 and Claude Opus 5. Anthropic said shortly afterward that it had identified the cause and deployed a fix, and it marked the issue resolved by 9:16 am PT, but the company never said what the cause actually was, and it declined to comment to WIRED on the episode. Claude Sonnet 5 also appeared to have brief, similar trouble shortly after 9 am PT, though that model was not named in Anthropic's own alert.

xAI posted "investigating outage" on its own status page for Grok at 6:30 am PT, saying the chatbot was experiencing issues across all of its platforms and services. The company marked the episode complete at 10:05 am PT, writing that it had "resolved the situation" and that "traffic is healthy again." SpaceX, xAI's parent company, did not respond to WIRED's request for comment, but separately said in its own public comments on Thursday that Grok's problems came from "an outage at our Memphis compute center this morning," adding: "we'd also like to apologize to our impacted compute partners." WIRED noted, as context rather than as an explanation, that Anthropic and xAI had announced a compute partnership with SpaceX back in May.

OpenAI spokesperson Kathleen Chaykowski told WIRED that a routing error starting around 7:43 am PT on Thursday, September 3, made ChatGPT and Codex unavailable for some users across platforms, and that as of about 8:17 am PT a solution had been successfully implemented and was continuing to be monitored. OpenAI did not say what caused the routing error or whose routing was affected.

The overlapping timing initially made the three incidents look connected, WIRED wrote, perhaps to a shared third-party service provider, but neither OpenAI nor Anthropic cited any external cause in their comments. Simultaneous outages across one sector would typically point to a shared cloud provider, content delivery network or other vendor running into trouble, but none of Cloudflare, Amazon Web Services or Microsoft Azure reported outages that Thursday. There were also scattered reports of a possible Google Gemini outage the same morning, but Google neither confirmed them nor logged any incident on its own service status dashboard, and it did not respond to WIRED's request for comment ahead of publication.

Key facts

  • Anthropic, OpenAI and xAI all had outages on the same Thursday morning (September 3), taking down ChatGPT, Codex, several Claude models and Grok for some users.
  • OpenAI said a routing error starting around 7:43 am PT hit ChatGPT and Codex; a fix was in place by about 8:17 am PT and monitoring continued.
  • Anthropic's status page reported a partial outage from 6:23 am PT affecting Claude Mythos 5.1, Claude Fable 5.1 and Claude Opus 5, marked resolved by 9:16 am PT; the page said the company had identified the cause and deployed a fix but never said what it was; Anthropic itself declined to comment to WIRED.
  • xAI's Grok outage ran on the company's status page from 6:30 am PT to 10:05 am PT; parent company SpaceX blamed "an outage at our Memphis compute center."
  • Neither OpenAI nor Anthropic cited a shared external cause, and Cloudflare, Amazon Web Services and Microsoft Azure all reported no outages that day; a rumored Google Gemini outage went unconfirmed by Google.

Why it matters

Three of the industry's most-used AI services, OpenAI's ChatGPT and Codex, several Claude models, and xAI's Grok, all ran into trouble that same morning, at a moment when many people and companies now route everyday and professional work through exactly these tools. Each provider describes its own incident as separate and contained, but three leading labs going down in the same morning, with only brief status-page notes to show for it, is itself notable: anyone depending on a single AI vendor for time-sensitive work has no way to tell, in the moment, whether a problem is an isolated hiccup or something wider.

Who it affects

OpenAI said its routing error made ChatGPT and Codex unavailable for some users across platforms. Anthropic's own alert named Claude Mythos 5.1, Claude Fable 5.1 and Claude Opus 5 as hit by elevated errors, and Claude Sonnet 5 appeared to have brief, similar trouble shortly after, though it wasn't part of that alert. xAI said Grok was affected across all of its platforms and services. And, per scattered but unconfirmed reports, some Google Gemini users may have hit trouble too, though Google never confirmed this or logged an incident.

How to use it

None of the three companies has published anything beyond a brief statement or a status-page note, so anyone who depends on these services for critical work currently has only the timestamps and one-line causes given here: a routing error for OpenAI, a Memphis compute center problem for xAI's Grok, and an undisclosed identified cause for Anthropic. Until a fuller post-incident report appears, the safest approach is to treat each company's own status page as the authoritative, real-time record, and to not assume that a future outage at one of these providers is connected to trouble at another without the companies themselves confirming a shared cause.

How solid is it

The sequence of events is well sourced: OpenAI's account comes from an on-the-record statement by named spokesperson Kathleen Chaykowski, and Anthropic's and xAI's timelines come from their own public status pages, quoted directly. Where the story is deliberately not solid is on causation. WIRED frames the shared timing as something that only appeared linked, flags the shared-provider idea as a perhaps, and states plainly that neither OpenAI nor Anthropic pointed to an external cause. Anthropic declined to comment to WIRED on the episode and never said what its identified cause actually was, SpaceX did not respond to WIRED's request for comment at all, and Google did not respond before publication. The Gemini portion of the story is explicitly unconfirmed, resting on scattered reports rather than any company statement.

Risks and caveats

Read the shared-cause angle as a question WIRED raised and could not answer, not as a finding: the source is explicit that neither OpenAI nor Anthropic cited an external source, and that Cloudflare, Amazon Web Services and Microsoft Azure, the infrastructure vendors most likely to explain a coincidence like this, reported no outages that day. The Anthropic and xAI compute partnership with SpaceX, announced in May, is mentioned only as context next to SpaceX's apology to its impacted compute partners, and the source does not claim that partnership caused, or was affected by, Thursday's incidents. The Claude Sonnet 5 issue is hedged as something that seemed to happen, not a confirmed part of Anthropic's alert. And no company has disclosed how many users were affected, what the financial or SLA impact was, or, in Anthropic's and xAI's cases, what the underlying cause actually was.

“A routing error starting around 7:43 am PT on Thursday, September 3, made ChatGPT and Codex unavailable for some users across platforms. As of about 8:17 am PT on Thursday, a solution was successfully implemented and is continuing to be monitored.”

— Kathleen Chaykowski, OpenAI spokesperson, to WIRED