Three coiled ethernet cables in one unbranded network switch on a lamp-lit walnut desk, with a brass screwdriver beside it.
NewsTech

Three rival chatbots failed Thursday while Microsoft Azure’s status page stayed quiet

Aaron Harme

ChatGPT, Grok, and Claude stopped answering around the same time Thursday. ChatGPT errors started around 11am ET after a routing error OpenAI timed at 7:43 a.m. PT, Grok overloaded from about 9:30am ET, and Anthropic restored Claude around 12:15 to 12:16pm ET.

A routing error is traffic sent to the wrong place. OpenAI told Newsweek that error made ChatGPT and Codex unavailable for some users across platforms, and that mitigation was in place by about 8:17 a.m. PT.

Emma Roth at The Verge writes the outage also hit logins, file uploads, voice, search, deep research, and image generation.

Anthropic described a partial infrastructure issue across Claude.ai, Claude Code, Claude Cowork, and the Claude API. Thomas Claburn at The Register and Greta Cross at USA Today put the Claude restore at 12:16pm ET.

Kyle Orland at Ars Technica times the Anthropic partial outage from 9:23am ET and OpenAI’s elevated errors from 10:43am ET, matching the 7:43 a.m. PT routing-error clock.

The brands compete. The computers often do not. Shared cloud means one company’s machines can run more than one chatbot even when those chatbots sell against each other.

Microsoft Azure is widely suspected as that shared layer. No named company has confirmed Azure as the root cause.

Josephine Walker at Axios writes that Azure, which provides cloud services for all three models, was also seeing outages, and that some outlets have suggested Microsoft’s problem may be contributing.

Ben Schoon at 9to5Google makes the same point: all three platforms rely on Azure among other providers. The Register notes AWS, Google Cloud, and Microsoft Azure status pages showed no relevant technical difficulties at filing, pages that do not always update in time.

Three rival dashboards went dark while the cloud’s own page stayed quiet. Switching chatbots only helps if they do not share the same machines.

OpenAI called it a routing error and Anthropic called it infrastructure. Grok answered with a prompt that the model was overloaded right now. Three different self-diagnoses on the same morning leave Azure as a suspicion, not a finding.

Google did not declare an official Gemini outage. Ars reports DownDetector Gemini reports moved from 23 around 10:30am to 412 just after 11am, and StatusGator flagged a likely Gemini API outage from 10:45 to 11:15am. A user-report spike is not the same as the ChatGPT, Claude, and Grok dashboards.

Cursor confirmed an outage because Grok and Claude were down, 9to5Google reports. Downstream tools inherit the shared rack even when they never sold a chatbot.

Downdetector logged more than 35,000 user-submitted OpenAI reports at peak, about 80 percent tied to ChatGPT, Newsweek writes. USA Today puts more than 38,000 ChatGPT reports as of 11am ET, and Ars has Grok reports moving from fewer than 10 just before 9am to 1,365 by 9:45am. Cloudflare told Newsweek it was not seeing significant disruptions, and that any reporting that says otherwise is incorrect.

Adrian Parham at TechTimes frames Azure East US as the common thread, citing a StatusGator user-submitted report of inbound traffic failures at 10:26 a.m. PT. That is a user report, not a company saying what broke.

ChatGPT and Claude were back to fully operational status by 1pm ET, USA Today writes. Ars times OpenAI’s elevated errors as resolved at 12:55pm ET. 9to5Google has all three back by 12:38pm PT.

The chatbots came back. Azure is still not on the ticket.

—

Sources

One thought on “Three rival chatbots failed Thursday while Microsoft Azure’s status page stayed quiet”

Leave a Reply

Your email address will not be published. Required fields are marked *