ChatGPT, Claude, and Grok all went down at once

ChatGPT, Claude, and Grok all went down within the same window on September 3, 2026. BleepingComputer reports the ChatGPT outage began around 10:58 a.m. ET. Anthropic's Claude and xAI's Grok showed problems on their own status pages at the same time. No company has confirmed a shared cause, and nobody should assume there is one yet.

What actually broke, and for how long

The outage was not narrow, at least on the OpenAI side.

BleepingComputer lists conversations, login, search, file uploads, Voice mode, GPTs, image generation, Deep Research, Agent, and ChatGPT Work among the affected services. OpenAI's own status page gave one line: "We are investigating the issue for the listed services."

9to5Mac reports that services appeared to recover within about 30 minutes. BleepingComputer confirms Claude had simultaneous outages affecting multiple models during the same window. The precise list of affected Claude products was not available from Anthropic's status page at the time of writing.

9to5Google adds a detail worth noting for anyone building on these platforms. Cursor, the AI coding tool, was affected too. Google's own Gemini was not.

What is confirmed, and what is only a guess

Two separate theories circulated within hours of the outage. Neither is confirmed by any of the companies involved.

The first theory points at shared infrastructure. 9to5Google notes that OpenAI, Anthropic, and xAI all rely on Microsoft Azure for cloud services, and frames that as a likely explanation, without confirmation from any of the three. The second theory ties the timing to OpenAI's next model launch. 9to5Mac reports that OpenAI's own accounts posted a cryptic line, "The stars are almost aligned," which some read as a hint at planned rather than accidental downtime. OpenAI has not confirmed any connection to a launch.

xAI's response was brief. 9to5Google's own wording for the acknowledgment reads: "SpaceX AI confirmed Grok is having a service issue." No further detail was given.

What we know	Source
Outage start, ChatGPT	About 10:58 a.m. ET, per BleepingComputer
Recovery time	About 30 minutes, per 9to5Mac
Companies affected	OpenAI, Anthropic, xAI, per all three outlets
Also affected	Cursor, per 9to5Google
Not affected	Google Gemini, per 9to5Google
Confirmed root cause	None, from any company

What this means for developers

Treat this as evidence for a design decision, not just as a news story. Three separate AI providers went down inside the same window. The working theory, still unconfirmed, is a shared cloud dependency. If your product calls a single model provider with no fallback, this is exactly the failure mode you are exposed to. That holds whether or not Azure turns out to be the actual cause here.

Build the fallback before you need it, not during the next outage. A second provider, even a slower or more expensive one, kept warm behind a feature flag turns an outage into a degraded-mode incident instead of a full stop. Test that path on a normal day. A fallback nobody has exercised is not a real fallback.

Watch your own status page discipline against OpenAI's example here. "We are investigating the issue for the listed services" is a real status update, timestamped and specific about which services are affected, even without a root cause yet. If your own service has an incident, match that specificity. Name exactly what is and is not working. That helps the people depending on you far more than a vague "we are aware of the issue."

Do not treat the Azure theory as fact, including in your own postmortems. It comes from press speculation based on which vendors share a cloud provider, not from any of the three companies. If you are debugging your own outage today and it correlates with this one, check your own logs before assuming the cause.