ChatGPT, Claude, and Grok all went down within minutes of each other this morning.

Gemini never reported an outage.

Everyone's making the same joke about withdrawal symptoms and doing their own thinking for an afternoon. But three competing labs failing at the same minute is the real story.

Three independent companies don't fail simultaneously. Unless they aren't independent.

No cause announced yet. Cursor and Copilot went down too, which is what happens when something upstream breaks and everything built on it follows.

And the one that stayed up designs its own chips and runs its own data centers. Proves nothing by itself. Quite a coincidence though.

For months the AI conversation has been about price. Metering, rationing, pay per token. I've written plenty of it myself.

This morning was about something else. You don't only rent the intelligence. You rent the uptime, from a supply chain you can't see.

If your product has one model provider, you don't have a vendor. You have a single point of failure with an invoice attached.

The good news is it was a cheap lesson. Blank prompt boxes, lost hours, nobody hurt. Better in year three than year ten, when this is wired into payroll and diagnostics and the grid.

What's your fallback provider? And have you actually tested it?