ChatGPT, Claude and Grok All Went Down at Once on September 3 — Here's What Happened

OpenAI's ChatGPT, Anthropic's Claude, and xAI's Grok all suffered outages within the same general window on September 3, 2026 — a notable anomaly, since individual platform outages happen regularly but three major, independently operated AI services failing at roughly the same time is unusual.

What Happened, Service by Service

• ChatGPT: a routing error beginning around 7:43 am PT made ChatGPT and Codex unavailable for some users across platforms; a fix was implemented by approximately 8:17 am PT
• Claude: Anthropic reported that Claude Opus 4.8 and Opus 5 continued experiencing outages while other Claude models had recovered to baseline error rates
• Grok: xAI's chatbot was also down, with the company acknowledging it was working to address the problem

The Suspected Common Thread

Several outlets reported that Microsoft Azure, which provides cloud infrastructure used by multiple AI providers, was separately experiencing its own outages around the same time, and suggested this may have contributed to the correlated AI service disruptions. This connection has been reported by multiple outlets but has not been uniformly confirmed as the direct cause by all of the AI providers involved.

What This Means for Enterprise AI Reliance

The immediate impact disrupted workflows across tech, media and finance, reflecting how deeply embedded AI assistants have become for coding, research and content generation. For enterprises building critical workflows on top of any single AI provider, this incident is a concrete reminder that shared underlying cloud infrastructure can create correlated failure risk across ostensibly competing AI vendors — meaning a single point of infrastructure failure can disrupt multiple 'redundant' AI tools simultaneously, undermining an assumption that using multiple AI vendors automatically provides meaningful failover redundancy.