Microsoft 365 Outage Halts Email and AI Connectors
A failure in Exchange Online disrupted Outlook, meeting scheduling and AI tools for U.S. users while Microsoft has yet to confirm full recovery.
Updated on Nov. 13, 2026 in Data Centers
Live Poll
Do you trust cloud services like Microsoft 365 to stay reliably available?
Microsoft 365 services experienced disruption and degradation beginning November 12, 2026, with Outlook users losing connections shortly before 1 p.m. ET and affected services depending on Exchange Online. ChatGPT and Claude connectors to Microsoft 365 also failed during the disruption; neither company reported an outage of its core AI models.
Why it matters
The incident shows how much enterprise email, calendar and AI-assistant workflows now depend on a single cloud provider's internal health-check and routing systems. It arrives as cloud infrastructure spending hits $168 billion in a single quarter, concentrated among a few providers.
Synergy Research reports third-quarter cloud infrastructure spending of $168 billion, up 55% year over year, with AWS at 26%, Microsoft at 19% and Google at 16% of the market. Microsoft has not confirmed full restoration or a permanent fix, and the precise trigger for the overload remains unexplained.
The players
Microsoft
Microsoft operates Microsoft 365, Exchange Online and Azure, and holds a 19% share of the cloud infrastructure market.
OpenAI
OpenAI makes ChatGPT, whose connectors to Microsoft 365 reported problems during the disruption.
Anthropic
Anthropic makes Claude, whose Microsoft 365 connector saw most email, calendar and search requests fail.
Synergy Research
Synergy Research is a market analysis firm that tracks cloud infrastructure spending and provider market share.
The details
Some servers failed automated health checks, the automated tests that decide whether a machine is working, and dropped out of service, leaving the remaining machines to absorb more requests. Engineers restarted unresponsive servers and tested a different health-check method, while redirecting Outlook connections away from affected infrastructure in the United States. Some servers returned to service only to become unresponsive again once traffic resumed, which pointed engineers toward an overloaded core process and a suspected fault in the health-check component.
Timeline
November 12, 2026, shortly before 1 p.m. ET: Outlook users began losing connections.
November 13, 2026, 2:14 a.m. UTC: Microsoft's service-health dashboard still reported degradation.
The Tech Race
Synergy Research places AWS, Microsoft and Google at a combined 61% of cloud infrastructure, leaving the three leaders with most of the enterprise workloads now affected by the outage. Gartner expects AI infrastructure spending to reach $2.4 trillion in 2027, which means more AI tools will rely on the same concentrated infrastructure.
Anyone whose Outlook, Teams or Copilot access runs through Microsoft 365 may have seen delayed email, failed meeting scheduling or Copilot Chat that would not load during the disruption. Teams chats, online meetings and SharePoint files stayed available, so workers can fall back on those while Microsoft completes recovery.
The takeaway
Watch for Microsoft's confirmation of full restoration and any published root-cause analysis of the overloaded core process. Teams that depend on Outlook or ChatGPT and Claude connectors should note which functions remained working, since the failure hit mail and connectors while chats and file storage held up.
Further reading
Learn more about how cloud providers run and protect these systems in our Data Centers coverage.
Live Poll
Do you trust cloud services like Microsoft 365 to stay reliably available?
