tech
What We Know About Microsoft's November 2026 365 Outage

Three separate technical failures, each manageable on its own, combined on November 12 to take down large portions of Microsoft 365 and Exchange Online for roughly 24 hours, according to a preliminary Post-Incident Report cited by Ars Technica. The company's own classification of the incident: "a technical issue not related to cyber or AI/agents."
What caused the outage?
A defect introduced into the cloud gateway in July sat dormant for months, Ars Technica reports. Under certain load conditions, the bug let CPU and memory resources on gateway servers become exhausted and unresponsive. Pre-release testing missed the flaw because the test hardware ran fewer CPU cores than some machines already in production, where higher core counts apparently helped trigger the lockout, per the report.
How did a routine traffic shift expose the bug?
About 15 hours before the outage began, traffic from Teams was moved from an alternate gateway to the primary gateway that also handles Exchange Online access, according to the report cited by Ars Technica. The move was described as a standard fail-safe tied to network maintenance, one of two independent gateway systems the company keeps for resilience. But it changed both the volume and the mix of requests hitting the primary gateway.
When did the outage start and end?
The report places the incident between 16:20 UTC on November 12 and 16:10 UTC on November 13, just under 24 hours, with additional monitoring through November 14, Ars Technica reports. The sequence, as described in the report:
- July: A gateway code change introduces a latent defect.
- ~15 hours before impact: Teams traffic shifts onto the primary gateway during scheduled maintenance.
- 16:20 UTC, Nov. 12: Gateway servers begin freezing as US business traffic hits daily peaks; the outage officially starts.
- 16:10 UTC, Nov. 13: The incident is marked resolved, just under 24 hours later.
- Nov. 14: Monitoring continues on affected systems.
What role did reduced server capacity play?
An existing fault had left the gateway server fleet running 14 percent below optimal capacity, the report states, as cited by Ars Technica. That missing headroom would normally have absorbed the shifted traffic without incident. Combined with the changed request patterns and the dormant defect, the shortfall produced lock contention as requests piled up, and servers began refusing new connections once US business hours hit their peak.
What happened to backup systems, and is the issue fixed?
Backup systems rerouted requests away from unresponsive servers as designed, Ars Technica reports, though the report does not specify why rerouting alone did not prevent broader disruption. The incident is tracked under IDs EX1550543 and MO1550821. Full details require admin credentials to view on the company's status pages; Ars Technica notes that Office Watch reproduced the preliminary report and the University of Pennsylvania maintains an archive of related service-health updates.
Key terms
- Post-Incident Report (PIR): A formal internal document explaining the cause, timeline, and fix for a service disruption.
- Cloud gateway: The front-door system that routes user requests to backend services like Exchange or Teams.
- Lock contention: A condition where multiple processes compete for the same limited resource, slowing or halting requests.
Questions
What caused the Microsoft 365 outage in November 2026?
A hidden gateway defect dating to July, a routine traffic shift that moved Teams requests onto the primary gateway, and a server fleet running 14 percent below optimal capacity combined to overload gateway servers, according to the company's preliminary incident report cited by Ars Technica.
How long did the outage last?
The incident ran from 16:20 UTC on November 12 to 16:10 UTC on November 13, just under 24 hours, with monitoring continuing through November 14, per the report.
Was the outage caused by a cyberattack?
No. The company's report classifies the incident as a technical issue unrelated to cyber or AI/agent activity, according to Ars Technica.