On Thursday, 12 November 2026, as office operations were underway as usual around the world, the Microsoft 365 network experienced a sudden disruption. Employees were left stranded, unable to send emails, view calendars, or access other collaboration tools.

Within an hour, the impact was felt by users across 74 countries, including the United States, the United Kingdom, Japan, Singapore, and Australia. Initially identified around 11:17 a.m. Eastern Standard Time (ET) as an Exchange Online issue, the situation soon escalated into a widespread outage. By 1:00 p.m., complaints regarding Outlook received on the Downdetector website stood between 130,000 and 138,058, a surge over 2,000 times higher than normal levels.

Key Points

  • On 12 November 2026, services were disrupted across 74 countries, including the US, the UK, and Japan.
  • Outlook incident reports lodged with Downdetector rose to a record 138,058.
  • Microsoft stated this was not a cyberattack, but rather a technical glitch on its servers.

Disrupted services and AI tools

The hardest hit by this disruption was the Exchange Online system. Outlook email, calendars, and message delivery ground to a complete halt. Furthermore, Microsoft Teams, SharePoint Online, Copilot, Microsoft Defender XDR, Planner, To Do, OneDrive, as well as several Outlook.com and Hotmail accounts, also experienced interruptions.

The issue extended beyond Microsoft services to heavily impact the artificial intelligence sector as well. Disruptions to Microsoft 365 connectors used by OpenAI and Anthropic halted the operations of several tools, including Claude for Outlook, the companies said.

What caused the outage?

Investigations conducted by Microsoft revealed that the issue was triggered when a large number of servers unexpectedly went offline following automated health check failures. Consequently, excessive loads were placed on the remaining servers, rendering systems unresponsive. However, the company stressed that this was not caused by user traffic surges or a cyberattack.

To restore the system, steps were taken to restart servers and temporarily suspend new deployments. Although restarting the servers yielded positive results by the early hours of 13 November, service recovery for certain accounts reportedly remains gradual.

What lies ahead?

A comprehensive post-incident review is scheduled to be released in due course.