One Process Took Down Email for 430 Million People

An Exchange Online connection process overwhelmed its servers on 12 November, taking Outlook, Teams, SharePoint and Copilot offline across 74 countries for roughly 20 hours.

Microsoft 365 suffered a global outage beginning around 1:45pm Eastern on 12 November 2026, when a process on the affected servers responsible for establishing secure connections became overwhelmed and stopped processing requests; Microsoft said a large number of servers became unexpectedly unavailable, shifting load onto the remaining infrastructure. Impacted services included Exchange Online, Teams, SharePoint Online, Copilot, Defender XDR, Planner, To Do and Purview, across about 74 countries, with the company extending its email timeout window to 48 hours to protect delayed messages. Recovery began on 13 November and services were still returning in batches into 14 November, with the company describing the fix as a validated configuration change plus an automated process to restart unresponsive servers, alongside the reversion of a recent network configuration change. Microsoft denied a cyberattack. Reporting put the potential exposure at about 430 million users, though that figure is an inference from the install base rather than a Microsoft disclosure. The commercial read is that the outage validated a risk that procurement teams had mostly treated as theoretical: a single vendor’s control plane is a single point of failure for email, chat, files and the AI layer sitting on top of all three, and the follow-up question from enterprise buyers will be about contractual recovery guarantees rather than feature roadmaps.