What Happened and Which Services Were Hit?
On Thursday, November 12, 2026, users globally began reporting issues accessing a wide range of Microsoft services. The problems, which started around 6 p.m. GMT, primarily centered on Exchange Online, the cloud service that powers Outlook's email functions.
The impact quickly cascaded to other connected applications. Users reported being unable to access their email inboxes, send or receive messages, and join meetings. The disruption affected not just Outlook but also Teams, SharePoint, OneDrive, and the AI assistant Copilot, impacting businesses, governments, and even healthcare services worldwide.
The Prime Suspect: A Server Cascade Failure
Microsoft's initial investigation pointed away from malicious activity and towards an internal technical failure. The core issue appears to be a domino effect that started when a large number of servers became 'unexpectedly unavailable'. This wasn't due to an increase in user traffic, but rather an internal problem. According to Microsoft, some servers failed automated health checks and were taken offline. This sudden loss of capacity placed an overwhelming load on the remaining infrastructure. A process responsible for establishing secure connections on the servers became overwhelmed, causing them to stop processing requests and leading to the widespread connectivity errors users experienced.
The Problem with the 'Fix'
Fixing a system of this scale is not as simple as flipping a switch. Engineers had to identify the root cause while managing a fragile system. Their attempts to bring services back online were complicated. At one point, Microsoft noted that as they restored servers and traffic began to return, the servers would become unresponsive again, prolonging the outage. This indicates a deep-seated issue, likely related to a recent configuration change that had unintended consequences. The company was forced to roll back recent changes and carefully limit traffic to diagnose the problem, which explains why the restoration process took so many hours.
Why 'Cloud' Doesn't Mean 'Failsafe'
This event serves as a powerful reminder that 'the cloud' is just someone else's computer—or in this case, a vast, interconnected network of them. While cloud services like Azure are designed with redundancy, with infrastructure split into different physical zones to prevent single points of failure, some problems can bypass these safeguards. Failures in core, globally shared components, like authentication systems or critical configuration data, can have a worldwide impact that regional separation cannot prevent. Incidents like this highlight the systemic risk in having so much of the world's digital infrastructure dependent on a small number of hyperscale providers.
















