Microsoft says a software flaw that escaped testing contributed to last week’s widespread disruption of its workplace services.
The bug was introduced in July but went unnoticed until a surge in traffic and reduced server capacity set off a chain reaction during peak U.S. business hours on Nov. 12, Microsoft executive Charles Lamanna said.
Lamanna apologized to customers, acknowledging that the outage interfered with their day-to-day work.
Outlook email, calendars and meeting invitations were among the services affected. The problem also disrupted other Microsoft 365 tools that businesses use routinely.
Three issues combined to overload a network gateway serving multiple Microsoft 365 services: a hidden software defect, heavier-than-usual demand and fewer available servers. The bug, Lamanna said, caused servers to become unresponsive under certain conditions.
The faulty change was created for Microsoft 365 gateway servers and does not affect general Windows deployments. It had been running around the world for more than three months without incident. Microsoft’s testing did not reproduce the combination of traffic and hardware that exposed the defect. In particular, the issue proved more severe on newer, more powerful machines, while tests tended to use less powerful hardware.
The outage began after Teams traffic was moved onto a system that also supports Microsoft email. Nearly 14% of that system’s capacity was already unavailable because of a separate problem, and the added traffic revealed the flaw.
When servers stopped responding, Microsoft’s systems redirected traffic to machines that were still functioning. Those servers then became overloaded as well. At the worst point, more than 80% of the servers were unresponsive, Lamanna said. Another issue prevented safeguards from containing the failure within parts of the network, allowing it to spread globally.
Restoring service took time. Servers that came back online were hit immediately by waiting traffic and failed again. Microsoft’s automated recovery systems were not built to handle an outage on this scale, and its usual rollback tools could not quickly reverse the months-old software change.
Engineers eventually applied a fix, restarted servers in groups and gradually brought them back into service. Teams, OneDrive, SharePoint and Office apps recovered sooner after being shifted away from the faulty infrastructure. Teams chats and meetings continued to work normally for most of the outage, Lamanna said, but Exchange Online, which runs Microsoft’s email services, could not be moved in the same way.
Lamanna said the incident was not caused by a cyberattack or artificial intelligence tools. Microsoft’s Copilot assistant was also affected, while airlines IndiGo and Akasa Air reported problems with customer-support email systems. OpenAI and Anthropic also reported issues with their Microsoft 365 integrations.
Microsoft plans to share more information after completing a review. Lamanna said the company would study what happened and work to prevent a similar disruption.


