Microsoft reported that several of its major services, including Outlook, Teams, and Microsoft 365, experienced outages affecting thousands of users in the United States on Thursday, November 12. The disruption also impacted integration connectors used by artificial intelligence providers OpenAI and Anthropic. According to the outage tracking website Downdetector, nearly 130,000 users reported issues with Outlook shortly after 1 p.m. ET.
The company stated that connectivity errors were caused by a portion of its infrastructure experiencing an elevated load across multiple regions. Microsoft clarified that this load was not due to an increase in traffic but resulted from a number of servers becoming unexpectedly unavailable, which placed a strain on the remaining infrastructure. The company noted that these servers were failing automated health checks.
To address the outage, Microsoft paused all deployments and began routing traffic away from the affected infrastructure. Technicians tested configuration changes and initiated restarts of the impacted servers. By late afternoon, the company reported it had validated an alternative health check method that allowed a subset of servers to return to service. While a previous Teams disruption in North America had been resolved earlier in the day, Microsoft noted that the broader issue could still affect Microsoft 365 connectivity in other regions.
The scale of the disruption was primarily focused on U.S. users, but Microsoft indicated that the connectivity issues had the potential to reach users in other global regions as the day progressed. OpenAI and Anthropic confirmed their services were impacted by the failure of Microsoft’s connectors.
As of Thursday afternoon, Microsoft was continuing to apply a new service health check method more broadly and was preparing automated recovery systems to detect and restart unresponsive servers. The company's goal is to maintain stability while the configuration changes progress. While some servers have returned to service, the company did not report a final deadline for full restoration of all impacted regions. Users may continue to see fluctuations in service availability as the automated recovery systems are deployed.