Microsoft attributed the November 12, 2026, global Microsoft 365 networking-gateway outage to three conditions: a latent defect in a gateway-specific fix, a shift in Teams traffic, and nearly 14% of gateway capacity unavailable for an unrelated issue. Charles Lamanna, a Microsoft executive, published the company’s explanation on November 16. Lamanna’s account describes how traffic redirection spread the failure and why recovery varied across services.

What caused the Microsoft 365 outage?

Microsoft said the defect involved lock contention inside an internal process. Under certain load conditions, that contention could make a gateway server unresponsive. The fix containing the defect was developed for Microsoft 365 networking-gateway servers; Lamanna said it did not apply to broader Windows deployments.

The change began rolling out in July 2026 and reached worldwide deployment over several weeks. Microsoft said it ran for more than three months without observed impact. The company attributed the testing gap to differences in traffic and product-SKU mixes, and to test labs’ tendency to use older machines with fewer cores. According to Microsoft, the issue became more pronounced on newer hardware with more cores.

When traffic shifts onto a gateway already serving other services, a latent defect and reduced capacity can turn a routine failover into a wider failure.

How did the failure spread?

About 15 hours before the outage, Teams failed over from an alternative gateway to the primary gateway, which also served Exchange Online. Microsoft described the failover as normal operation and said it changed traffic patterns and increased load on the primary gateway.

At the same time, nearly 14% of gateway capacity was out of service because of an unrelated issue from the previous day. Microsoft said that reduction alone had no impact because the service had enough spare capacity. It became part of the wider failure when the other conditions converged.

As servers stopped responding, traffic-management systems redirected requests to remaining endpoints. That concentrated demand on fewer servers, where the same defect could recur. Microsoft said the failure spread across its gateway fleet in less than two hours. At the lowest point, more than 80% of the affected gateway servers were unresponsive. A separate issue also prevented normal partition boundaries from containing the cascade across the global gateway infrastructure.

Why did recovery differ between services?

Microsoft said Teams, OneDrive, SharePoint and Office apps could fail away from the faulty gateway and recover sooner. Exchange Online could not. Microsoft also said Teams chats and meetings functioned without issue for most of the outage.

Restoration was complicated by queued traffic: returning servers were overwhelmed as waiting requests reached them. Microsoft said ordinary rollback, flighting and feature-flag processes could not quickly reverse a defect introduced months earlier. Engineers restored the code path to a previously known-good state, then took server groups out of service, applied the change, restarted the servers and returned them to service.