Microsoft traced its global Microsoft 365 outage to three converging faults in the networking gateway fronting Exchange Online and other services, which triggered a cascading failure that left more than 80% of gateway servers unresponsive in under two hours. A latent lock contention bug deployed in July went undetected for over three months because test labs used lower-core hardware, until a Teams failover 15 hours before the Nov 12 incident combined with a nearly 14% capacity reduction from an unrelated prior-day issue to overwhelm servers during peak US traffic.
Automated routing shifted traffic from stalled machines to remaining healthy endpoints, spreading the failure across global regions after partition bulkheads failed to contain it. Recovery stretched past 35 hours before all services were restored on Nov 14 because waiting traffic repeatedly froze restarted servers and standard rollbacks could not revert the months-old code defect, forcing engineers to take regional server groups offline in batches of 50% or more to patch and restart them.
The disruption affected users across 74 countries and territories, halting email and workplace tools for banks, courts, hospitals, police forces and government agencies while delaying earnings releases at Richemont and Workhorse. Anderson Economic Group estimated lost productivity at up to $7 billion, with insurer claims projected at $300 million to $500 million, while Microsoft Threat Intelligence warned of an active phishing campaign using real incident code EX1550543 and fake 24-hour message deletion alerts.




