On this page · 4 sections ▾
Microsoft has traced the global Microsoft 365 outage of November 12 to a software defect introduced in July, according to a preliminary post-incident report published on November 15 and summarized by The Register. The bug sat unnoticed in network gateway servers until a load surge and a drop in capacity triggered it, producing a cascading failure that took Exchange Online offline for roughly a day.
Key facts
- The defect arrived with a July 2026 update prepared for Microsoft 365 network gateway servers. It caused lock contention inside an internal Windows process, exhausting CPU and memory.
- Two events coincided: Teams failed over to the primary gateway that supports Exchange Online, and almost 14% of gateway capacity was out of service.
- At the worst point, more than 80% of servers were unresponsive, and the defect was hit across the fleet within two hours.
- Microsoft's fix returned the code path to a previously known good state in Windows. It was applied to server groups in batches.
- A final post-incident report is due on November 20.
How the failure unfolded
Microsoft's report says its load testing focused on weaker hardware with fewer cores, while the contention problem worsens on newer, multi-core machines. XDA notes that EVP Charles Lamanna described the same mechanism on X.
Neither trigger was a problem alone. The Register reports that the Teams failover was not unusually large, and Microsoft said the service had enough headroom despite the reduced capacity. Together they pushed the gateway into the defect. Traffic from failed servers was rerouted to healthy ones, which then hit the same bug.
Recovery was slow. Restored servers were overwhelmed again by pent-up traffic, and automated recovery was not designed for such a broad unresponsive state. A simple rollback was also impossible because the defect was months old. TechRadar adds that Microsoft said a separate issue kept its protections against cascading failure from working properly.
The Register's timeline puts the Teams failover at 0107 UTC on November 12 and the first detection of connectivity issues at 1603 UTC. Microsoft identified a mitigation at 0720 UTC on November 13, confirmed health at 1610 UTC, and sent its final customer update at 0415 on November 14. Services such as Teams, OneDrive and SharePoint could fail out of the faulty gateway, but Exchange Online could not for scale reasons.
Why it matters
Earlier statements were less specific. According to Android Authority, Microsoft first cited an issue during an earlier network capacity upgrade and said the incident was unrelated to cyber activity or AI agents. The report now gives a fuller root cause and rules out the theories that circulated during the outage.
The real-world effects were wide. Android Authority lists disruptions at courts, government agencies, schools, airlines and the NHS call handling in the UK. In Germany, WDR reported that a chlorine dosing fault at a Phantasialand hotel went unnoticed because the alert was sent by email. Fifteen guests were triaged and four went to hospital.
What to watch next
The November 20 final report should show whether Microsoft changes its testing to cover newer, high-core hardware and whether it redesigns recovery for large-scale unresponsive states. The Register questions how a defect that survived three months reached production with load tests that did not reflect conditions Microsoft controls.




