Microsoft 365 Outage November 2026: What Happened and Why?
On November 12, 2026, a significant global disruption affected Microsoft 365 services, including Exchange Online, Outlook, Microsoft Teams, SharePoint and OneDrive. Microsoft has now shared a detailed explanation of what happened, identifying three factors that combined to cause the outage.
The explanation shared by Charles Lamanna at Microsoft provides a clearer picture of how a software defect, changing network traffic and temporarily reduced server capacity interacted. It also explains why the incident spread beyond the initial infrastructure and why recovery took longer than usual.
This article summarizes Microsoft's published findings and engineering response. It offers a useful look at the complexity of operating cloud services at global scale and the work involved in restoring them when multiple safeguards are challenged at once.
What Caused the Microsoft 365 Outage?
Microsoft identified three main factors behind the November 12 incident:
- A latent software defect in the Microsoft 365 networking gateway.
- An increase in traffic after Microsoft Teams failed over to the primary networking gateway.
- Approximately 14% of the gateway's server capacity being temporarily out of service.
Microsoft said none of these factors would have caused the global outage independently. Their combination triggered a failure across the networking gateway infrastructure that fronts many Microsoft 365 services.
1. A Software Defect That Had Gone Undetected
The primary technical issue was a latent defect involving lock contention inside an internal process running on Microsoft's networking gateway servers. Under certain load conditions, the contention caused servers to become overwhelmed and unresponsive.
Microsoft explained that the defect originated in a fix prepared specifically for Microsoft 365 networking gateway servers. The defect and the fix do not apply to broader Windows deployments.
The change began rolling out in July 2026 and was deployed globally over several weeks through Microsoft's safe deployment process. It operated for more than three months without any observed impact.
Microsoft's normal load testing did not expose the issue because the traffic patterns and hardware mix in its test environments differed from production. The contention was more pronounced on machines with more processor cores and newer hardware, while testing tended to focus on less powerful hardware configurations with fewer cores.
As a result, the defect remained latent until production encountered the particular combination of conditions that triggered it.
2. Increased Traffic Following a Teams Failover
Microsoft operates two independent and heterogeneous networking gateway systems to provide isolation and high availability for Microsoft 365 services.
Approximately 15 hours before the outage, Microsoft Teams failed over from the alternative gateway to the primary gateway, which also supports Exchange Online.
Microsoft described this failover as part of normal operations and an important element of its resiliency approach. However, the change increased the load on the primary gateway and altered its traffic patterns.
The additional load made it more likely that the latent software defect would be triggered.
3. Temporarily Reduced Server Capacity
At the same time, approximately 14% of the networking gateway's server capacity was out of service because of an unrelated issue from the previous day.
Microsoft said this capacity was already on track to return to active service through routine maintenance and health processes. The remaining infrastructure had sufficient capacity headroom, so the reduced capacity did not cause an outage on its own.
However, when combined with increased traffic and the software defect, it contributed to the conditions that caused the gateway servers to become unresponsive.
How the Problem Spread Across the Network
Microsoft's traffic management systems are designed to redirect users to healthy endpoints when servers lose connectivity. This is a normal resiliency mechanism intended to maintain service when individual hardware, software or network components experience problems.
During this incident, redirected traffic placed additional load on the remaining healthy endpoints. As more servers became unresponsive, traffic was redirected again, spreading the load across fewer available servers.
Microsoft described this as a cascading failure. In less than two hours, the defect had been encountered across the affected fleet. At the lowest point, more than 80% of the networking gateway servers were unresponsive.
The gateway normally uses thousands of capacity units, organized into regions. These units provide separation between different portions of the infrastructure. Microsoft said that, due to a separate issue, the normal bulkheads between these partitions did not prevent the cascading failure from spreading across the global gateway infrastructure.
Why Were Some Microsoft 365 Services Restored Earlier?
The affected networking gateway supported many Microsoft 365 services, but not all services depended on it in the same way.
Microsoft said services including Teams, OneDrive, SharePoint and Office applications were able to fail out of the faulty gateway infrastructure. This helped those services recover earlier. Teams chats and meetings continued to function for most users during much of the outage.
Exchange Online, however, could not fail out of that networking gateway. This difference in service dependencies helps explain why Exchange Online and some related functionality took longer to recover.
Microsoft 365 is an interconnected platform, but its workloads do not all have identical infrastructure dependencies or recovery paths.
Why Did Recovery Take Longer Than Usual?
Restoring the affected servers was not as simple as bringing them back online. Microsoft said that as servers returned to service, they were immediately overwhelmed by waiting traffic and became unresponsive again.
Its normally automated recovery processes were not designed to restore service from such a widespread unresponsive state. The age of the defect also made recovery more difficult: because the problematic code had been deployed months earlier, the usual rollback, flighting and feature-flag mechanisms could not quickly reverse it.
After engineers confirmed the defect, they developed an update that returned the affected code path to its previously known good state. Applying the fix required a controlled process:
- Groups of servers were taken out of service.
- The corrective update was applied.
- The servers were restarted.
- The updated servers were gradually returned to service.
Microsoft said this process was applied to large quantities of infrastructure at a time, including groups representing more than 50% of a region.
What This Incident Shows About Cloud Resiliency
The November 12 outage highlights the complexity of operating a global cloud platform. Microsoft uses independent gateway systems, traffic management, capacity partitions and automated recovery processes to help maintain availability.
In this case, a software defect that had operated for months without observed problems was triggered by a particular combination of traffic and hardware conditions. The resulting server failures then placed additional pressure on the remaining infrastructure as traffic was redirected.
The incident also illustrates why safe deployment and load testing are ongoing engineering challenges. Testing can reduce risk, but production environments contain a wide range of hardware, traffic patterns and operating conditions that are difficult to reproduce completely.
Microsoft's explanation gives customers and IT professionals a clearer understanding of how the incident developed, why the impact spread, and why restoring service required a carefully staged process.
Microsoft's Post-Incident Review
Microsoft acknowledged the disruption to customers and apologized for the impact on normal business operations. The company also indicated that a final Post-Incident Review (PIR) would provide additional details later in the week.
The PIR may provide further information about the investigation and any additional improvements identified following the incident. This article is based on the explanation shared by Charles Lamanna and should be read alongside Microsoft's subsequent review when it becomes available.
For administrators, the Microsoft 365 Admin Center's Service Health section remains an important place to review service incidents, affected workloads and Microsoft's communications.
Final Thoughts
The November 12, 2026 Microsoft 365 outage resulted from three contributing factors: a latent software defect in the networking gateway, increased traffic following a Teams failover, and temporarily reduced server capacity.
Microsoft's detailed explanation provides a valuable look at how these factors combined, how the failure spread across the infrastructure, and why recovery required a controlled engineering effort.
Operating cloud services at this scale is a significant engineering undertaking. Sharing the technical findings helps customers better understand the incident and gives the wider IT community useful insight into the challenges of designing and maintaining resilient services.
I'll update this article if Microsoft's final Post-Incident Review adds significant details to the findings discussed here.
Source: Charles Lamanna, Microsoft, November 2026 statement on the Microsoft 365 outage. The statement notes that further details are available in the Microsoft 365 Admin Center and that a final Post-Incident Review was forthcoming.
#Microsoft365 #MicrosoftOutage #ExchangeOnline #MicrosoftTeams #SharePoint #CloudComputing #MicrosoftMVP



