Microsoft explains global 365 outage: How a four-month-old software bug triggered disruption and delayed recovery
Microsoft has explained the technical failures behind the global Microsoft 365 outage on November 12, saying a combination of a software defect, increased network traffic and reduced server capacity triggered a cascading failure across its networking infrastructure.
Charles Lamanna, Executive Vice President of Copilot, Agents and Platform at Microsoft, said the outage was caused by three factors that occurred simultaneously. Together, they overwhelmed the company’s networking gateways, leaving a large proportion of servers unresponsive and disrupting services including Teams, Exchange Online, OneDrive, SharePoint and Office applications.
At its worst, more than 80% of the affected networking gateway servers were unresponsive.
LaManna said the underlying software defect had existed since July 2026 but remained undetected for more than three months because its regular testing did not replicate the hardware and traffic conditions that eventually triggered the failure.
Here is an explainer of what happened, why the outage spread globally and how Microsoft restored its services.
What caused the Microsoft 365 outage?
According to Lamanna, three factors combined to trigger the disruption.
1. A software bug that had gone undetected for months
The first problem was a latent software defect introduced through a code change prepared specifically for Microsoft 365 networking gateway servers.
The defect involved lock contention inside an internal process. Under certain traffic conditions, this caused servers to become overwhelmed, stop responding and eventually become unavailable.
The change was first deployed in July 2026 and gradually rolled out across Microsoft’s global infrastructure over several weeks.
However, the rollout did not trigger any noticeable problems for more than three months.
Microsoft said its regular load testing failed to identify the issue because the test environments used different traffic patterns and a different mix of server hardware.
The problem was particularly severe on machines with more processing cores and newer hardware, while the company’s testing focused largely on older, less powerful server configurations.
Microsoft clarified that the defect and the code change were specific to Microsoft 365 networking gateway servers and did not affect broader Windows deployments.
2. A sudden increase in network traffic
The second factor was a surge in traffic reaching the affected networking gateway.
Microsoft operates two independent network gateway systems to maintain service availability during disruptions.
Around 15 hours before the outage, Teams failed over from the alternative gateway to the primary gateway, which also supports Exchange Online.
Such failovers are part of Microsoft’s normal resilience and recovery operations.
However, the shift increased the traffic load on the primary gateway and changed its traffic patterns, making it more likely that the latent software defect would be triggered.
3. Nearly 14% of server capacity was unavailable
The third factor was a reduction in available server capacity.
Nearly 14% of the networking gateway’s server capacity was already out of service because of an unrelated issue from the previous day.
Microsoft said the affected capacity was expected to return through routine maintenance and system health processes.
On its own, this reduction was not enough to disrupt services because the network had sufficient spare capacity.
However, when combined with increased traffic and the existing software defect, the reduced capacity contributed to the failure.
How did the outage spread across Microsoft 365 services?
The disruption escalated when the affected servers started becoming unresponsive.
Microsoft’s traffic management systems are designed to automatically redirect requests from unavailable servers to healthy ones.
This normally helps maintain service availability during hardware, software or network failures.
But in this instance, the mechanism made the situation worse.
As servers stopped responding, incoming traffic was redirected to the remaining operational servers. This increased their workload and made them more likely to encounter the same software defect.
The process repeated across the network, triggering a cascading failure.
In less than two hours, the previously undetected defect had affected servers across Microsoft’s networking gateway infrastructure.
At its worst, more than 80% of the servers were unresponsive.
The failure also spread across different regions because the usual isolation mechanisms between network partitions did not prevent the cascading disruption.
Microsoft said a separate issue prevented the normal bulkheads between different capacity partitions from containing the failure.
These partitions, known as capacity units, are designed to isolate infrastructure and prevent problems in one section from spreading to others.
Why were some Microsoft services restored earlier than others?
The outage did not affect every Microsoft 365 service in the same way.
Services such as Teams, OneDrive, SharePoint and Office applications were able to fail over from the affected networking gateway infrastructure.
This helped Microsoft restore these services earlier.
Microsoft said Teams chats and meetings continued to function without issues for most of the outage.
However, services such as Exchange Online could not fail over from the affected gateway in the same way.
As a result, some services experienced longer disruptions than others.
Why did Microsoft take so long to restore services?
Microsoft said recovery was more complicated than usual because its automated restoration systems were not designed to handle such a widespread failure.
As servers came back online, they were immediately hit by accumulated requests waiting to be processed.
This caused some of them to become overwhelmed and unresponsive again.
Another challenge was that the underlying defect had been introduced months earlier.
Since the change had already been deployed globally, Microsoft’s usual rollback mechanisms, gradual deployment controls and feature flags could not quickly reverse it.
Engineers eventually identified the software defect and developed an update that restored the code to its previously stable state.
To implement the fix, Microsoft took groups of servers offline, applied the update and restarted them before returning them to service.
The process was carried out across large portions of the infrastructure, sometimes involving more than 50% of a region at a time.
This gradual restoration helped the company bring services back online without immediately overwhelming the recovering servers.
What happens next?
Microsoft said its final Post-Incident Review (PIR) would provide additional details later in the week.
The review is expected to offer further insight into the software defect, the failure of infrastructure isolation mechanisms and the challenges faced during recovery.
Microsoft 365 outage hits users across 74 countries
Microsoft 365 services, including Outlook, Teams and AI assistant Copilot, suffered a global disruption on Thursday, affecting users across more than 70 countries and territories.
Downdetector, an online outage tracker, received reports from 74 countries and territories within an hour of problems emerging around midday in the US. The disruption affected governments, courts, healthcare services, businesses, airlines and banks, with several governments reporting operational impacts.
Microsoft, headquartered in Redmond, Washington, apologised for the hours-long disruption on Friday, clarifying that the outage was “not related to cyber” or artificial intelligence agents.
The exact number of affected users remains unclear. Microsoft says more than 430 million people use its Microsoft 365 apps, while it reported over 450 million paid commercial subscriptions to its 365 services in January.