On Thursday 12 November, Microsoft experienced a major outage that disrupted access to several of its major services.
The outage affected systems across over 70 countries: widespread reporting became apparent during the evening of Thursday 12 November, GMT.
Among the services hit were Outlook, Teams and Copilot. Over 430 million people utilise these products and at least 51 million people reported a loss of productivity that impeded economic activity.
Numerous groups of people and organisations were affected, from governments to businesses to UK universities. King’s College London is also heavily reliant on Microsoft Outlook and Teams for work and communications.
Healthcare providers such as GP surgeries in Suffolk reported disruptions due to being unable to access emails confirming prescriptions.
Meanwhile in Japan, workers returned to fax machines, “going full analog” in the wake of the outage.
The technology giant has apologised for the outage on X, reporting that it was “not related to cyber or AI/agents” and was introduced during an “earlier network capacity upgrade”.
What caused the outage?
The outage lasted for nearly 24 hours, with stable restoration completed by Saturday morning. According to a Preliminary Post Incident Report published by Microsoft, this event was primarily caused by three factors that collectively contributed to an outage of the networking gateway serving as the foundation for multiple Microsoft 365 services.
In the case of software such as Outlook, emails are not solely stored on the devices used to access them. Instead, they are primarily stored on remote servers – powerful computers located in Microsoft data centres.
These servers are used to handle emails, Teams calls and millions of operations simultaneously. A primary cause of the outage was a latent defect that overwhelmed Windows servers, causing them to become unresponsive under specific load quantities.
This was a result of lock contention, where numerous threads compete for access to one locked resource, causing delays. This occurred inside an internal Windows process and overloaded the machine’s CPU and memory.
The defect was introduced by a software update deployed in July 2026 for Microsoft 365 gateway networking servers, to increase their capacity. However, its effects did not immediately become evident.
Another cause of the outage was an increase in network traffic, which made the defect more likely to disrupt operations. Microsoft 365 operates two independent network gateway systems, allowing services to switch between them in the event of a disruption, so as to maintain availability.
Thirdly, almost 14% of server capacity for the network had been out of service because of another unrelated issue. This portion of server capacity was undergoing maintenance and health checks at the time.
However, Microsoft still had enough server capacity to function, hence its reduction may have had little to no contribution to the outage.
These three factors contributed towards the network gateway being affected by the bug that had been present since July, causing an increasing backlog of requests for servers to process until they eventually stopped responding.
Once servers returned to service, they were overwhelmed by excessive traffic, causing them to become unresponsive once again. This prolonged the recovery time.
To solve the problem, engineers programmed a fix that restored the previously functioning code path for Windows servers. They subsequently applied this to all of the affected servers, in groups.
The prominence of the outage’s affect raises a question: is the world too dependent on a few major technology corporations?
The provision of digital infrastructure and services by “tech giants” such as Microsoft and Google is incredibly convenient for everyday operations in domains such as business operations. This maximises efficiency and accessibility, as well as optimising costs.
However, it could pose risks in such situations. The occasional unavailability of essential digital services will inevitably cause disruptions such as economic loss.
Therefore, firms should contemplate planning ahead, in the case of a software outage. These could include alternative methods of accessing crucial information, in addition to continuing the utilisation of their current software system.
Microsoft started successfully running again on Saturday morning. But the prevalence of the impact of the outage highlights how dependent the world has become on a small number of technology corporations – when one experiences a significant issue, millions of people across multiple industries can be affected.






