Microsoft 365 outage lasted over 24 hours, disrupting Outlook, Teams and Copilot worldwide.
Microsoft blamed an earlier network capacity upgrade, not any cyberattack or AI fault.
Services were restored gradually; a full post-incident report is still promised.
Microsoft has confirmed that the Microsoft 365 outage that knocked out Outlook, Teams and Copilot for more than a day is fixed. In its final status update, the company said it had completed its efforts to address the underlying issue and restored service.
The disruption reached users in at least 74 countries. Businesses, governments, hospitals and airlines were pushed back onto phone lines, paper records and, in some offices, fax machines.
The company has since said it is “deeply sorry” for the disruption and has maintained that the incident was not caused by a cyberattack or AI.
The Cause: A Network Upgrade Gone Wrong
A Microsoft spokesperson said the incident was introduced by an issue during an earlier network capacity upgrade, which surfaced following a recent network change. The company described it as a technical issue unrelated to cyber threats or AI agents.
That explanation arrived only after a long night of trial and error. Microsoft’s first updates on X blamed “elevated load” across several regions. Hours later, the company reversed that diagnosis, saying traffic had not risen at all. Instead, a number of servers had become unexpectedly unavailable after failing automated health checks, and their workload fell onto the servers still running, straining those as well.
Engineers later traced part of the fault to a core process, responsible for secure connections, that became overwhelmed and stopped processing requests. At one stage, Microsoft admitted its actions had not provided the anticipated relief.
Why Recovery Took Almost A Full Day
The hardest part was keeping servers stable once they were back. Microsoft said restored servers could stop responding again as soon as traffic returned. Engineers worked on several fronts at once: they limited the types of traffic hitting affected infrastructure, rolled some servers back to an earlier configuration, patched an authentication and routing component, and added virtual machines for extra capacity.
Email made the job slower still. Mail servers outside Microsoft kept retrying undelivered messages, so each restored server faced a backlog of queued mail the moment it came online. To ensure no messages sent during the outage were lost, Microsoft extended its email timeout window to 48 hours.
“Given the scale of demand, we restored services gradually, validating stability at each stage,” the company said, adding that although this took longer, it ensured a stable restoration.
Asia Pacific was among the first regions to stabilise on November 13, followed by Europe and North America. By that afternoon (GMT), email queues were significantly reduced and approaching normal levels. Teams chats and meetings had largely returned earlier, though calendar features that depend on Exchange Online stayed patchy for longer. Personal-account services, including Outlook.com, OneDrive, OneNote and Microsoft To Do, were also affected.
A Global Reach: Airlines, Hospitals And Courts Disrupted
Microsoft said in January that it had more than 450 million paid commercial subscriptions to its 365 services, and the breadth of the outage reflected that scale.
In India, email support at IndiGo and Akasa Air was hit, while Air India rerouted its emails to keep receiving customer messages and its flights ran as usual. At one large brokerage, a backup covered Outlook, but client calls on Teams were harder to manage.
In the United Kingdom, NHS staff switched to paper records, pharmacies warned of prescription delays, and a judge in Manchester sent jurors home. In the United States, a sheriff’s office in Colorado had to ask people to post bond in person, because online payments relied on Outlook email notifications, and in Japan, some companies went back to fax.
The technology and industrial sectors were not spared either. Even OpenAI and Anthropic ran into problems with their Microsoft 365 connectors, while electric truck maker Workhorse Group pushed its quarterly results to November 16.
Markets Stay Calm; Loose Ends Remain
Investors were far calmer than users. Microsoft shares traded largely flat on November 13, even as the outage stretched into a second day.
For users, a few loose ends may remain. Emails held in queues carry the time they were delivered rather than sent, so a message stamped November 13 may well have been written on the night of November 12.
What Comes Next
Microsoft has promised a full post-incident report once its review is complete. That report should explain how a capacity upgrade knocked servers offline across regions for more than a day, and what the company is changing so the next network update does not do the same.
