Technology · Cloud Outage
The Microsoft 365 outage that began on Thursday afternoon knocked out Outlook email, Teams calendars and Copilot for organisations in 74 countries, reached into the US Capitol, British police forces and the Japanese government, and was still not fully over on Friday evening — with Microsoft blaming a network change of its own, promising a full post-incident report, and giving no time for a complete fix.
REDMOND, Washington, November 13, 2026 — For the second day in a row, the working world woke up on Friday to an inbox it could not open. A global Microsoft 365 outage that began on Thursday, November 12, continued into Friday, November 13, disrupting Outlook email, Microsoft Teams and the Copilot AI assistant for companies, governments and millions of individual users, according to Microsoft’s own service-health updates and reporting by the BBC and the Associated Press. Microsoft said on Friday that it was seeing “broad recovery and stabilisation” across its services — but its status page was still listing service degradation for Microsoft 365 products late on Friday afternoon in the United States, queued email was still draining, and the company pointedly declined to give an estimated time for full recovery.
The scale is what separates this from the routine wobbles that ripple through cloud services every few weeks. Outage tracker Downdetector recorded reports of Outlook problems from 74 countries, peaking at nearly 140,000 reports at about 18:00 GMT on Thursday — more than 2,000 times the normal level, the BBC reported. Downdetector counts only people who bother to file a report, so the true number of affected users is far larger: Microsoft 365 is the default office software for much of the world’s business and government, and when its email layer fails, the failure is measured not in users but in organisations. Among those caught, according to the BBC and Politico reporting carried by the Associated Press: staff of the US House of Representatives, police forces, schools and doctors’ offices in the United Kingdom, the Singaporean government, and government offices in Japan, where officials switched to text messages as an emergency measure. In some regions, the Associated Press reported, the outage disrupted police emergency calls and court operations.
What actually broke: an Exchange problem that would not stay in its lane
The failure began where a modern office can least afford it: in Exchange Online, the engine underneath Outlook email and calendars. Microsoft’s incident record tells the story in two reference numbers that IT administrators around the world now know by heart — EX1550543 for the Exchange Online failure, opened on Thursday afternoon, and MO1550821 for the wider Microsoft 365 incident declared as the damage spread. Users found they could not connect to their mailboxes through Outlook on the web, the desktop app, mobile sync or the programming interfaces that other software uses to reach email.
From email, the damage travelled along the connections that make Microsoft 365 a single product rather than a box of separate apps. Teams kept its core alive — Microsoft says messaging and online meetings were largely functional — but Teams calendars and meeting scheduling, which lean on Exchange, faltered. Copilot, the AI assistant Microsoft has spent two years weaving into every corner of the suite, was disrupted. SharePoint-connected features, the Microsoft 365 admin centre and a tail of security and administration tools were also affected at points during the incident. It is the defining trait of the modern cloud outage: nothing fails alone, because nothing is built alone.
Microsoft’s explanation: its own network change, not an attack
Microsoft has been consistent on one point from the first hours: this was not a cyberattack. “We will continue to share more information as soon as we have it, and for now I can confirm that there is no indication this is the result of malicious or cyber activity,” Microsoft executive Charles Lamanna wrote on X late on Thursday, apologising for the disruption. By Friday, the company’s account of the cause had sharpened. Computerworld, reviewing Microsoft’s service-health dashboard on Friday, reported the company’s conclusion that the outage was a technical issue “introduced by an earlier network capacity upgrade that surfaced following a recent network change” — in plain language, a change Microsoft made to expand its own network capacity lay dormant until a later change woke it up, in the worst possible way.
The mechanics, as Microsoft described them through the incident, were a cascade. A number of servers became “unexpectedly unavailable”; their traffic landed on the servers still standing, producing what the company called “elevated load” across multiple regions; and a process responsible for establishing secure connections began to be overwhelmed, leaving more servers unable to answer requests. Engineers rolled back a network configuration change that coincided with the start of the problem, and later applied a patch to an upstream authentication and routing component that Microsoft said improved server stability, rolling it out region by region. One protective measure will matter to anyone still missing mail: Microsoft extended the window its servers hold pending messages before giving up, from 24 hours to 48 hours, so delayed email should arrive rather than bounce while the backlogs clear.
The recovery: real progress, carefully worded
Friday’s updates traced a genuine but uneven recovery. Just after 6 a.m. Eastern on Friday, Microsoft was still telling customers that “users may be unable to connect to multiple Microsoft 365 services,” while reporting progress. By mid-morning in the US, the company described “encouraging recovery progress, with service availability and email delivery performance improving across a growing portion of affected Microsoft 365 infrastructure.” By Friday afternoon, Microsoft reported broad recovery and stabilisation, and the Seoul Economic Daily noted the outage had passed the 24-hour mark — among the longest in the company’s history, and long enough for exasperated users to rename the product “MS364” online. Yet the qualifiers never left the page: at 3 p.m. Eastern, Microsoft’s public status page still showed degradation for both business and consumer Microsoft 365 services, and an 8 p.m. London-time update reviewed by Computerworld said engineers were “still draining queued messages and rebalancing traffic,” adding, “We do not have an ETA for full recovery yet.” Microsoft closed the day with an apology — “We are deeply sorry for the disruption this incident has caused you and your business” — and a promise of a full post-incident report once its review is complete.
Why this matters: the world’s office has a single point of failure
It is tempting to file a day of broken email under inconvenience. That undersells what Thursday and Friday demonstrated. Email is not merely a messaging tool; it is the identity layer of modern work — the address to which password resets, contract signatures, court filings, hospital referrals and two-factor codes are sent. When Exchange Online stops, organisations do not just lose a way to talk; they lose a way to prove who they are and to move paper that legally has to move. That is why the affected list reads the way it does: a national legislature, police forces, courts, schools, doctors. The outage did not target the critical machinery of government. It did something more revealing — it showed how much of that machinery quietly runs on one vendor’s office suite.
The concentration question is no longer theoretical. This was, by The Guardian’s count, the third major Microsoft 365 outage of 2026, after failures in January and August. Each one restarts the same argument with fresh evidence. Defenders of the status quo note, correctly, that Microsoft’s scale is also its defence: few organisations on earth could restore a planetary email system as fast as Microsoft’s engineers did, and the historical comparison cuts both ways — the July 2024 CrowdStrike failure, which crashed some 8.5 million Windows machines, was caused by a security vendor’s update, not by Microsoft’s cloud at all. Critics answer that the lesson of both episodes is identical: when a handful of interlocking platforms carry the world’s work, a single bad change — a network upgrade in Redmond’s cloud this week, a faulty sensor file in 2024 — becomes a global event within the hour. The difference this time is that the change was Microsoft’s own, made to add capacity: the system failed, in part, while being made bigger.
There is a second, quieter exposure that this outage priced in real time: artificial intelligence. Copilot is no longer an experiment bolted onto Office; Microsoft has sold it to enterprises as the connective tissue of the working day, summarising the meetings and drafting the emails that this week it could not reach. An AI layer is only as reliable as the plumbing beneath it. Every chief information officer who spent Friday fielding calls about blank inboxes now has a concrete data point for the renewal negotiation — and Google, whose rival Workspace suite stayed up, has a sales story it did not have to write. Our coverage of Google’s push to put its own agent inside Microsoft 365 (Google’s Gemini agent for work) and of Microsoft’s own AI hardware ambitions (Microsoft’s Maia 300 AI chip) shows how hard both companies are fighting for exactly the customers who spent this week offline.
What happens next
Three things now matter more than the outage itself. First, the post-incident report: Microsoft has promised a root-cause account with what it calls defence-in-depth actions, and the credibility of that document will be judged against a simple question — how did a capacity upgrade and a network change, neither of them an attack, combine to take down email on six continents? Second, the tail: mail queues are still draining, and organisations should expect delayed messages to keep surfacing into the weekend even where sign-in already works; service access, as Microsoft itself warned, recovers before mail flow fully normalises. Third, the reckoning inside IT departments: expect a fresh round of continuity planning — secondary email routes, offline contact trees, and hard questions in procurement about what a productivity suite’s bad day does to a police force, a court or a legislature. None of that will move most organisations off Microsoft 365; the switching costs are real, and Friday’s recovery was, in the end, a demonstration of formidable engineering. But resilience has a way of becoming a budget line only after it has been a headline.
For now, the accurate summary is the narrow one Microsoft’s own dashboard supports: the Microsoft 365 outage is receding, not finished — broad recovery across most services, degradation still listed for parts of the suite that depend on Exchange Online, no confirmed time for full restoration, and a root cause the company attributes to its own network changes rather than to any attacker. If your inbox went silent this week, was it a blip your organisation absorbed — or the moment it started planning for the day the cloud goes down? If you find our reporting useful, like this story, follow Signal Post News for continuing coverage of the recovery and the post-incident report, and share your experience in the comments.
Sources
BBC News, November 13, 2026 (Downdetector recorded nearly 140,000 Outlook problem reports from 74 countries at about 18:00 GMT on Thursday, more than 2,000 times normal levels; Microsoft’s account of unexpectedly unavailable servers and elevated load; Charles Lamanna’s statement that there was no indication of malicious or cyber activity; impact on US House of Representatives staff, UK police forces, schools and doctors’ offices, and the Singaporean government). Associated Press, November 13, 2026 (the global outage entering its second day, affecting Outlook, Teams and Copilot; disruption to police emergency calls and court operations in some regions; Microsoft treating the incident with urgency until full resolution; recovery progress on Friday). Computerworld, November 13, 2026 (Microsoft’s service-health dashboard: a technical issue introduced by an earlier network capacity upgrade that surfaced following a recent network change; continued degradation for Exchange-linked services; queued messages still draining with no ETA for full recovery; Microsoft’s apology and promised post-incident report). Reuters reporting on the outage, via Dawn, November 13, 2026. Microsoft 365 service health and status updates, November 12–13, 2026 (incidents EX1550543 and MO1550821; email timeout window extended to 48 hours). The Guardian, November 13, 2026 (the outage as the third major Microsoft 365 failure of 2026, after January and August; Japanese government offices switching to text messages). Photos: Hay Kranen and Coolcaesar, via Wikimedia Commons, as credited in the captions.