Microsoft 365 Outage Drags Past 21 Hours as Services Slowly Come Back

The trouble started around 6pm GMT on Thursday, when Down Detector logged a surge of users reporting problems with Microsoft 365 — the cloud suite spanning Teams, Word, Excel, PowerPoint and Outlook. Recovery only began to show up at 11:20 GMT on Friday, according to data from research firm Infoblox, putting the outage past the 21-hour mark. By Friday morning Microsoft had narrowed most of the trouble to Exchange, the email engine behind Outlook, while Teams meetings and chats were already back.

Microsoft says this was not a cyber-attack. Instead, a configuration update overloaded some servers and pushed them offline, the company indicated. When engineers applied a fix, servers that came back up turned unresponsive again as traffic flooded back. Traffic was then funnelled onto the remaining servers, and Microsoft began limiting what could reach affected locations and rolled some machines back to an earlier software configuration to test for a fix. By midday UK time it reported “encouraging recovery progress in Europe”, improvements in North America and Asia Pacific, and Japan as “stable”, with email queues “significantly reduced and approaching normal levels” — though some users may still hit issues.

The blast radius was enormous. In the UK, police services, the Bank of England, ambulances, the NHS and schools were affected, and the government said it was “working closely” with Microsoft while pointing to “robust contingency plans”. The NHS in England holds a five-year deal to use Microsoft 365. Australia's councils, courts, schools and a spy agency reported problems; government agencies in the Netherlands, Belgium, Germany and France flagged issues, with one French municipality telling residents to phone instead. Japan saw workers reverting to faxes, with shipping firm Sagawa Express unable to take some online reservations and pharmaceutical company Taisho handling orders by phone. In the US, the House of Representatives, a Maine court and cities in Colorado and Florida reported trouble, while Indian airlines IndiGo and Akasa Air said email support was hit and Taiwanese security firms, life insurers and investment trust companies saw fund management disrupted.

Steven Murdoch, a professor of security engineering at University College London, said the root cause is still a mystery. “We don't know what actually went wrong, and that will take weeks to discover,” he said, adding that the engineers are “too busy fixing to understand or report on the cause”. He expects a proper postmortem on an outage this size, but not for a few weeks, and noted such failures are usually “the result of some surprising interaction between systems”.

The outage has sharpened a familiar worry: the world's reliance on a handful of US cloud giants — Amazon, Google and Microsoft. That unease has grown louder as the Trump administration's hostility towards Europe pushes some governments to weigh alternatives to US software. Mark Gregory of Australia's RMIT University said organisations should test updates thoroughly before pushing them live, and keep backup options such as Gmail or the open-source LibreOffice. Forrester analyst Dario Maisto called the incident “the perfect quasi-natural experiment for stress testing an organisation's – and even an entire continent's – digital sovereignty posture”, forcing governments to confront what happens when core tools like Copilot go dark.