Microsoft 365 Services Recover After an Outage of More Than 20 Hours

"We're seeing encouraging recovery progress, with service availability and email delivery performance improving across a growing portion of affected infrastructure," the company said in an update at 10:39 a.m. Eastern time. "We know our customers rely on our services to get important work done every day, and regret the impact these outages had on those who depend on us."

The failure shows how far large organisations now depend on a single provider for email, scheduling and, increasingly, AI assistants. Microsoft has made Copilot central to its Microsoft 365 sales pitch, and Copilot was among the services that went down. The company said it had seen no evidence of malicious activity and would publish a post-incident report on the root cause once service is fully restored.

The Long Recovery

The problems began around 1 p.m. Eastern on Thursday. Microsoft first said a number of servers were "unexpectedly unavailable", which put extra strain on the rest of its infrastructure across several regions, and blamed a surge in traffic and a fault in a component that checks server health. By Thursday afternoon it had found another way to run health checks, which brought some servers back, and it started applying that fix more widely.

It did not hold. At around 10 p.m. Eastern on Thursday, Microsoft said some of its efforts to improve serviceability had not worked as hoped. At 2:45 a.m. on Friday it reported positive results from new traffic optimisation steps. At about 8:30 a.m. it described broader improvements in North America, sustained gains in much of Europe and encouraging signs in Asia Pacific. Engineers took affected systems offline for repair and restored them in stages, putting priority on keeping delayed emails and clearing backlogs. Microsoft extended the retry window for delayed messages to 48 hours.

Some users were still seeing delayed email and patchy connections on Friday morning, Microsoft said. It earlier warned that servers it had restored could become unresponsive again and said it was preparing automated systems to find and restart them.

Hospitals, Fax Machines and Earnings Releases

The effects reached well beyond office email. The UK's National Pharmacy Association warned that some prescriptions could be delayed, and a hospital group in London and Surrey told patients they might wait a long time on the phone. Some Japanese businesses went back to fax machines for a while. Air India rerouted email so it could keep receiving customer messages and said flights were not affected. Richemont's earnings release came out slightly late, and Workhorse Group moved its third-quarter results from November 12 to November 16.

Australia's Home Affairs Minister Tony Burke said Microsoft regarded the incident as somewhat similar to the 2024 CrowdStrike outage, which a faulty software update caused.

Microsoft shares were roughly flat in Friday trading after falling earlier in the session, and cybersecurity stocks including CrowdStrike and Palo Alto Networks gained. The outage began a day after Nvidia's leadership crisis had already unsettled technology investors.

Microsoft has not given a date for full restoration or for its post-incident report. That report will show whether the cause was a capacity problem, a faulty health-monitoring component or a configuration change, and it is likely to shape how enterprise customers think about relying on one provider for email, collaboration and AI tools.