Microsoft outage cuts off Outlook, Teams and Copilot users worldwide for hours after servers go offline

What happened

Microsoft’s cloud services suffered a major outage on Thursday. Wired reported that at about 10:40 am Eastern Time Microsoft began investigating problems with Teams in North America, and by just before 1 pm it was aware of problems reaching Exchange Online mailboxes in the Americas. The issue was later said to affect “any region”, TechCrunch reported.

Besides Outlook email, Microsoft listed Teams, SharePoint, its Copilot AI assistant and the Defender XDR security system as affected, and consumers could not reach OneDrive or Office on the web, according to TechCrunch. Reports on the tracking site Downdetector peaked at nearly 130,000 shortly after 1 pm ET, Al Jazeera and the Guardian reported. France 24 said users in 74 countries and territories were hit, and the South China Morning Post reported that bankers in Hong Kong, including at Goldman Sachs and Nomura, had their work disrupted on Friday morning.

The cause

Microsoft first said part of its infrastructure was under “elevated load”, but later blamed “a large number of servers being unexpectedly unavailable” after failing automated health checks. Wired reported that the company said the problem was servers going offline, not a rise in traffic.

“We've paused all deployments. We're moving additional traffic off the affected infrastructure, and we're restarting affected servers to return them to service.”

Microsoft, Wired

A Microsoft spokesperson told the Guardian the company had “no indication of a cyber-attack”. Microsoft said it was also assessing whether reverting a small number of recent configuration updates could speed up recovery.

A slow recovery

By about 6 pm ET, The Verge reported, Microsoft said it had begun returning some of the affected servers to service and had seen limited recovery in some regions, but there was still no time for a full fix. Five hours after the peak, Downdetector reports had fallen to about 13,000, the Guardian said.

The outage also broke Microsoft connectors used by AI services, including OpenAI’s. Mark Gregory, an associate professor at RMIT University, told the Guardian that when part of a server pool is lost, all users crowd onto the remaining servers, which then overload and fail too, a cascading problem.

Microsoft’s last major incident was in September, when a days-long outage hit Microsoft 365; the company blamed a core authentication configuration issue, TechCrunch noted.