Skip to content
RecortNews
Economy

Microsoft declares its global outage over after some 37 hours and blames an earlier network capacity upgrade

Outlook, Teams and Copilot failed across at least 74 countries from November 12, and Microsoft marked the incident remediated at 3:51 a.m. UK time on November 14

Building 92 on Microsoft's headquarters campus in Redmond, Washington, photographed in 2016.
Building 92 on Microsoft's headquarters campus in Redmond, Washington, photographed in 2016.Photo: Coolcaesar/Wikimedia Commons, CC BY-SA 4.0

Microsoft said it had fully restored Microsoft 365 on November 14, closing an outage that had begun a day and a half earlier and had left mailboxes unreachable across dozens of countries.

The company's own incident log puts the first symptoms at 2:45 p.m. UK time on November 12, when Teams chats began failing for users in the Americas. Mailbox access went next, at 4:17 p.m., under Exchange Online incident EX1550543. By 6:26 p.m. Microsoft had opened a second, wider incident, MO1550821, covering every service that depends on Exchange Online. The company marked that incident remediated at 3:51 a.m. on November 14, roughly 37 hours after the first reports, and its public status page showed all products operational at 6:41 a.m.

What stopped working

The failure spread through everything built on top of corporate email. Outlook on the web, desktop and mobile; Teams calendar scheduling and meeting invitations; external file sharing in SharePoint Online; Microsoft 365 Copilot Chat; Defender for Office 365 and Purview; the Graph mail, calendar and contacts endpoints; Planner, To Do and Intune. Consumer accounts were hit as well, including Outlook.com and OneDrive.

Reports on Downdetector peaked at about 130,000 for Outlook alone late on November 12. Microsoft first described the problem as affecting the Americas and then conceded it "may affect Microsoft 365 service connectivity for users in any region"; impacts were reported in at least 74 countries.

A day and a half of workarounds

The disruption landed hardest on institutions that run on email and have no second channel. In the United Kingdom, NHS staff reverted to paper records and pharmacies warned of prescription delays, and a judge in Manchester sent jurors home. A sheriff's office in Colorado asked people to post bond in person. Some Japanese companies went back to the fax machine. Email support lines at the Indian carriers IndiGo and Akasa Air failed, and Air India rerouted its mail. OpenAI and Anthropic both reported problems with their Microsoft 365 connectors.

The cause, as far as Microsoft has said

Microsoft's explanation shifted over the two days. Early updates pointed at the automated health-check system that pulled failing servers out of service, shifting their load onto the remaining ones; overnight the company said a process that sets up secure connections on the affected servers had been overwhelmed and had stopped answering requests. At 5:57 p.m. on November 13 it gave its clearest account yet, attributing the incident to "an issue during an earlier network capacity upgrade that surfaced following a recent network change," and apologised, saying it was "deeply sorry."

The company says there is no indication of malicious activity, and no artificial-intelligence system involved. Recovery was deliberately slow: engineers paused all deployments and configuration changes, throttled traffic to the affected infrastructure, reverted a recent network configuration, patched a component and added capacity, region by region, with Asia-Pacific first, then Europe and North America. "Given the scale of demand, we restored services gradually, validating stability at each stage," Microsoft said.

One side effect outlasted the outage itself. With external mail servers retrying delivery into a system that was not accepting it, Microsoft extended its email timeout window to 48 hours so that queued messages would not be bounced, and the backlog took hours more to drain. Calendar features that depend on Exchange Online stayed unreliable longest. Microsoft shares traded roughly flat through November 13, and the full post-incident report has not yet been published.

Reporting from Microsoft's service status updates as compiled by Azure Blog UK, and Outlook India.