Microsoft said on Friday that its services were recovering after a global outage disrupted email, workplace communications and public-sector operations for more than a day, although it continued monitoring delayed messages and intermittent access problems.
Microsoft's incident record places the start of the Exchange Online disruption at 16:17 GMT on Thursday, November 12.
Downdetector received outage reports from 74 countries and territories within about an hour, according to Al Jazeera, but Microsoft has not disclosed how many users were affected.
Outlook, Teams, Copilot and other Microsoft 365 applications experienced problems, with the effects varying by service and region.
Microsoft said the fault originated during an earlier network-capacity upgrade and emerged after a subsequent network change.
The company said it had addressed the underlying technical issue and found no link to cyberattacks or artificial-intelligence agents.
Earlier Microsoft service notices showed that servers becoming unexpectedly unavailable transferred workloads onto remaining infrastructure, rather than an overall rise in demand causing the overload.
By 22:00 GMT on Friday, Microsoft reported sustained recovery but kept its service incident open while checking that backlogged emails were processed.
The disruption affected institutions as well as individual users.
In Britain, NHS England employees turned to paper records, although the UK government's cyber minister said emergency calls and frontline NHS 111 services had not been disrupted.
Canada's VIA Rail warned passengers that emailed travel information might not arrive.
In the United States, the House of Representatives and some local governments experienced problems, according to Reuters reporting cited by Al Jazeera.
Microsoft's administrative notices provided a distinction within the reported service failures: most core Teams chats and calls remained functional, while linked calendar and meeting-scheduling features experienced problems.
The company also extended its email timeout window to 48 hours to protect delayed messages.
It cautioned that restored mailbox access could precede normal email delivery and said a detailed post-incident report would follow.