A major Microsoft 365 outage that began on November 12 left millions of users worldwide struggling to access Outlook, Microsoft Teams and the Copilot AI assistant. The disruption continued well into Friday, November 13, more than 12 hours after it started.
The problems began around 16:17 UTC, when Exchange Online, the service behind Outlook, faced widespread mailbox access failures. Users on web, desktop and mobile could not open inboxes, and email delivery was delayed. Within about an hour, reports came from 74 countries and territories. Downdetector recorded peaks of more than 130,000 Outlook complaints.
Microsoft said unexpected server failures caused the issue. Many servers failed automated health checks, which cut available capacity and put heavy load on the remaining infrastructure. Engineers found that a process responsible for setting up secure connections became overwhelmed. The company and its executives said there was no evidence of a cyberattack or malicious activity. They also said the cause was the sudden loss of servers, not a broad traffic spike.
The disruption spread beyond email. Teams users faced problems with calendar access, meeting scheduling and some chat features, though core messaging and online meetings mostly kept working. Microsoft 365 Copilot users could not load chat or reach work data such as emails, calendars and files. Other affected services included:
- SharePoint Online
- Defender XDR
- Intune
- Power BI
- Various admin and productivity tools
The outage also affected real-world operations. Banks, government offices, courts and airlines reported disruptions. City governments in the United States reported email outages, and financial institutions in hubs such as Hong Kong saw workflow interruptions. Some companies postponed planned events, and authorities warned of phishing scams that could exploit the confusion. Microsoft 365 connectors used by OpenAI and Anthropic were also disrupted, limiting those AI assistants’ access to user data.
Microsoft tried several fixes, including restarting servers, rerouting traffic, using alternate health-check methods and adding capacity with virtual machines. It saw limited recovery in some regions, but acknowledged that fixes were “not taking as expected.” Restored servers often became unresponsive again once traffic returned. As of Friday morning, sustained recovery had not been achieved. Microsoft said it is pursuing multiple workstreams to stabilize its infrastructure and is posting updates on its status page and social channels.
