Microsoft outage prolongs disruption to email and workplace services

Microsoft struggled to restore Outlook, Teams and Copilot services on Thursday night after a widespread outage left customers unable to access essential workplace software for more than nine hours.
The disruption extended into Friday in parts of Asia, despite efforts to restart failed servers and redirect traffic across the company’s global infrastructure. Microsoft acknowledged limited recovery in some regions but did not provide a timetable for full restoration.
The company said numerous servers had unexpectedly become unavailable, placing excessive pressure on remaining systems. Engineers discovered that some machines brought back online stopped responding again when customer traffic returned, complicating attempts to stabilise services.
Microsoft initially investigated elevated infrastructure loads across several regions before determining that the problem was not caused by a general surge in demand. The company also examined configuration changes as possible contributors, without confirming a definitive underlying cause.
The failures affected Exchange Online, which supports business email, as well as Microsoft Teams, SharePoint Online, Microsoft 365 Copilot and Microsoft Defender XDR. Other services, including Planner, To Do and elements of Microsoft’s security and administration platforms, were also identified as potentially affected.
Customers encountered errors opening Outlook mailboxes, sending and receiving messages, accessing calendars and connecting to collaboration tools. The scale of the interruption complicated routine operations for organisations that depend on Microsoft 365 to communicate and share documents.
Microsoft said it had suspended deployments and configuration changes while technicians worked to contain the disruption. It redistributed connection requests across its worldwide network, shifted traffic away from struggling equipment and restarted servers to recover lost capacity.
An alternative health-check procedure successfully returned a subset of affected machines to service, prompting engineers to apply the approach more widely. Microsoft also prepared automated measures to identify and restart unresponsive servers, although the company warned that recovery could be uneven as applications reconnected.
“We are aggressively pursuing all pathways to reduce impact,” Microsoft said during its response, as initial recovery measures failed to deliver the expected improvement.
Downdetector, which aggregates complaints submitted by users, registered almost 130,000 reports involving Outlook shortly after 1pm Eastern time on Thursday. Complaints also rose for Microsoft 365, Teams and Copilot, although the figures represented individual submissions rather than a verified count of affected customers.
The volume of complaints subsequently declined, but falling reports did not establish that every service had recovered. Microsoft continued to warn that users in any geographical region could encounter connection failures.
The incident had consequences beyond Microsoft’s own applications. OpenAI and Anthropic identified problems involving connections to Microsoft 365, disrupting some features that allow their artificial intelligence services to interact with workplace email and other information.
Government and education systems also experienced difficulties. Japan’s government acknowledged disruption affecting official information technology infrastructure, while university technology teams in the United States alerted staff and students to intermittent access problems.
Northeastern University advised its community that Outlook email and calendars, Teams messaging and meetings, SharePoint, OneDrive and Copilot Chat could be affected. It cautioned that services might return at different times for different users as Microsoft restored capacity.
The broader incident followed a separate Teams messaging problem earlier on Thursday affecting users in North and South America. Microsoft said that interruption, which caused delayed or failed messages and difficulties retrieving conversations, had been resolved. It did not establish whether the two incidents were connected.
The main Microsoft 365 incident was logged under advisory MO1550821, while an Exchange Online connectivity problem was tracked separately as EX1550543. Those identifiers enabled enterprise administrators to follow technical updates even when normal access to Microsoft’s management portal was impaired.
Microsoft’s public service-status page directs business and enterprise customers to detailed information through the administration centre. That distinction meant some organisations relied on their own technology departments for updates about the particular services and accounts affected.
The incident also exposed the practical difficulty of recovering shared cloud systems when dependent services compete for reduced computing capacity. Restoring individual servers was insufficient if those machines became unresponsive again under normal customer demand.
The company said its restoration work included reviewing a small number of configuration updates and checking whether reversing them could accelerate recovery. Engineers continued monitoring the stability of servers returned to operation while shifting connections to alternative routes.