A failure affecting Exchange Online infrastructure has rippled across Microsoft 365, Defender, Purview, Intune and Copilot, while OpenAI and Anthropic also reported problems with their Microsoft 365 integrations.
Microsoft is still working to restore services after an infrastructure failure disrupted Exchange Online and a wider range of Microsoft 365 services, including security, identity management, collaboration and AI applications. The outage, which began on Thursday, has persisted into Friday, exposing how a problem in a shared cloud service can affect multiple enterprise workloads.
The disruption initially surfaced as connectivity problems with Exchange Online, leaving users unable to access mailboxes through Outlook and other connection methods. Microsoft subsequently identified problems across a much wider set of services, including Microsoft Defender, Purview, Intune, Teams, SharePoint Online and Copilot for Microsoft 365.
In its latest incident update available on Friday morning, Microsoft said a process on affected servers responsible for establishing secure connections was becoming overwhelmed, causing the servers to stop processing requests. The company was preparing a configuration change to address the behavior and testing it on a subset of servers before broader deployment.
Microsoft was also reverting a recent network configuration change that correlated with the start of the disruption, although it had not established that the change caused the outage.
The company said it was imposing temporary traffic restrictions in certain locations while monitoring whether affected servers could remain stable with a narrower range of connections.
Server failures triggered wider disruption
Microsoft’s incident reports distinguish between two related problems: EX1550543, covering Exchange Online mailbox access, and MO1550821, covering connectivity to multiple Microsoft 365 services.
The Exchange Online incident began at 16:17 UTC on November 12, while the broader Microsoft 365 incident was recorded at 18:26 UTC.
Microsoft initially attributed the elevated infrastructure load to a large number of servers becoming unexpectedly unavailable, rather than an overall increase in user traffic. Their failure shifted additional load onto the remaining servers, creating further connectivity problems.
The company initially suspected that affected servers were failing automated health checks, potentially because of a component involved in processing those checks. It tested an alternative health-check method and attempted to restart affected servers.
However, the recovery effort encountered further complications. Microsoft found that some servers returned to service but became unresponsive again after receiving traffic. It subsequently began rebuilding a subset of servers using an earlier software configuration to determine whether that would prevent the failures.
Microsoft also paused deployments and configuration changes as a precaution, redirected traffic away from affected infrastructure and introduced measures to limit the load on struggling servers.
In its latest update, the company said it was reviewing traffic flows and infrastructure to identify opportunities to improve routing and stabilize affected services. It had not yet published a confirmed root cause or declared the incidents resolved.
Security and compliance workloads also affected
The breadth of the disruption is particularly significant for enterprises because the impact extended beyond email and collaboration.
Microsoft’s incident advisory lists potential failures in Defender for Office 365, including Safe Links and Safe Attachments, protection policies, quarantine access and threat investigation features. Users clicking URLs rewritten by Safe Links could encounter navigation failures because the service was unable to process the links.
Microsoft Defender XDR users could also be unable to load portal pages or solution tabs.
The disruption extended to Microsoft Purview, affecting authorization, data loss prevention, auto-labeling, information protection, communication compliance and data classification. Audit searches, eDiscovery operations and access to protected content could also fail.
Microsoft Intune customers experienced intermittent problems with device enrollment, check-ins and compliance policies. Microsoft Graph requests involving Exchange Online data, including mail, calendar and contacts, could time out, potentially affecting applications and workflows that depend on those APIs.
Teams’ core messaging, channel conversations and online meetings remained largely functional, according to Microsoft’s advisory. However, features dependent on Exchange Online, including calendar access, meeting scheduling, invitations and some chat operations, were affected. SharePoint file access remained functional, although features dependent on Exchange Online could fail.
AI integrations caught in the disruption
The outage also affected services that depend on Microsoft’s infrastructure beyond its own product portfolio.
OpenAI reported an issue with its Microsoft 365 connectors, warning that new connections could fail and tools might not work as expected. Its status page classified the incident as degraded performance and said it was investigating the problem.
Anthropic also identified an issue with an upstream provider affecting its Microsoft 365 connector, according to the company’s outage communication.
The disruption illustrates the dependencies that can emerge when enterprise applications and AI assistants rely on third-party cloud services to access corporate email, calendars and files.
The business impact was not limited to technology services. Electric truck maker Workhorse Group postponed its third-quarter financial results announcement, originally scheduled for November 12, to November 16, citing disruptions caused by the global Microsoft outage.
Microsoft has not provided a definitive restoration timeline. Its incident reports continued to classify both Exchange Online and the broader Microsoft 365 disruption as service degradation in the latest updates reviewed.
