Microsoft 365 outage: cause, affected services and recovery, Nov. 12–13

Latest verified update: 13 November 2026, 17:57 UTC (Microsoft advisory). The incident record was still marked "service degradation" when checked at 19:50 UTC. This article is a reporting snapshot, not a live status dashboard.

Microsoft says its worldwide Microsoft 365 disruption is broadly recovering but not yet fully restored. The company says an issue introduced during an earlier network-capacity upgrade surfaced after a more recent network change. Microsoft says the underlying issue has been addressed, while restoration continues for remaining customers and regions. It explicitly says the incident was not related to cyber activity or AI/agents. A full post-incident report has not yet been published.

The disruption began on Thursday, 12 November, and continued into Friday, 13 November. Exchange Online mailbox connectivity was the central affected service, with knock-on failures in Microsoft 365 features that depend on email, calendars or related Microsoft services. Microsoft tracked the incident under EX1550543 (Exchange Online, beginning 16:17 UTC on November 12) and MO1550821 (broader Microsoft 365, beginning 18:26 UTC). Microsoft's administrator advisories record both IDs and their updates.

What is the outage status?

In its 17:57 UTC update on November 13, Microsoft reported broad recovery and service stability. It said it had corrected the underlying issue and was restoring access gradually to remaining users and regions, validating server stability as demand returned. Both incident records were still classified as service degradation in the administrator feed available at 19:50 UTC; Microsoft had not yet declared a complete restoration.

Earlier on November 13, Microsoft reported that email delivery queues had been substantially reduced and were approaching normal levels. It still warned that some users could see intermittent mailbox access problems or delayed delivery. The company had extended its email timeout window to 48 hours to help preserve delayed messages while queues were processed; this was a mitigation, not a guarantee that every individual message had been delivered.

Independent measurements from Cisco ThousandEyes also showed recovery. At 14:00 UTC, its monitoring indicated that Outlook availability had largely returned and most monitored users could connect, while some locations continued to show packet loss and timeouts. That observation does not override Microsoft's later incident classification.

Which Microsoft services were affected?

Microsoft said features dependent on Exchange Online could be interrupted; it did not report every function of every listed product as unavailable.

The impact descriptions above follow Microsoft's MO1550821 notice. They describe services and features that could be affected, not a confirmed percentage of customers in each product.

Timeline: November 12–13, 2026

All times UTC. Microsoft times are from the administrator service-health notices; the ThousandEyes observations are independently sourced.

What caused the Microsoft 365 outage?

Microsoft's latest explanation is more specific than the initial reports of unusually heavy load. At 17:57 UTC November 13, it said a problem introduced during a previous network capacity upgrade became apparent after a subsequent network change. Microsoft says the underlying issue has been addressed.

The incident advisories document how the problem manifested and why recovery was difficult:

• Some affected servers failed health checks or became unresponsive, reducing available service capacity.
• Traffic was redistributed onto fewer healthy servers; Microsoft later clarified that the high load was not simply a rise in overall customer traffic.
• Some restarted servers failed again when traffic returned.
• Engineers tested traffic restrictions, server isolation, an upstream authentication/routing component patch, added capacity and phased restoration.

These are Microsoft's reported incident observations and mitigation steps, not an independently verified, complete technical postmortem. Microsoft has promised a post-incident report setting out root-cause details and follow-up safeguards. As of this reporting snapshot, it had not provided that document.

Microsoft explicitly ruled out cyber activity and AI/agents as causes. There is also no established causal link between this Microsoft 365 incident and separate reports about Azure VM provisioning constraints in European regions. They concern different services and failure scenarios.

How far did disruption extend?

The affected functions went beyond employee email. The Associated Press reported that a court trial in Manchester was interrupted, some UK police forces had problems receiving online crime reports, and institutions in Norway, Greece, Singapore, the Philippines and Japan reported communication or service difficulties. The reported impact differed by organization.

OpenAI and Anthropic also reported problems with integrations connecting to Microsoft 365 tools. That is not evidence that their underlying AI models were unavailable.

Downdetector recorded more than 130,000 US Outlook problem reports near the November 12 peak, according to contemporary Reuters reporting. These are user-submitted reports, not a count of distinct people affected, and do not establish the total global impact. Microsoft has not released a verified total of affected users for this event.

What remains unconfirmed?

Microsoft has identified a network-change sequence and reported broad recovery, but has not yet published a final post-incident report or confirmed full global restoration in the notices cited here. The exact internal failure chain, complete customer-impact figures and detailed prevention measures are therefore not established in this article.

For the latest status, see Microsoft's public service-health page or Health → Service health in the Microsoft 365 admin center, using incident IDs EX1550543 and MO1550821. The university-hosted Microsoft notice archive provides timestamped copies of Microsoft administrator updates.

Quick answers

Is Microsoft 365 fully restored? Not in the last Microsoft advisories checked: the company reported broad recovery but still classified the incidents as service degradation.

Were Teams calls and chat down for everyone? No. Microsoft said most core Teams calls, meetings, group chats and standard channel messages remained functional; Exchange-dependent features were affected.

Were emails lost? Microsoft confirmed delayed delivery and a 48-hour email timeout mitigation. It did not report that every delayed message had been delivered, nor did it establish a universal loss of email.

Was this a cyberattack or caused by AI agents? No. Microsoft explicitly said this was a technical problem unrelated to cyber activity or AI/agents.

Did this prove an Azure-wide infrastructure shortage? No. Microsoft identified a problem in the Microsoft 365 service stack involving network changes. No direct connection to separate Azure provisioning-capacity issues has been established.