Problems began in North America on Thursday afternoon before spreading across Microsoft’s international network. Businesses lost access to email, communications and other cloud services, while the disruption also reached artificial intelligence platforms operated by OpenAI and Anthropic.
A major Microsoft outage disrupted businesses and individual users around the world on Thursday, 12 November, leaving people unable to access email, workplace communications and a growing range of cloud-based services.
The disruption began with a Microsoft Teams failure in the Americas at 2:45 p.m. UK time, followed by a larger Exchange Online incident at 4:17 p.m. Within hours, Microsoft was warning that its network problems could affect customers in any region.
By early evening, nearly 130,000 reports of problems accessing Outlook had been registered with outage-monitoring service Downdetector. The number of complaints subsequently declined, but significant disruption continued into the night.
Microsoft’s own service-health records identified problems affecting far more than Outlook and Teams. SharePoint, Microsoft 365 Copilot, security monitoring, device management, document services and other business systems were among the services that could be affected.
The failure also reached third-party artificial intelligence services that depend on Microsoft’s infrastructure, including integrations operated by OpenAI and Anthropic.
Microsoft began returning servers to operation late on Thursday, but had not declared the wider incident resolved at the time of its 10:58 p.m. UK update.
The Outage Began With Teams in North America
Microsoft’s incident records establish that the first major disruption of the afternoon began at 2:45 p.m. UK time.
The initial incident affected Teams messaging services hosted in, or routed through, North and South America.
Users experienced problems sending chat messages, creating conversations and retrieving existing messages.
Microsoft identified excessive resource consumption within the underlying service infrastructure as the preliminary cause of that incident.
Automated recovery measures were introduced, and the company reported that the initial Teams problem had been resolved by 3:53 p.m.
However, a separate and considerably wider incident was already developing.
Microsoft’s Exchange Online records place the beginning of the email disruption at 4:17 p.m. UK time.
Exchange Online is the cloud infrastructure supporting business email, calendars and mailbox access for organisations using Microsoft 365.
Problems arose in the infrastructure responsible for receiving and processing requests to connect to those mailboxes.
By late afternoon, customers were experiencing failures when attempting to open Outlook, retrieve messages or access their email through applications connected to Exchange Online.
Microsoft subsequently opened a broader incident covering multiple Microsoft 365 products at 6:26 p.m.
The company initially concentrated its public announcements on the Americas but soon confirmed that connectivity problems could affect customers elsewhere.
Reports Climbed Across the United States
The most substantial publicly reported disruption occurred in the United States, where the outage coincided with the working day.
Downdetector recorded a rapid increase in complaints concerning Outlook, including reports that users could not send or receive messages and were unable to open their inboxes.
Shortly after 1 p.m. Eastern Time, equivalent to 6 p.m. in Scotland, complaints approached 130,000.
The figure later rose to approximately 138,000 before declining.
At 2:07 p.m. Eastern Time, the monitoring service recorded more than 97,000 reports concerning Outlook, over 38,000 reports involving Microsoft 365 and approximately 4,500 concerning Teams.
These figures represent reports submitted to an outage-monitoring service, not a verified count of individual customers affected.
One person may submit more than one report, while many customers experiencing problems may never report them.
Businesses can also have large numbers of employees affected by a single interruption.
Consequently, Downdetector figures provide evidence of the scale and timing of complaints but cannot establish the total number of people or organisations affected.
By 6 p.m. Eastern Time, reports concerning Outlook had fallen to approximately 34,500.
The decline suggested that some customers were regaining access, although it did not establish that Microsoft had fully restored its services.
Problems Reported Beyond North America
Microsoft confirmed that the affected infrastructure was experiencing elevated demand across multiple regions.
The United States was the principal source of the largest publicly reported complaint figures, but the disruption was not confined to American customers.
In the United Kingdom, outage monitoring showed a rise in reports involving Microsoft services during Thursday evening.
These reports included problems with Outlook and other Microsoft 365 applications.
Microsoft did not publish a separate total for Scotland, and there is insufficient evidence to establish how many Scottish businesses or public bodies were affected.
Nevertheless, organisations in Scotland using Microsoft’s cloud infrastructure were exposed to the same potential service failures as other customers connected through the affected systems.
Canadian users also reported problems, with the disruption extending across North American business networks.
In Australia, local outage monitoring recorded more than 1,500 reports, with complaints rising from shortly before 5 a.m. Australian Eastern Daylight Time on Friday, 13 November.
The international timing meant that the outage occurred during different parts of the working day.
For North American businesses, it interrupted Thursday’s daytime operations. In Britain and Europe, the problems emerged during the afternoon and evening. In Australia, they appeared during the early hours of Friday morning.
Microsoft’s service announcements allowed for customers in any region to be affected, but the company had not released a comprehensive country-by-country account of confirmed outages.
The available evidence therefore supports describing the incident as international rather than claiming that every Microsoft region or customer experienced a complete service failure.
Which Microsoft Products and Services Were Affected?
The initial reports focused on Outlook and Teams, but Microsoft’s internal service-health notices identified a much broader range of affected or potentially affected products.
Exchange Online and Outlook experienced the most prominent disruption. Users encountered errors accessing mailboxes through Outlook on the web, desktop applications, mobile connections and other systems using Exchange services. Microsoft’s incident notice specifically identified MAPI, Exchange ActiveSync and REST-based connections.
Microsoft Teams experienced messaging failures during the earlier Americas incident and was subsequently included among the services potentially affected by the wider Microsoft 365 disruption.
SharePoint Online, used by organisations to store and manage documents, was also listed. Problems connecting to SharePoint can affect access to shared records, internal websites and collaborative working environments.
Microsoft 365 Copilot, the company’s artificial intelligence assistant integrated into workplace applications, was among the affected services.
Microsoft Planner and Microsoft To Do, used for task management and work organisation, were included in the broader incident.
Microsoft Defender XDR and Microsoft Defender for Office 365, which provide security monitoring and protection against threats affecting computers, accounts and email, were also named.
Microsoft Intune, used by organisations to manage computers, mobile devices, applications and security policies, was listed among the potentially affected services.
Microsoft Purview, Microsoft’s information governance and compliance platform, experienced problems affecting several functions. These included authorisation, automatic labelling, data loss prevention, communication compliance and data classification.
Power BI users could encounter intermittent failures when exporting reports containing Microsoft Information Protection sensitivity labels to PDF.
Microsoft 365 Apps Admin Center services, including Cloud Update, Cloud Policy and Inventory, were affected or at risk of disruption.
Universal Print, Microsoft’s cloud printing service, could produce errors when users submitted print jobs or attempted administrative operations.
Windows 365 customers using a particular Microsoft Purview Customer Key configuration could encounter problems connecting to their cloud-based computers.
These were not necessarily complete outages of every product.
Microsoft’s notices identified services where users might encounter failures, with some effects limited to particular functions, configurations or customer environments.
Separate reports also identified difficulties accessing Word, Excel, PowerPoint and OneDrive.
Microsoft was dealing with an additional Word for the web incident on Thursday involving documents failing to open or remaining on the loading screen. Its records attributed that problem to a browser compatibility issue introduced by a configuration change. The company had not established that it shared the cause of the wider Exchange incident.
Artificial Intelligence Services Were Also Disrupted
The consequences extended beyond Microsoft’s own applications.
OpenAI confirmed problems affecting its Microsoft 365 connectors in ChatGPT.
These integrations allow authorised users to connect Microsoft services to ChatGPT so that information from supported accounts can be retrieved and used within the assistant.
OpenAI’s status notice reported that new connections might fail and that tools relying on existing connections were not working as expected.
The incident concerned Microsoft-connected functionality rather than an established general shutdown of ChatGPT.
Anthropic, which operates the Claude AI assistant, reported similar problems.
Its first public incident notice appeared at 5:52 p.m. UK time.
By 6:34 p.m., Anthropic had identified an upstream provider problem affecting its Microsoft 365 connector.
The company warned that mail, calendar, search and other connected tools could fail, while new connections might not complete.
Later updates reported that its Claude for Outlook integration was also affected, with some users unable to open the add-in or perform mail and calendar actions.
The incidents demonstrate how disruption within one major technology provider can spread into services operated by other companies.
A business using an AI assistant to search its Microsoft documents, retrieve correspondence or manage connected information may lose those functions even when the assistant’s underlying language model remains available.
The greater the number of applications dependent on a shared cloud platform, the wider the potential consequences when essential infrastructure fails.
Why Microsoft’s Servers Went Offline
Microsoft initially reported that parts of its infrastructure were experiencing unusually high loads.
Further investigation showed that the problem was not simply an increase in customer demand.
A large number of servers had unexpectedly become unavailable.
This forced additional traffic onto the remaining working servers, increasing demand on systems that were still operating.
Microsoft subsequently identified failures involving automated health checks.
These checks are intended to determine whether a server is healthy enough to continue handling requests.
In this case, some servers were failing the checks and being removed from service.
By 7:43 p.m. UK time, the company was testing a configuration change intended to bypass a suspected component involved in processing those checks.
Engineers also began restarting affected servers.
At approximately 8:56 p.m., Microsoft reported that an alternative health-check method had allowed a subset of servers to return to service.
However, later attempts to restore additional infrastructure encountered further difficulties.
At 10 p.m., Microsoft stated that its earlier actions had not provided the expected improvement.
The company’s investigation indicated that a core process on affected servers was becoming overwhelmed, leaving them unable to respond.
Engineers then began isolating portions of the infrastructure, restarting components and gradually reintroducing traffic while monitoring stability.
They also examined recent configuration changes as possible contributors.
Microsoft had not established a definitive underlying cause in the available incident reports.
There was no verified evidence that the disruption resulted from a cyberattack.
Recovery Efforts Continued Into Thursday Night
Microsoft’s efforts to restore services involved several measures operating simultaneously.
It paused software deployments and configuration changes while investigating whether recent changes had contributed to the failures.
Engineers redirected connections away from affected systems, using alternative infrastructure where capacity remained available.
The company restarted unresponsive servers and introduced modified health checks intended to prevent otherwise usable machines from being removed from service.
It also explored ways to isolate traffic between regions to reduce the possibility of disruption spreading through shared systems.
By 10:58 p.m. UK time, Microsoft reported that a proportion of the affected servers had returned to operation and remained stable.
Some regional recovery was becoming visible.
The company was continuing to restore capacity and implement traffic-management measures, allowing customers to reconnect gradually as systems became available.
However, the wider Microsoft 365 incident was still classified as a service degradation.
There was no confirmed time for full restoration.
Businesses were therefore advised through Microsoft’s service-health channels to continue monitoring the incident rather than assume that falling complaint numbers meant every service was functioning normally.
Businesses Beyond Microsoft Reported Consequences
The interruption affected organisations that depend on Microsoft’s cloud systems for daily operations.
Email failures can prevent staff from communicating with customers, suppliers and colleagues.
Problems with SharePoint and Microsoft 365 can interrupt access to documents and records stored in the cloud.
Failures involving device management, security monitoring and compliance systems may also interfere with the administrative work needed to operate larger organisations.
The disruption had at least one publicly documented consequence for corporate reporting.
American electric vehicle manufacturer Workhorse Group postponed its scheduled third-quarter financial results announcement from 12 November to 16 November, citing disruption caused by the Microsoft outage.
The problems affecting OpenAI and Anthropic’s Microsoft integrations provided further evidence of consequences beyond Microsoft’s own customer-facing applications.
There was no reliable worldwide estimate of the total financial cost.
Nor had Microsoft published a complete list of affected organisations, industries or public services.
It would therefore be premature to attribute unrelated hospital, transport, banking or government technology failures to this incident without confirmation from the organisations concerned.
The Wider Risk of Dependence on Cloud Services
The disruption highlights the extent to which modern businesses depend on infrastructure operated by a relatively small number of technology companies.
Microsoft 365 is used for email, shared documents, workplace messaging, identity-related functions, security administration and records management.
These systems may appear to users as separate products, but they can depend on common network services, underlying infrastructure and connections between data centres.
When a shared component fails, several apparently unrelated applications can become difficult or impossible to use.
The effects can also reach third-party companies whose products rely on Microsoft’s services.
Thursday’s incident illustrated this through the interruption of AI connectors operated by OpenAI and Anthropic.
It also raised questions about the resilience of business systems that require continuous internet access or cloud authentication.
Locally stored documents may remain accessible during an outage, but applications relying on remote services can lose functionality.
Similarly, backup internet connections do not necessarily restore access when the cloud provider itself is experiencing a failure.
Organisations seeking stronger resilience may therefore need to consider locally available copies of critical information, alternative communications arrangements and processes that can operate temporarily without the principal cloud platform.
Those precautions do not eliminate cloud outages, but they can reduce the risk that an external service failure prevents an organisation from carrying out essential work.
Microsoft Had Not Established When Normal Service Would Return
The available evidence shows that Thursday’s disruption was not a single isolated failure of Outlook.
It involved a sequence of incidents beginning with Teams in the Americas, followed by Exchange Online connectivity problems and a much wider Microsoft 365 service degradation.
By evening, Microsoft’s list of affected systems included communications, collaboration, artificial intelligence, cybersecurity, device management and business administration.
Customers in the United States, United Kingdom, Canada and Australia reported problems, while Microsoft warned that the infrastructure failures could affect users in other regions.
The company had begun recovering servers late on Thursday, but its published incident records continued to show service degradation.
The final cause, number of affected customers, financial consequences and complete recovery time had not yet been established.
For organisations relying on Microsoft’s services, the immediate issue remained whether access to essential systems could be restored and sustained.
The longer-term question is how many businesses and public institutions can continue operating when a major external technology provider becomes unavailable for several hours.
Sources
Microsoft 365 Service Health — Incidents EX1550543, MO1550821 and TM1550488. Microsoft, 12 November 2026.
https://status.cloud.microsoft/
Microsoft 365 Incident Reports for PennO365. University of Pennsylvania Information Systems and Computing, reproducing Microsoft service-health reports, 12 November 2026.
https://mailservices.isc.upenn.edu/computing/email/penno365/alerts/ms-incidents.html
Elevated Errors Affecting the Microsoft 365 Connector. OpenAI Status, 12 November 2026.
https://status.openai.com/incidents/01M78GAT3AK9P9W8Z230KHFVE8
Elevated Errors Affecting the Microsoft 365 Connector. Anthropic, Claude Status, 12 November 2026.
https://status.claude.com/
Microsoft Outage Hits Tens of Thousands of Users. Al Jazeera Staff and Reuters, Al Jazeera, 12 November 2026.
https://www.aljazeera.com/news/2026/11/12/microsoft-outage-hits-tens-of-thousands-of-users
Microsoft Outage Drags on for Hours, Disrupting Outlook and Teams for Thousands. Reuters, 12 November 2026.
https://www.reuters.com/technology/microsoft-365-outlook-down-thousands-us-users-downdetector-shows-2026-11-12/
Microsoft Services Down in Global Outage. Nine News Australia, 13 November 2026, Australian local date.
https://www.nine.com.au/australia-news/microsoft-services-down-in-global-outage-20261113-p61dup.html
Microsoft 365 Outage Continues Globally. Global News Canada, 12 November 2026.
https://globalnews.ca/news/12148204/microsoft-outage-2026/
