Ongoing Since November 12, 2026
Exchange Online: worldwide outage of mailbox access
Since the evening of 12 November 2026 (European time), many companies worldwide have had no or only intermittent access to their email, calendars and contacts in Microsoft 365. The affected service is Exchange Online, Microsoft's email service: Outlook on the computer, Outlook in the browser and the mail app on the phone often cannot connect, and new emails arrive late. Because many other Microsoft services access the mailboxes, Teams calendars, Copilot and security features are disrupted as well. On the morning of 13 November, Microsoft reports first improvements after a fix; the outage is not resolved yet.
Key facts
- Status: Microsoft reports broad recovery and stable service across many areas. Service is still being restored for remaining customers and regions. The incident is not fully resolved and remains classified as “Service degradation.”
- What is happening: Email, calendar, and contacts in Microsoft 365 may still be affected. Other services may be affected, including admin and security portals, Copilot, Planner, To Do, Universal Print, and some Teams, SharePoint, and Intune features. According to Microsoft, Teams calls, meetings, group chats, and channel messages are not affected. The cause is at Microsoft, not on your device or with your IT department.
- Since when: November 12, 2026, early evening Swiss time; Microsoft’s first public notice was at 18:54 Swiss time.
- Cause according to Microsoft: An issue during an earlier network-capacity upgrade surfaced after a later network change. Microsoft has addressed the underlying issue and is restoring services gradually. According to Microsoft, the outage was not related to a cyberattack or AI features. Microsoft will publish a full report after its review is complete.
- Email: New email may still arrive late. Microsoft extended the time before a delivery failure to 48 hours so delayed messages can be delivered during recovery.
- For IT administrators: Reference numbers in the Microsoft 365 admin center are EX1550543 for Exchange Online and MO1550821 for Microsoft 365.
What does this mean for me?
- Emails, appointments and contacts that were already loaded stay visible in Outlook on your computer. So do not reinstall Outlook or set up the account again: a newly set up Outlook no longer has any stored emails.
- Do not reset your password. The problem is on Microsoft's side; a new password does not change anything.
- New emails arrive late. People who write to you do not need to do anything: the sender's mail server retries delivery automatically.
- Teams meetings, calls and chats mostly work. Calendars and invitations are not reliable; meeting links can be shared in Teams chat.
- Many companies rewrite links in emails for security reasons (Microsoft “Safe Links”). Such links often cannot be opened at the moment. Ask the sender for important links through another channel.
- Handle urgent matters by phone or Teams chat, and take the outage into account for important appointments and deadlines.
Timeline
- First public post on EX1550543, initially for the Americas
- Microsoft pauses all deployments, restarts affected servers and opens MO1550821; users in all regions may be affected
- Cause of the high load: failed servers, not more traffic
- Microsoft tests a configuration change because servers fail automated health checks
- Microsoft assesses reverting recent configuration changes
- The measures bring no relief; an overloaded core process is named as the reason for the capacity loss
- Part of the servers is back in service and stable for now
- Restarted servers fail again as soon as they receive traffic
- Restarting parts of the infrastructure in isolation shows positive results
- Partial recovery in some regions, but not sustained
- Routing services send markedly more requests to the backend servers; Microsoft scales up additional virtual machines
- According to Microsoft, no signs of malicious activity
- A patch for a sign-in and routing component shows positive results and is being expanded
- Gradual improvements; Microsoft extends the time before an email delivery failure to 48 hours
- Further progress in Europe and North America; some users still have intermittent problems
- Email backlog significantly reduced, but intermittent access issues and delays continue
- Microsoft continues to monitor recovery and lists additional affected Microsoft 365 services
- Microsoft reports broad recovery; the issue from an earlier network-capacity upgrade has been addressed
Our own measurement: reachability of Exchange Online
Independently of Microsoft's status posts, a rafaelpfister.ch server checks every 5 minutes whether the public Exchange Online endpoints accept connections. Sign-in (Entra ID), Microsoft Graph and a neutral reference service are measured for comparison. This shows whether only Exchange is affected or the network in between.
This measurement runs permanently and has its own page: Microsoft 365 Monitor →
Loading measurements …
Last 30 minutes
Share of successful connections per endpoint (3 connections per run) and average time to the first response.
History
Each cell covers 15 minutes (24 hours) or one hour (7 days). Hover over or tap a cell to see the values.
- all connections successful
- partly successful
- no connection successful
- no measurement
Why connections fail
Failed connections in the selected period, by type of error.
Affected Microsoft server networks
The Exchange Online host names point to many servers. Each measurement goes to a randomly chosen address. The table groups the addresses into networks (/24) and shows whether some server groups are affected more than others.
How we measure
- Every 5 minutes the measuring server opens three connections to each endpoint, one after another: name resolution (DNS), TCP connection, TLS encryption and the server's first response (HTTP status or SMTP greeting).
- Successful means: the server answered each step within 10 seconds, for HTTP with a status below 500. No mailbox sign-in takes place.
- The measuring server is located in a data centre in Europe. Its internet exit changes location regularly; in the selected period measured via: ….
- Entra ID (sign-in) and Microsoft Graph run on separate infrastructure and serve as a comparison; the reference service (Cloudflare) checks the measuring server's own network.
Limits of the measurement
- This is a single measuring point. Other regions and networks may be affected differently.
- It measures reachability, not whether mailboxes, calendars or mail flow work in your tenant. Service health in the Microsoft 365 admin center remains authoritative for your tenant.
- The measurement is deliberately light (about 20 connections every 5 minutes) and puts no load on the services.
Scope
Microsoft lists more than 15 affected services. The outage first hit the Americas and, since the evening of 12 November, all regions, including Europe and Switzerland. Reports on Downdetector rose sharply from about 17:35 UTC; according to media reports, there were at times more than 100,000 from the US alone.
Analysis
Microsoft checks its servers automatically. Servers that fail the check are switched off. On 12 November, so many servers failed that the rest had to take over their work and became overloaded too. Restarted servers failed again right away because too many requests were waiting. Experts call this a cascading failure.
That is why the fix takes so long: Microsoft has to slow down and reroute requests until the servers run stably. Since 07:29 UTC, a fix (patch) has been helping and is being rolled out step by step. At midday, Microsoft reports further progress in Europe and North America.
The exact cause is still open. Microsoft stopped all software rollouts at the start and is assessing whether to undo recent changes. This points to a change as the trigger. According to Microsoft, there are no signs of an attack. Microsoft intends to name the cause in a final report a few days after the fix.
Affected services
| Service | Impact according to Microsoft |
|---|---|
| Exchange Online | No mailbox access via Outlook on the web, Outlook desktop (MAPI), ActiveSync and REST; mail delivery delayed. |
| Microsoft Teams | Calendar, invitations and meeting edits fail, as do new live events; private and shared channels cannot be created or deleted. Meetings, calls and chats mostly work. |
| Defender for Office 365 | Safe Links and Safe Attachments, quarantine, Threat Explorer and message trace disrupted; rewritten links cannot be opened. |
| Defender XDR | Some Defender portal pages do not load. |
| Microsoft Purview | DLP, sensitivity labels, Communication Compliance, eDiscovery, audit search and protected content disrupted. |
| Microsoft Copilot (Microsoft 365) | Copilot Chat may not load; no access to mail, calendar and files. |
| Microsoft Graph | Requests for mail, calendar and contacts time out; Teams management via Graph affected. |
| Microsoft Intune | Enrollment, check-ins and compliance policies intermittently affected. |
| SharePoint Online | Features that depend on Exchange disrupted; files in SharePoint remain available. |
| Microsoft 365 admin center | Some pages do not load. |
| Microsoft Planner, To Do | Plan and task operations unreliable; To Do may not load. |
| Microsoft 365 Apps, Apps admin center | Recently used files missing; Cloud Update, Cloud Policy and Inventory disrupted. |
| Power BI, Microsoft Fabric | Exporting and editing content with sensitivity labels fails. |
| Windows 365 | Cloud PCs using Purview Customer Key cannot be reached. |
| Universal Print, Microsoft 365 Education | Print jobs and admin features disrupted; in Education, Assignments and Reading Progress affected. |
Notes for IT staff
- Track the status: Microsoft 365 admin center, Service health, EX1550543 and MO1550821. If the admin center does not load (it is affected itself), status.cloud.microsoft and @MSFT365Status on X show the public version.
- Do not recreate Outlook profiles, reinstall Outlook or reset passwords: the problem is on Microsoft's side. In Cached Exchange Mode, mail, appointments and contacts that were already synchronized stay readable. A new profile would have no local cache and therefore no data at all.
- Do not change MX records. Exchange Online Protection keeps accepting mail; where delivery fails, sending servers retry on temporary errors (4xx) for hours to days. A short-notice MX change is more likely to create delivery problems than to solve them.
- Hybrid with Exchange Server: if Exchange Online accepts mail with delays, it backs up on the on-premises servers in the queue to the tenant (Get-Queue). Exchange Server drops messages once MessageExpirationTimeout expires, 2 days by default. If the outage continues, raise it temporarily, e.g. Set-TransportService -MessageExpirationTimeout 4.00:00:00.
- Safe Links: links rewritten by Defender for Office 365 cannot be opened during the outage. Do not turn Safe Links off globally; users should ask for urgent links through another channel.
- According to Microsoft, Teams meetings, calls and chats mostly work. Calendars, invitations and new live events fail: share meeting links in chat if needed.
- Quarantine releases, message trace, the Tenant Allow/Block List and Purview features (DLP, labels, eDiscovery, audit) are disrupted. Collect requests and process them after recovery.
- Intune: enrollment, check-ins and compliance evaluation may fail intermittently. Do not loosen Conditional Access hastily; check device compliance status after recovery.
Check after recovery
- Check message trace in the Exchange admin center for delayed, rejected (NDR) or stuck messages.
- Hybrid: let the Exchange server queues drain (Get-Queue, Retry-Queue if needed) and reset a raised MessageExpirationTimeout.
- Catch up on quarantine releases and Tenant Allow/Block List entries; review changes made during the outage.
- Check third-party journaling and archiving solutions for gaps.
- Read the Post Incident Report in the admin center (Service health, History) and complete your own incident documentation.
Template for informing users
Subject: Email and calendar disruption (Microsoft 365)
Since 12 November 2026, Microsoft has been experiencing a worldwide outage of Exchange Online. Outlook, Outlook on the web and the mail app on your phone may be unable to connect at times; new emails arrive with a delay, and calendar features in Teams are limited. The cause is on Microsoft's side, not on your device.
Please do not reinstall Outlook or reset your password. Emails and appointments that have already been loaded remain readable in Outlook. Teams meetings and chats mostly work; share meeting links in chat if needed.
Emails are generally not lost but will be delivered once the issue is resolved. We will let you know as soon as the service is running normally again.
Terms explained
- Exchange Online
- The Microsoft 365 service that stores email, calendars and contacts. Outlook is the program used to access it.
- Outlook on the web, MAPI, ActiveSync, REST
- The ways Outlook in the browser, Outlook on the computer, mail apps on the phone and other programs access the mailbox.
- Microsoft 365 admin center
- The management portal of Microsoft 365. Under “Service health”, IT staff see the official status of outages.
- EX1550543, MO1550821
- Microsoft's incident IDs for this outage.
- Health check
- An automatic check whether a server is working properly.
- Patch
- A fix to software or configuration.
- Safe Links
- A Microsoft security feature that rewrites links in emails and checks them for threats when clicked.
- Hybrid
- Companies that still run their own Exchange servers alongside Exchange Online.
- Post Incident Report
- Microsoft's final report on the cause and measures after a major outage.
Comments
Comments are loaded from GitHub / Giscus.