Microsoft 365 outage: recovery continues, moving to extended monitoring. Microsoft is confirming that the backlog of email has been processed

What you should do

Even after you can connect again, email delivery may take a while to catch up. If an email has not arrived, do not rush to resend it. Wait a little
Do not reinstall Outlook or change your password. The problem is on Microsoft’s side, and neither will fix it
For urgent matters, use a Teams chat or a phone call. Do not rush to resend emails you already sent. Microsoft has extended the period it keeps trying to deliver delayed email to 48 hours. After recovery, check whether your message arrived
Do not open links in emails or messages that tell you to “log in again because of the outage”

Disaster news for each company

This page reads the same for everyone

But what stopped at your company, and what each person in sales, the factory and management should do now, differs from company to company. We have begun using agents to make disaster news for each company: checking it against the company's map, writing it separately for each employee, and delivering it by another means.

Try the notice that reaches each employee

For companies, and for individuals through "My Disaster Preparedness News," we will announce it in our newsletter once we have decided how to start.

Emails are sent in Japanese. You can unsubscribe at any time. (Privacy policy)

What we know, what we don’t know yet, and what is being done

What we know

The outage affecting the Microsoft 365 connector for Claude and Claude for Outlook lasted from 1:03 a.m. JST on November 13 to 1:40 a.m. JST on November 14, and is resolved (Anthropic, 6:49 a.m. JST)
The cause was a problem in earlier work to increase network capacity that surfaced after a recent network change. It is a technical problem unrelated to cyberattacks or AI agents (2:57 a.m. JST on November 14)
It began with Outlook at 1:17 a.m. JST on November 13, and became an outage across all of Microsoft 365 at 3:26 a.m.
Traffic did not surge. Servers that failed automated health checks were removed one after another, and the load piled up on the rest
The process that sets up encrypted connections was overwhelmed, and the servers could no longer handle requests (announcement at 10:06 a.m. JST)
The impact is limited to features that exchange email (revised at 9:45 a.m. JST)
Email systems run on a company’s own servers may also be affected where they exchange mail with Exchange Online (3:16 p.m. JST)
Microsoft says there is no sign that the outage was caused by a malicious attack (12:48 p.m. JST)
Microsoft says Japan is relatively stable (9:00 p.m. JST)
The period for which delivery of delayed email is retried was extended to 48 hours (6:05 p.m. JST)
Consumer services (Outlook.com, OneDrive, OneNote, Microsoft To Do, Teams and the web version of Microsoft 365) may also be unavailable (7:26 p.m. JST. No relationship to the business outage is stated)

What we don’t know yet

When full recovery will come (Microsoft says it does not yet have an outlook. 5:01 a.m. JST)
The detailed sequence of events (Microsoft says it will publish a post-incident report after verification)
How many companies in Japan were affected

What Microsoft is doing

Microsoft confirmed that recovery is continuing. It is watching whether the measures it applied keep working and whether the backlog of email is being processed, and is addressing the remaining impact on related products (11:13 a.m. JST)
Because the impact was large, it is monitoring for longer to see whether any further measures are needed. It will call the incident resolved only after confirming that all the backlogged email has been processed (7:00 a.m. JST)
It is flushing the backlogged email and rebalancing traffic. Access comes back first, and email delivery catches up afterward (5:01 a.m. JST)
It has applied a fix for the underlying problem. It is restoring service to the remaining users and regions in stages, confirming stability at each step (2:57 a.m. JST on November 14)
It is focused on resolving the remaining impact and restoring service reliably. It is confirming recovery for each connection method and watching whether stability holds (12:52 a.m. JST)
It is shifting its focus to resolving the remaining problems and restoring stable service for all users. It is confirming recovery for each connection method (11:55 p.m. JST on November 13)
It applied additional measures and is investigating the remaining problems while bringing back the remaining infrastructure. The backlog of email has dropped significantly (11:01 p.m. JST)
It is stabilizing the infrastructure that has recovered, finishing repairs where problems remain, and working down the backlog of delayed email. It is confirming recovery for each connection method (9:59 p.m. JST)
Recovery is progressing in Europe and North America is improving too. It is fixing disruptions in the remaining regions and stabilizing them (9:00 p.m. JST)
Recovery work has finished on part of the infrastructure in the Asia-Pacific region, and it is extending this to infrastructure in other regions. It is adding capacity to stabilize service (8:01 p.m. JST)
It is carrying out recovery in stages. Some features are improving little by little, but conditions are uneven and capacity has not yet returned to normal levels (6:57 p.m. JST)
It is cycling through the infrastructure one unit at a time: draining traffic, restarting, repairing, and restoring in stages. Signs of stability have appeared, but capacity has not yet returned to normal levels (6:05 p.m. JST)
To avoid losing delayed email, it extended the period for which it keeps trying to deliver it to 48 hours
A fix to components involved in authentication and routing settings stabilized the infrastructure it was applied to. It is extending it to other infrastructure, starting with regions currently in business hours (5:06 p.m. JST)
It is building additional changes to improve stability and capacity further
After restarting infrastructure, it is checking whether the fix it built is working. So far the improvement is slight (3:54 p.m. JST)
To rebuild the front-end servers, it is considering traffic controls such as throttling traffic, adjusting traffic rules, and shifting to another region (3:14 p.m. JST)
It is running two efforts in parallel: tracing why calls to the back-end servers surged, and recovering the back-end servers, identifying the affected machines and repairing them (2:43 p.m. JST)
It is adding capacity by increasing the number of virtual machines
It is trying to force a recent network change onto the malfunctioning machines, but the machines are frozen and cannot take it
It is adjusting traffic rules, rebalancing traffic, and blocking targeted traffic. In regions where it intervened, part of the service comes back but does not stay back
On the persistent traffic imbalance, it is re-examining traffic flows and infrastructure and looking for ways to improve routing
Because the change it applied is not working as expected, it is looking for another way to add capacity
It is restarting the affected infrastructure and extending repairs while confirming stability
It is adding infrastructure capacity to support traffic in regions that are recovering
It is rolling back a network configuration change whose timing matches the start of the outage
In some locations, it is limiting the types of traffic that reach the servers

Status by service

Service	Status	Details
Outlook (email and calendar)	Partial	In parts of Asia-Pacific, connections and the processing of delayed email are improving (8:01 p.m. JST). Some users still cannot connect intermittently, cannot reach their mailbox, or see delayed email delivery. Email on a company’s own servers may also be affected where it exchanges mail with Exchange Online
Microsoft 365 Copilot	Down	Chat will not open or does not respond, and email and calendar cannot be used
Teams	Partial	Features that depend on email, such as calendar and meeting booking. Chat and meetings are mostly usable
SharePoint	Partial	Features that depend on email. Files saved directly can still be opened
Defender for Office 365	Partial	Safe Links, which checks whether links are safe, may not be working
Defender XDR · Purview	Partial	Some admin screens will not open. Data loss prevention (DLP) and similar features are affected
Microsoft Graph and connectors	Partial	Requests that handle email, calendar and contacts may time out. The Microsoft 365 connectors for ChatGPT and Claude also failed. The Microsoft 365 connector for Claude and Claude for Outlook show signs of recovery (12:40 a.m. JST; some SharePoint searches still return errors). The Microsoft 365 connector for Claude is resolved (6:49 a.m. JST)
Consumer services	Partial	Outlook.com, OneDrive, OneNote, To Do, Teams and others. They may be unavailable, or some operations may not work (7:26 p.m. JST). Caused by the same problem, and has recovered widely and become stable (3:57 a.m. JST)
Azure	Available	No company-wide outage is shown on the service status page

Based on Microsoft’s incident log (revised at 9:45 a.m. JST). The consumer row is based on the Microsoft 365 service status (7:26 p.m. JST). This does not mean every feature of each service stopped.

Text you can forward inside your company

[Microsoft 365 outage (11:42 a.m. JST, November 14)]
A Microsoft-side outage affecting Microsoft 365 (Outlook and others) is continuing. It is not a problem with your PC or account, so please do not reinstall Outlook or change your password.
- For urgent matters, use a Teams chat or a phone call
- Do not open links in emails that say “please log in again”
- Do not rush to resend emails you sent during the outage (delivery keeps being retried for 48 hours). After recovery, check that the recipient received it
Microsoft explains that the cause was a problem in network work, and that it is unrelated to cyberattacks or AI. Service has recovered widely, and the remaining regions are being restored in turn. If you still cannot connect, please try again after a while. Even once you can connect, email delivery may take a while to catch up. Microsoft says it will call the incident resolved only after confirming the backlogged email has been processed. Personal Microsoft accounts (Outlook.com and others) have also been affected.
Latest status: https://teleport.jp/dataiscity/en/news-microsoft-365-outage-live/

Timeline (newest first)

11:42
OngoingOur updates every 30 minutes end. Microsoft has not yet announced a resolution
The next announcements are at 1:00 p.m. (consumer) and 1:30 p.m. (business)
Source: Microsoft’s announcement
11:13
OngoingRecovery is confirmed to be continuing. No announcement of resolution yet
Most of the impact on related products has been addressed. The next announcement is at 1:30 p.m.
Source: Microsoft’s announcement
9:39
OngoingConsumer services also move to extended monitoring
The next announcement for consumer services is at 11:00 a.m.
Source: Microsoft’s announcement
7:00
DowngradedRecovery continues. Because of the scale of the impact, moving to extended monitoring
It will be called resolved only after confirming that all backlogged email has been processed
Source: Microsoft’s announcement
6:49
AnnouncedThe outage affecting the Microsoft 365 connector for Claude is resolved
Source: Anthropic’s announcement
5:01
OngoingFor business too, no timeline yet for full recovery
Access returns first, and email delivery catches up later
Source: Microsoft’s announcement
5:03
OngoingAccess comes back first, and email delivery catches up afterward
An announcement for consumer services. There is no timeline yet for full recovery. The next announcement is at 9:00 a.m.
Source: Microsoft’s announcement
3:57
OngoingFor consumer services too, the same apology and explanation of the cause
Consumer services such as Outlook.com had the same cause
Source: Microsoft’s announcement
2:57
AnnouncedMicrosoft apologizes and explains the cause. It is unrelated to attacks or AI
A problem in earlier work to increase network capacity surfaced after a recent network change. A fix for the underlying problem has been applied, and the remaining regions are being restored in turn
Source: Microsoft’s announcement
1:24
OngoingFor consumer services too, signs that recovery is continuing
The investigation into the cause continues. The next announcement for consumer services is at 3:30 a.m.
Source: Microsoft’s announcement
0:52
OngoingMoving to resolving the “remaining impact”. The next announcement is in 2 hours
This is not full recovery
Source: Microsoft’s announcement
0:40
OngoingSigns of recovery for the Microsoft 365 connector for Claude as well
Some SharePoint searches still return errors
Source: Anthropic’s announcement
23:55
OngoingImprovement continues. Focus shifts to resolving the remaining problems
Some users still cannot connect intermittently, and email is delayed
Source: Microsoft’s announcement
23:10
OngoingConsumer services may take more time to resolve
The next announcement for consumer services is at 1:30 a.m. on November 14
Source: Microsoft’s announcement
23:01
OngoingThe backlog of email has dropped significantly and is nearing normal levels
Some users still see intermittent problems
Source: Microsoft’s announcement
21:59
OngoingRecovery continues in North America, Europe and Asia-Pacific
Some users still cannot connect intermittently, and email is delayed
Source: Microsoft’s announcement
21:00
Ongoing“Japan is relatively stable.” Europe and North America improve too
Some users still see intermittent problems
Source: Microsoft’s announcement
20:55
OngoingRecovery is progressing for consumer services too
The next announcement for consumer services is at 11:30 p.m.
Source: Microsoft’s announcement
20:01
DowngradedRecovery work is finished in parts of Asia-Pacific. Connections to mailboxes improve
Some users still see intermittent problems. Other regions come next
Source: Microsoft’s announcement
19:26
AnnouncedConsumer Outlook.com and OneDrive are affected too
An announcement from the status page for consumer services. No relationship to the business outage is stated
Source: Microsoft’s announcement
18:57
OngoingRecovery is proceeding in stages, with small improvements in some features
Conditions are uneven, and capacity has not yet returned to normal
Source: Microsoft’s announcement
18:05
OngoingSigns of stability. The period for retrying delivery of delayed email is extended to 48 hours
Capacity has not yet returned to normal levels
Source: Microsoft’s announcement
17:06
OngoingThe fix is confirmed to stabilize infrastructure, and is extended starting with regions in business hours
This is not full recovery
Source: Microsoft’s announcement
16:24
OngoingProgress with the fix on some infrastructure. Extending it to other servers
A fix to components involved in authentication and routing settings
Source: Microsoft’s announcement
15:54
OngoingAfter restarts, checking whether the repairs are working. Slight improvement
Capacity is being added with virtual machines. This is not full recovery
Source: Microsoft’s announcement
15:16
AnnouncedEmail on a company’s own servers may also be affected where it exchanges mail with Exchange Online
Source: Microsoft’s announcement
15:14
OngoingConsidering traffic controls to rebuild the front-end servers
Such as throttling traffic and shifting to another region. The explanation that there is no sign of an attack is unchanged
Source: Microsoft’s announcement
14:43
OngoingInvestigating why calls surged, and identifying and repairing machines (Outlook)
Source: Microsoft’s announcement
13:56
OngoingCalls to the back-end servers surged. Repairs cannot be applied to frozen machines
Adding virtual machines to reduce the load
Source: Microsoft’s announcement
13:10
OngoingPart of the service comes back in regions where it intervened, but does not stay back
Source: Microsoft’s announcement
12:48
Revised“No sign that this was caused by a malicious attack”
Re-examining traffic flows and infrastructure, and looking for ways to improve routing
Source: Microsoft’s announcement
11:54
OngoingThe change it applied is not working as expected
Source: Microsoft’s announcement
11:09
OngoingAdding infrastructure capacity to support regions that are recovering
Source: Microsoft’s announcement
10:45
OngoingProgress with restarts. Extending repairs while confirming stability
Source: Microsoft’s announcement
10:06
RevisedThe mechanism of the outage is understood. The process that sets up encrypted connections was overwhelmed
Began rolling back a network configuration change whose timing matches (not stated to be the cause)
Source: Microsoft’s announcement
9:45
DowngradedThe list of impacts is revised. It is limited to features that exchange email
Teams chat and meetings are mostly usable. Safe Links may not be working
Source: Microsoft’s announcement
8:59
OngoingServers that were restored stop responding again once they receive traffic
Source: Microsoft’s announcement
7:58
OngoingLimited recovery in some regions
Source: Microsoft’s announcement
7:03
RevisedChanging the recovery approach. Isolating part of the infrastructure and restoring it little by little
Source: Microsoft’s announcement
4:43
OngoingIdentifies that servers are being pulled out after failing automated health checks
Source: Microsoft’s announcement
4:18
RevisedRevises its assessment: “this was not an increase in traffic”
Source: Microsoft’s announcement
4:06
OngoingHalts new deployments and configuration changes
Source: Microsoft’s announcement
Around 4:00 a.m.
OngoingReports surge in Japan as well. More than 300 for Outlook by around 6:00 a.m.
Source: Nikkei (Downdetector)
3:26
UpgradedBecomes an outage across all of Microsoft 365 (Teams, SharePoint, Copilot, Defender XDR and more)
Source: Microsoft’s announcement
1:17
AnnouncedAn Outlook (Exchange Online) outage begins
Source: Microsoft’s announcement

Sources

Microsoft incident records (EX1550543, MO1550821; a copy published by the University of Pennsylvania; times are in UTC) https://mailservices.isc.upenn.edu/computing/email/penno365/alerts/ms-incidents.html
Microsoft, "Microsoft 365 service status"https://status.cloud.microsoft/
Microsoft, "Azure status"https://azure.status.microsoft/ja-jp/status
Anthropic, “Claude Status” (Microsoft 365 connector) https://status.claude.com/
OpenAI, “Status” (Microsoft 365 connector) https://status.openai.com/incidents/01M78GAT3AK9P9W8Z230KHFVE8
Nikkei (Downdetector reports in Japan, November 13, 2026) https://www.nikkei.com/article/DGXZQOGN12CY20S6A111C2000000/
The full list of sources for the background and sequence of events is in the references at the end of the article

About this page

Our updates every 30 minutes ended at 11:30 a.m. JST on November 14. For any announcements after that, check Microsoft’s service status, the original source.

This page covers the same Microsoft 365 outage in the format of disaster reporting. At the top are the current stage and what you should do, and the timeline stacks up below, newest first. If you want to read what happened and why it spread as a narrative, see the article.

The format draws on crisis communication from the U.S. Centers for Disease Control and Prevention (CDC), the Japan Meteorological Agency’s warning levels, the U.S. Federal Emergency Management Agency (FEMA) alert guidance, how IT outages are communicated (status pages), and live news updates. The stage levels are the judgment of the Data is CITY Media editors, not Microsoft’s categories.

All times are Japan time (JST). Sources are Microsoft’s incident record (a copy published by the University of Pennsylvania) and news reports.