On Thursday night, 12 November 2026, Outlook stopped opening mailboxes for people all over the world, and Nepal was not spared. By the time offices in Kathmandu opened on Friday morning, Microsoft had been fighting the problem for more than twelve hours and email was still not back to normal. I work in IT and Microsoft 365 is a big part of my day job, so I followed this one update by update.
This post puts the facts in one place: when it started, what broke, how it hit Nepal and the rest of Asia, what Microsoft has and hasn't said about the cause, and what an admin should actually do while it's happening. I've added Nepal time next to UTC, because working that out in your head at midnight is no fun. I wrote this on Friday evening, 13 November, and the incident was still open at that point.
Quick answer
The Microsoft 365 outage of November 2026 began on Thursday 12 November at 16:17 UTC (10:02 pm Nepal time), when users lost access to Exchange Online mailboxes. It spread to Teams calendars, SharePoint sharing, Copilot and other services that depend on Exchange, and hit users in every region, including Nepal, India, Japan and the rest of Asia. Microsoft tracks it as EX1550543 and MO1550821. The company says many servers became unavailable, a process that sets up secure connections was overwhelmed, and there is no sign of a cyberattack. More than 20 hours later it was recovering region by region, with no confirmed root cause. Email was delayed, not lost.
What happened in the Microsoft 365 outage?
Exchange Online, the service behind Outlook and work email in Microsoft 365, lost a big share of its server capacity. Mailboxes would not open in Outlook on the web, the desktop app, or on phones. Because so many other Microsoft products lean on Exchange for mail and calendar data, the trouble spread well beyond email.
Downdetector, the site where people report service problems, logged about 130,000 Outlook reports from US users at the peak, and reports from 74 countries and territories within roughly an hour, according to Al Jazeera's summary. Those numbers only count people who bothered to file a report, so the real figure is much higher. Microsoft has not said how many users were affected.
There was also a smaller Teams incident earlier the same day. Chat was delayed for users in the Americas from 14:45 to 15:53 UTC, tracked as TM1550488, and Microsoft closed that one. Whether it's related to what came next has not been stated.
Timeline in UTC and Nepal time
I built this from Microsoft's status updates as recorded by azureblog.co.uk and LazyAdmin, and checked the key points against the Microsoft 365 Status account. Nepal is 5 hours 45 minutes ahead of UTC.
| UTC | Nepal time | What happened |
|---|---|---|
| Thu 16:17 | Thu 10:02 pm | Users start losing mailbox access. EX1550543 begins. |
| Thu 18:26 | Fri 12:11 am | MO1550821 opened for services that depend on Exchange Online. |
| Thu 19:06 | Fri 12:51 am | Microsoft pauses its own deployments and configuration changes. |
| Thu 19:18 | Fri 1:03 am | Microsoft says the load is not from extra traffic. Servers have dropped out and the rest are carrying their work. |
| Thu 19:43 | Fri 1:28 am | Affected servers are failing automated health checks and being pulled from service. |
| Thu 20:56 | Fri 2:41 am | A different health check method brings some servers back. |
| Thu 22:00 | Fri 3:45 am | Microsoft admits the fixes so far haven't given the relief it expected. A core process on the servers is overwhelmed. |
| Fri 00:00 | Fri 5:45 am | Restored servers go unresponsive again once they take traffic. Traffic limits applied in some locations. |
| Fri 01:50 | Fri 7:35 am | Secure connection process named as the overwhelmed part. Rollback of a network configuration change under way. Restarts show positive results. |
| Fri 03:07 | Fri 8:52 am | A change is not taking effect as expected. Other ways to add capacity are explored. |
| Fri 04:11 | Fri 9:56 am | Microsoft confirms it sees no sign of malicious activity. |
| Fri 06:57 | Fri 12:42 pm | Only marginal success after patches. Power Apps added to the affected list. |
| Fri 09:10 | Fri 2:55 pm | Early signs of stabilisation. Email timeout extended to 48 hours so delayed mail isn't dropped. |
| Fri 11:04 | Fri 4:49 pm | Recovery completed in parts of Asia Pacific. Mailbox access improving. |
| Fri 12:00 | Fri 5:45 pm | Europe recovering, Japan more stable, North America improving. Some users still have intermittent issues. |
Read that table from top to bottom and one thing stands out. For the first eight hours, almost every fix either did nothing or worked for a while and then fell over again. That's not a criticism from the cheap seats. It's what a hard outage looks like from the inside.
Which Microsoft 365 services stopped working?
Anything that talks to Exchange Online was at risk. Things that don't need it mostly carried on, which is why Teams calls kept working while Teams calendars didn't.
| Service | What broke |
|---|---|
| Exchange Online / Outlook | No mailbox access on web, desktop or mobile. Email delivery delayed. |
| Teams | Calendar, meeting invites, live events and channel creation could fail. Calls, meetings and chat mostly fine. |
| SharePoint Online | Sharing with external users could fail. Existing files still opened. |
| Copilot Chat | Might not load, or couldn't reach mail, calendar and files. |
| Defender for Office 365 / Purview | Data loss prevention and protection policies delayed or failing. Portal pages not loading. |
| Microsoft Graph | Mail, calendar and contact requests timing out, which breaks third-party apps and scripts. |
| Power Apps | Intermittent sign-in failures in the mobile apps. |
| Also listed | Planner, To Do, Intune, Universal Print, Fabric and Power BI (labelled content), parts of the admin center. |
The Graph line matters more than it looks. Outside tools that read Microsoft 365 mail and calendars use Graph, so they failed too. Both OpenAI and Anthropic reported problems with their Microsoft connectors during the incident.
Was Nepal affected? Yes, along with the rest of Asia
Nepal didn't make the international headlines, and I haven't found a Nepali organisation named in the foreign coverage. But this was not a US or Europe problem that we watched from a distance. Microsoft's own advisory said users in any region could be affected, and that included us. If your office in Nepal keeps its email in Microsoft 365, as many banks, INGOs, colleges and IT companies here do, your mailboxes were on the same struggling service as everyone else's.
The timing made it feel different here. It began at 10:02 pm on Thursday, so most people noticed nothing until Friday morning, and then walked into a full working day with shaky email. Rappler in the Philippines described it well: the outage spilled into Asia-Pacific business hours even while Microsoft was reporting partial recovery. Offices in Nepal that share a tenant with a head office abroad had the worst of it, because the people who could explain what was going on were asleep.
Across the region, this is what has been reported so far:
| Country | What was reported |
|---|---|
| Nepal | Microsoft 365 tenants here were affected like every other region. Hit during Friday office hours. |
| India | Outage reports early on Friday. The airlines IndiGo, Akasa Air and Air India were listed among organisations reporting disruption. |
| Japan | Government systems disrupted. Staff at companies couldn't reach Outlook, Teams or Copilot, and some reportedly went back to fax machines. |
| Singapore | Government and court systems affected. |
| Philippines | Communications at the Supreme Court disrupted. |
| Australia | Federal agencies reported disruption, and local experts warned of large losses for businesses. |
The one piece of good news for this side of the world: Asia Pacific was the first region where Microsoft said recovery work was complete, in its update at 11:04 UTC (4:49 pm in Nepal) on Friday. The National reported the same. That was still most of a working day gone.
Outside Asia, the names in the news included parts of the NHS and many schools in the UK, US local governments, and companies such as Maersk, Aviva and Euronext, according to NBC's live coverage.
What caused it? What Microsoft has said so far
No root cause has been confirmed. Microsoft has promised a Post-Incident Report after everything is back, and until that's published, anyone telling you the exact cause is guessing. What we do have is how Microsoft's own explanation changed through the night.
- It's not a traffic spike. Early on, Microsoft said the high load wasn't coming from more traffic. A number of servers had become unavailable without warning, and the ones left were struggling to carry everything.
- The health checks looked guilty. Servers were failing automated health checks and being removed from service. Microsoft first suspected the component that processes those checks and tested a way around it.
- Then the real pressure point showed up. A core process on the affected servers was being overwhelmed. Later updates described it as the process that sets up secure connections. When it choked, the server stopped answering requests. So the health checks were probably right to fail those servers.
- A network change is being reversed. Around 20:00 UTC Microsoft said it hadn't found a recent change that matched the start of the trouble. Overnight it began rolling back a network configuration change that lines up with the start time. Lining up in time is not the same as being the cause, and Microsoft hasn't claimed that it is.
On security, Microsoft's status page said on Friday that it has "no evidence of malicious activity associated with this incident". Charles Lamanna, the Microsoft executive in charge of Copilot, agents and platform, said the same on X late on Thursday.
One more thing people are connecting to this: the outage came two days after November's Patch Tuesday, which SecurityWeek reports fixed 1,094 vulnerabilities. Microsoft has not linked the two. I'd treat that as a coincidence until the report says otherwise.
What to do if your office is affected
There is nothing you can fix on your side, and that is useful to know. Your job during an outage like this is to stop people making things worse and to keep them informed.
- Confirm it's Microsoft, not you. Open Service health in the Microsoft 365 admin center and look for EX1550543 and MO1550821. If the admin center won't load, use status.cloud.microsoft, which is hosted separately for exactly this reason.
- Tell users through another channel. An email saying "email is down" reaches nobody. Use Viber, WhatsApp, SMS or a phone call. Teams chat was mostly working this time, so that was an option too.
- Stop the self-repair. Users will want to reinstall Office, delete their Outlook profile or reset their password. None of that helps, and a deleted profile means a long re-sync later on a slow connection.
- Don't resend emails. Mail is delayed, not lost. Sending servers keep retrying, and Microsoft extended its email timeout to 48 hours so queued messages aren't dropped. Resending only means the recipient gets everything twice.
- Freeze your own changes. If you change DNS, conditional access or a mail flow rule in the middle of this, you won't be able to tell your problem from Microsoft's.
- Expect a slow return. Recovery has come region by region, and restored servers have failed again more than once. Don't announce "it's fixed" on the first good sign.
If Outlook shows "Disconnected" on a normal day with no incident posted, that's a different problem, and I've covered six fixes for a disconnected Outlook separately. The same goes for Intune: devices may look stale for a while after this, and you can force an Intune sync once the service is healthy again.
Can you get money back? The SLA credit
Microsoft's service agreement promises 99.9% monthly uptime for Exchange Online. In a 30-day month that allows about 43 minutes of downtime. This outage went far past that, so affected tenants can ask for a service credit.
| Monthly uptime | Roughly how much downtime | Service credit |
|---|---|---|
| Below 99.9% | More than 43 minutes | 25% |
| Below 99% | More than 7.2 hours | 50% |
| Below 95% | More than 36 hours | 100% |
A few details that catch people out, as LazyAdmin's write-up explains well:
- It isn't automatic. You open a service request in the admin center with the incident IDs, the start and end times for your users, and how many users were affected.
- There's a deadline. Claims for a November incident have to be in by the end of December 2026.
- It's a credit, not a refund, and it only covers the share of your licence that pays for the affected service. On a suite like Business Premium, that's the Exchange Online portion, not the whole bill.
- If you buy through a reseller, which is how most offices in Nepal get their licences, the claim goes through them. Ask before you file anything yourself.
So write down your own times now, while you remember them. When did your first user report it, and when did the last mailbox come back?
The catch: your status page was inside the outage
Here's the part that should bother admins more than the downtime itself. During this incident, parts of the Microsoft 365 admin center were on the affected list. So were the Defender and Purview portals. The place you go to find out what's broken was partly broken.
The same thing applies to your own office. If your emergency contact list is a file in SharePoint, your "IT is aware" notice goes out by Outlook, and your on-call plan is a Teams calendar entry, you have one vendor holding every tool you need to respond to that vendor failing. Most small offices are set up exactly like this, and nobody notices until a night like Thursday.
It's also not a one-off. In January 2026, Microsoft 365 was down across North America for almost ten hours, which Microsoft put down to heavy load and reduced capacity during maintenance. Twice in one year is enough to plan for.
A few things I learned
- Keep one channel outside Microsoft. A Viber or WhatsApp group for all staff, and a printed or offline list of phone numbers for the people who matter. It costs nothing and you set it up once.
- Bookmark status.cloud.microsoft today. You won't remember the address when the admin center is timing out.
- Early explanations change. Microsoft's own theory moved from health checks to an overloaded connection process to a network change in about six hours. Tell your manager what's known, and say clearly which parts are still open.
- Save the Post-Incident Report when it arrives. Service health only keeps history for 30 days. If you need it for an audit or a claim in January, it'll be gone unless you kept a copy.
- Delayed is not lost. Email was designed to queue and retry. Knowing that calms people down faster than any status update.
I'll update this post when Microsoft publishes its root cause. I don't think the answer to this is "leave the cloud". Running your own mail server is a second job. But it's worth keeping a few small things on your own hardware, which is why my newsletter runs on a self-hosted listmonk and not on someone else's platform. If your office needs a simple outage plan or a second look at how its Microsoft 365 is set up, that's part of my IT and network services.
Frequently asked questions
When did the Microsoft 365 outage start in November 2026?
The Exchange Online incident EX1550543 started on Thursday 12 November 2026 at 16:17 UTC, which is 10:02 pm Nepal time and 11:17 am US Eastern. Microsoft opened a second incident, MO1550821, at 18:26 UTC for other services that depend on Exchange. It continued through Friday 13 November.
Was Nepal affected by the Microsoft outage?
Yes. Microsoft said users in any region could be affected, and offices in Nepal that use Microsoft 365 had the same Outlook and Exchange Online problems. It started at 10:02 pm Nepal time on Thursday 12 November, so most people here felt it during office hours on Friday 13 November.
Which Asian countries were affected by the Microsoft outage?
Reports named India, Japan, Singapore and the Philippines, along with Australia, and Nepal was affected as well. Government systems in Japan and Singapore, the Supreme Court of the Philippines and several Indian airlines reported disruption. Asia Pacific was also the first region where Microsoft said recovery was complete.
Was the Microsoft outage caused by a cyberattack?
Microsoft says no. Its service health updates on Friday 13 November said there was no evidence of malicious activity, and executive Charles Lamanna said the same on X. Microsoft points to servers becoming unavailable and an overwhelmed secure connection process, but it has not confirmed a root cause yet.
Did I lose emails during the Outlook outage?
Almost certainly not. Email sent to you during the outage was queued and retried by the sending servers, and Microsoft extended its email timeout to 48 hours so delayed messages would not be dropped. Expect them to arrive late and out of order. There is no need to ask people to resend.
Why did Teams keep working when Outlook was down?
The outage was in Exchange Online. Teams calls, meetings and chat run on separate infrastructure, so they mostly kept working. The parts of Teams that read from Exchange did fail for many users: the calendar, meeting scheduling and invites, live events, and creating or deleting channels.
How do I claim a Microsoft 365 SLA credit for the outage?
Open a service request in the Microsoft 365 admin center and include the incident IDs EX1550543 and MO1550821, the start and end times for your users, and the number of users affected. Claims for November must be filed by 31 December 2026. If you buy licences through a reseller, claim through them.