On Thursday 12 November 2026, millions of people opened Outlook and got nothing. No inbox, no calendar, no sent mail. Within hours, the same failure had spread to Microsoft Teams, Copilot and SharePoint. The Microsoft 365 outage became the biggest internet disruption of the year — and it lasted more than a full day. Here is what happened, why it happened, and what it means for the next time a service you depend on goes dark.
What actually went down
The outage centred on Exchange Online, the service behind Outlook mailboxes. When it failed, Outlook on the web, on desktop and on phones could not reach people's mail. Email stopped arriving, calendars would not load, and sending a message went nowhere.
It did not stop there. The same incident, tracked by Microsoft as MO1550821 (with the Exchange Online part logged as EX1550543), also hit:
- Microsoft Teams — core chats and meetings mostly kept working, but anything that depends on a mailbox failed: scheduling meetings, creating new chats and calendar features
- Microsoft 365 Copilot — the AI assistant was disrupted
- SharePoint Online — document libraries and team sites were affected
- Defender XDR, Planner, To Do and Intune — also listed as affected
The failure started in North America and spread worldwide within hours — a reminder of how much of daily life rides on the same shared infrastructure, as explained in our guide to how the internet actually works. By the evening of 12 November, users in dozens of countries were reporting the same problems.
How big was it?
By the usual measures, this was enormous. Downdetector, the site that collects problem reports from users, logged more than 7.4 million reports about Microsoft's services worldwide. Reports peaked at nearly 140,000 in a single hour, and came from 74 countries and territories.
| Measure | What happened |
|---|---|
| Total reports | 7.4 million+ on Downdetector |
| Peak reports | ~140,000 in one hour |
| Countries affected | 74 |
| Duration | ~30 hours (incident record) |
| Largest outage of 2026 | Yes — 2.5× the next largest |
According to Ookla, which runs Downdetector, it was the largest outage of the past 12 months and the largest of 2026 so far — about two and a half times as many reports as the YouTube outage of February 2026. Reports also stayed high for about 20 hours, roughly twice as long as any other major outage of the year.
The timeline
- Thursday 12 November, ~17:25 GMT: reports of Outlook problems start climbing sharply
- 16:20 UTC: Microsoft opens incident EX1550543 for administrators and says it is investigating mailbox access problems
- Evening of 12 November: the disruption spreads beyond North America; a separate, shorter Teams problem in the Americas is fixed by about 21:25 UTC
- Friday 13 November: Microsoft says the underlying fault is fixed and reports sustained improvement, but recovery is still under way
- Friday 13 November, 16:10 UTC: Microsoft's incident record marks the end of the impact window — about 30 hours after it began
- Saturday 14 November, morning: Microsoft declares services fully recovered and confirms delayed mail has been processed
So from the first alarms to the final all-clear, roughly 36 hours passed, even though the core impact window was about 30 hours.
What caused it?
Microsoft's explanation, in its own words, is short: an issue introduced during an earlier network capacity upgrade surfaced after a more recent network change. In practice, that meant a large number of Microsoft's own servers became unexpectedly unavailable — they started failing automated health checks — and a process that sets up secure connections between servers became overwhelmed. The failures then cascaded: one system depending on another, each waiting on a service that was not answering.
Microsoft also extended its email timeout to 48 hours during the outage, so messages sent while mail was down would keep retrying for two days instead of bouncing back to the sender.
A full post-incident report has been promised, with a preliminary version within two business days and a final report within five. Until it appears, the one-sentence capacity-upgrade explanation is the official account — though The Verge, citing sources, has linked the failure to configuration changes in Azure virtual networks.
Why did it take so long to fix?
Restoring a service like Exchange Online is not a matter of flipping a switch. Engineers had to restart affected servers gradually, run health checks on each one, and confirm that the secure-connection process could cope before letting real mail flow again. Rushing that process risked a second collapse. On top of that, millions of delayed emails had to be processed in order once the servers were back. That is why the incident record shows impact ending on Friday afternoon, but Microsoft did not declare full recovery until Saturday morning — the last hours were spent clearing the backlog and confirming nothing else was broken.
Was it a cyberattack?
No. Microsoft said there was no evidence of malicious activity, and the outage was not related to AI or AI agents — a point Microsoft executive vice-president Charles Lamanna confirmed publicly. The fault was in Microsoft's own infrastructure, introduced by its own upgrade work. That is, in some ways, the more worrying lesson: the world's biggest productivity suite was knocked out for a day not by hackers, but by a routine change that went wrong.
What did it cost?
The bill is still being counted, but early estimates are large. The Anderson Economic Group estimated up to $7 billion in lost productivity. Morningstar DBRS put insured losses at $300–500 million. Microsoft's shares closed roughly flat on the Friday, at $523.88 — investors, for now, treated it as a one-off.
The human cost was more visible. Hospitals and government offices in several countries reverted to phone calls and paper records. In the UK, parts of the NHS went back to paper. New Zealand paused visa processing. In Australia, the Federal Police and ASIO were among the affected organisations. For a day, workplaces that run entirely on Outlook and Teams had to remember how work was done before them.
What happened to your emails?
If you sent an email during the outage, it was probably not lost — just delayed. Because Microsoft extended the retry window to 48 hours, messages queued up and were delivered once servers recovered. By the morning of Saturday 14 November, Microsoft said the backlog of delayed mail had been processed. If something you sent on 12 or 13 November still has not arrived, it is worth checking your sent folder and asking the recipient to check spam — but a permanent loss is unlikely.
What happens next
Two things are already in motion. First, Microsoft's post-incident report should explain exactly which change broke what, and what safeguards will stop it happening again. Second, affected business customers can claim service credits under their service-level agreement (SLA) — the compensation Microsoft owes when it misses its uptime promises. If your organisation pays for Microsoft 365, your IT team should check what you are owed.
Takeaway checklist
- Check the status page first: when Outlook or Teams fails, visit Microsoft's service health page before reinstalling anything — the problem is usually theirs, not yours.
- Keep a backup channel: a second way to reach colleagues (a phone number, a personal email, another chat app) turns an outage from a crisis into an inconvenience.
- Do not assume a cyberattack: most big outages are routine changes gone wrong, not hackers.
- Give delayed mail time: messages sent during an outage can take up to 48 hours to arrive.
- Know your SLA: if your business pays for Microsoft 365, an outage may entitle you to service credits — ask your IT team.
- Expect a report: Microsoft has promised a preliminary report within two business days and a final one within five — worth reading if your business depends on its services.
The Microsoft 365 outage was a reminder of a simple fact: when one company runs the email, calendar, chat and documents of a large share of the world's offices, a single bad upgrade can pause work in 74 countries. The fix, when it comes, will be technical. The lesson is organisational — have a plan for the day the cloud does not answer.