Microsoft 365

Microsoft 365 down
for almost 24 hours

·Microsoft 365Exchange OnlineEmailBusiness Continuity

In short: between 17:20 on Thursday 12 and 17:10 on Friday 13 November 2026 (Central European Time), most Exchange Online users could not open their mailbox, and email arrived late. In its preliminary report Microsoft names three causes that combined and admits two things that concern every company: email goes through a single front door it cannot move away from, and the isolation between regions did not work. No email appears to have been lost, but there are a few checks to make and one question to answer before the next outage: how do you work when email is down?

Length of the outageAlmost 24 hours
Unresponsive servers at the peakMore than 80%
Defect in productionSince July 2026
Final report expected20 November 2026

What happened, in Central European Time

The outage began on Thursday 12 November at 17:20 CET, at the peak of traffic in the United States, and was declared over on Friday 13 at 17:10. For people working in Italy that meant the end of Thursday afternoon and almost all of Friday morning: Microsoft states that availability in Europe returned to healthy levels at 12:35, and that web access was back to normal worldwide at 15:17.

The service hit hardest was Exchange Online: mailboxes unreachable from Outlook, Outlook on the web, phones and the applications that use email, with delayed delivery. The other Microsoft 365 services largely stayed accessible, with problems in the features that depend on Exchange Online: among them Teams, SharePoint, OneDrive, Copilot, Defender and Purview. Teams chats and meetings worked for most of the outage.

Three causes together, none enough on its own

The failure hit the networking gateway, the servers that connections to Exchange Online and many other Microsoft 365 services pass through. Microsoft describes three factors which, taken one by one, had caused no problems:

CauseSince whenWhy it was not enough on its own
Latent defect in a Windows fix for the gateway serversIn production since July 2026It caused no issues for more than three months: it only appears under certain loads, on servers with more cores
Teams traffic moved onto the same gateway as Exchange OnlineAbout 15 hours before the outageIt is an automatic move that is part of normal operations
Nearly 14% of gateway servers out of serviceSince the day before, because of an unrelated issueThe remaining capacity was enough to carry the traffic

The defect was lock contention inside a Windows process which, under certain conditions, exhausted processor and memory until the server became unreachable. Microsoft’s load testing had not detected it because the lab used less powerful machines, with fewer cores, and a traffic mix different from the real one.

Why email stayed down for so long

  • 1The failure spread by itself — when a server stops responding, traffic is moved automatically to the healthy ones. It is the mechanism that normally makes the service reliable; here it loaded the remaining servers until they fell one after another. In less than two hours more than 80% of servers were unresponsive.
  • 2The isolation between regions did not hold — the gateway is split into thousands of capacity units grouped by region, designed to contain a failure. Because of a separate issue, which the preliminary report does not explain, these barriers did not stop the spread and the outage became global.
  • 3Exchange Online had no other route — Teams, OneDrive, SharePoint and the Office apps moved to the second gateway. Exchange Online did not: Microsoft writes that it cannot move out of that gateway “for scale reasons”.
  • 4Restarting did not help — servers returned to service were immediately overwhelmed by the waiting traffic and locked up again. The automated recovery processes were not designed for such a broad unresponsive state.
  • 5There was no quick way back — the faulty change was months old, so the tools normally used to undo a recent release did not apply. The solution was an update that returns the code to its previous state, rolled out to groups of servers starting at 8:20 CET on Friday.

What to check now

  • 1The important messages sent in those hours — quotes, orders, confirmations and deadlines sent between Thursday afternoon and Friday: check that they are not still in the Outbox or in Drafts and, if an expected reply is missing, ask the recipient to confirm.
  • 2Whatever uses email without anyone watching — business software, multifunction printers, websites and applications that send through Exchange Online, automated flows and shared mailboxes fed by rules. If a send failed during the outage, it may not have been retried.
  • 3Meetings and calendar invitations — invitations travel by email: it is worth rechecking the appointments set or moved in those hours.
  • 4The service credit claim — Microsoft’s service level agreement provides a credit when monthly uptime falls below 99.9%, but it has to be claimed by the end of the month following the incident. Customers who buy licences through a partner go through the partner.

What you need to keep working

There is a limit worth stating plainly: email is useful if someone on the other side reads and replies. In a global outage like this one, being able to write and receive while most of your customers and suppliers are down helps little. Email continuity really matters in partial failures, which are the most frequent: a region, a tenant, a single company down while everyone else is working. And not everyone you deal with uses Microsoft 365: those on another service keep writing and expect an answer.

An outage this long is rare, and no provider of this scale is immune to a similar failure: changing platform because of one incident is not an answer. The answer is deciding in advance how you work when email is not there. Four things really help, and one does not:

  • 1An alternative channel chosen in advance — who you talk to, and how, if email is down: phone, chat, an emergency address on a different service. It should be written down, shared with your main customers and reachable without opening the mailbox.
  • 2Email already received, readable outside Exchange — with our email archiving the archive is independent of Exchange Online and stays accessible from the web during an outage: a quote, an attachment or a phone number can still be found.
  • 3A mail gateway in front of Microsoft 365 — if Exchange Online does not respond, the gateway queues incoming mail instead of rejecting it. With our Email Security Gateway users can also keep receiving and writing email during the outage: the feature is active for every customer of the service. In a global outage it matters less, because whoever should reply is down too; when a single company or a region stops, it makes the difference.
  • 4Contacts and essential documents not only in email — customer contacts, procedures and the documents you need every day should also live somewhere that does not depend on the mailbox. In this incident SharePoint, OneDrive and Teams largely remained usable.
  • 5Backup does not bring email back — a Microsoft 365 backup is there to recover deleted or damaged data, and remains necessary. During a service outage, however, it does not give you a working mailbox: they are two different problems, with two different answers.

AtWorkStudio has been operating from Piacenza since 2000. We are certified ISO/IEC 27001, 27017, 27018 and ISO 9001, ACN-qualified (Italian National Cybersecurity Agency) for cloud services, members of Clusit (Italian Association for Information Security), of Confindustria Piacenza, in the RICT cluster, and of Confapi Industria Piacenza.

Status as of 17 November 2026

This article is based on Microsoft’s preliminary report, dated 15 November 2026, which Microsoft itself describes as provisional and subject to change. The final report, with the full root cause analysis and the list of long-term actions, is expected by 20 November 2026: we will update the article when it is published.

Sources

  • Microsoft — “Preliminary Post Incident Report”, incident EX1550543/MO1550821, 15 November 2026
  • Microsoft Learn — Microsoft 365 service description, “Service health and continuity”
  • Microsoft — “Service Level Agreement for Microsoft Online Services”

Frequently asked questions

Answers to the most common questions about the Microsoft 365 outage of 12-13 November 2026.

What if email goes down again?

We review with you what depends on email in your company and prepare the way to work when Exchange Online does not respond. Let’s talk.