Key takeaways
- The Exchange Online incident ran from 16:20 UTC on 12 November to 16:10 UTC on 13 November 2026, about 23 hours 50 minutes by Microsoft's preliminary report.
- Microsoft says the cause was a latent defect in a Windows change on its own gateway servers, combined with a traffic shift and reduced capacity. It was not a cyberattack.
- Microsoft's report is preliminary and a final version is still to come, so details may change.
- Every business that relies on cloud email should have a written plan for the day email stops.
What happened
On Thursday 12 November 2026, Microsoft 365 users around the world began losing access to Exchange Online, the service that hosts business email. Microsoft's incident record for Exchange Online is EX1550543, and a wider advisory, MO1550821, covered knock-on effects on other services. By Microsoft's preliminary post-incident report, the impact started at 16:20 UTC on 12 November and ended at 16:10 UTC on 13 November. That is about 23 hours and 50 minutes. In Malaysia time (UTC+8), the start fell just after midnight on Friday morning.
Many people could not open mailboxes in Outlook or in mobile and other mail clients, and some mail was delayed. Reports say Teams, SharePoint and OneDrive recovered earlier because they can move traffic away from a failing path. Exchange Online could not do that fully at its scale.
What Microsoft says caused it
According to the preliminary report as described by the outlets below, three things combined.
- A latent defect. A Microsoft 365 specific change to a Windows process on its network gateway servers, rolled out globally from July, contained a lock-contention problem. Lock contention means processes compete for the same resource until the server stalls. The change ran for more than three months without visible trouble. Microsoft says it does not affect ordinary Windows PCs or Windows Server.
- A traffic shift. About 15 hours before the outage, Teams traffic moved from an alternative gateway to the primary gateway that Exchange Online also uses. This is described as normal behaviour.
- Reduced capacity. Roughly 14% of gateway capacity was unavailable for an unrelated reason.
A separate weakness in the barriers meant to contain failures let the problem spread globally. Recovery was also slow, because servers that restarted were overwhelmed by queued traffic and failed again. Microsoft restored the earlier working code, took large groups of servers out of service, updated and restarted them, and returned them gradually. It also kept a long window for email delivery so delayed messages would not be rejected as undeliverable. Both reports state there was no sign of a cyberattack.
What is still uncertain
This is a preliminary report. A final post-incident report is expected and may add repair items or change the account. The sources also disagree on small details, such as the exact date the preliminary report was issued. Download counts and user complaint figures from outage trackers are not measures of how many customers were affected, so treat them with care.
Why it matters
Email is where invoices, quotations, password resets and customer replies arrive. A day without it is a real cost, even when the cause is on the provider's side. Two lessons stand out. First, a cloud provider can fail through a defect that sat unnoticed for months, so having a good provider does not remove the need for a plan. Second, the impact reaches beyond the inbox: services that depend on Exchange, such as approval workflows, ticket alerts and notifications, can quietly stop working too.
What businesses can do
- Agree an "email is down" plan. Decide who watches the Microsoft 365 Service Health page, who tells staff, and how you reach key customers by phone or messaging.
- Keep a second channel. Reports note that Teams chat and meetings mostly kept working. Make sure staff know to use it, and keep a list of customer and supplier phone numbers outside email.
- Map what depends on email. List alerts, approvals, ticketing and security notifications that rely on Exchange so you know what to check after an outage.
- Check your service credits. One source notes credits are not automatic and must be claimed through the admin center or your reseller, with November 2026 claims due by the end of December 2026. Confirm this against your own agreement.
- Review the final report. When Microsoft publishes it, read the repair list and ask your IT partner whether anything changes for you.
Sources
- Microsoft Reveals How a Hidden Windows Bug Took Down Microsoft 365 Services Worldwide (Cyber Security News)
- The Great Exchange Online Outage of November 2026 (Office 365 for IT Pros)
- Exchange Online Outage: Microsoft Traces November 2026 Disruption to Gateway Bug (WindowsForum)
If you want help planning for email and cloud downtime, our team can review your setup through servers and cloud, or you can book a free consultation.