A cybersecurity team monitors a cloud email outage, with warning alerts on affected servers and restored service paths.
Microsoft's November 12–13 outage is worth a proper post-mortem. The commentary around it mostly says "the cloud is fragile". Microsoft's own preliminary account is more specific than that, and the specifics are the useful part.

A cybersecurity team monitors a cloud email outage, with warning alerts on affected servers and restored service paths. What happened​

Exchange Online was the main casualty. Most users could not reach their mailbox through Outlook or other mail clients. Microsoft's report puts the start at 16:20 UTC on Thursday, November 12 and the end of impact at 16:10 UTC on Friday, November 13. That is close to a full day. These are the report's incident boundaries. They don't mean every user was down for the whole period.

Microsoft tracked the incident in the admin center as EX1550543. The wider advisory was MO1550821. Microsoft's admin post said features that interact with Exchange Online and email services were affected, while parts of those services that don't depend on Exchange kept working. That caveat matters when reading the "Outlook, Teams and Copilot went down" headlines. Some Teams and Copilot features that touch mail were hit, but that is not the same as a full outage of each product.

Microsoft also said it was not aware of any malicious activity causing the outage. This was a failure in Microsoft's own infrastructure, not an attack.

Microsoft's explanation​

Microsoft's report is dated November 15 and is preliminary. The final report will supersede it. Note also that Microsoft's own admin messaging during the incident had pointed to an issue during a network capacity upgrade. The preliminary report gives a different and more detailed cause, so expect the account to change further.

According to the report as relayed by 365fanatic and The Register (via AI Weekly), three things lined up:

  • A latent defect. A lock-contention bug had sat inside Microsoft's network gateway servers since a July 2026 update. Under load it could exhaust CPU and memory and leave machines unresponsive. Per the report, the defect came from a fix made specifically for Microsoft 365 gateway servers. It does not apply to ordinary Windows deployments. Windows PCs and Windows Server were not affected.
  • A shift in traffic. About fifteen hours before the outage, Teams automatically failed over from its alternative gateway to the primary one, which also serves Exchange Online. That happened at 1:07 AM UTC on November 12. Microsoft describes this failover as normal, intended behavior.
  • Reduced capacity. Roughly 14% of gateway server capacity was temporarily offline. The report says that alone had not caused a problem, because there was enough headroom.

The bug surfaced only on high-core-count hardware, and Microsoft says its load testing focused on weaker machines with fewer cores. The change had been deployed for more than three months without visible impact before the incident.

Why it spread and why recovery was slow​

When servers stopped responding, traffic-management systems redirected users to other endpoints. That pushed more load onto the survivors, which then failed in turn. Microsoft says more than 80% of servers were unresponsive at the lowest point. It also says a separate issue meant the normal isolation between gateway partitions did not stop the failure going global.

Recovery was hard for three reasons:

  • Returning servers were swamped by queued traffic and fell over again.
  • Microsoft's automated recovery was not designed to restore service from such a broad unresponsive state.
  • Ordinary rollback and feature-flag tools could not quickly reverse a defect that had been latent for months.

Engineers wrote corrective code to restore the previous paths, deployed a patch reverting the defective change at 07:20 UTC on November 13, and declared services healthy at 16:10 UTC. Microsoft also kept a 48-hour email timeout window in place so delayed messages would not expire while processing caught up.

Which services could keep working​

Microsoft says some services could route around the faulty gateway, including Teams, OneDrive, SharePoint and Office apps. Teams chat and meetings reportedly worked for most of the outage. Services that use Exchange Online mostly stayed online, but the parts that work with Exchange were affected. Exchange Online itself could not fail out of the gateway for scale reasons.

The report lists a long set of affected services, including Copilot, Defender, Purview, Fabric, Intune, Planner, Power Apps, To Do, Power BI and Universal Print. The common thread is Exchange-dependent functions, not blanket downtime.

What this means for IT teams​

This is analysis from general industry practice, not Microsoft guidance. The reviewed sources establish no specific backup product, SLA credit or universal fix.

  • Map your mail dependencies. Alerts, approvals, ticketing, workflows and security notifications often run through Exchange Online. An email outage can break them even when the apps themselves are up.
  • Plan an out-of-band channel. If email and anything tied to it are unavailable, decide in advance how staff and incident responders will reach each other.
  • Know what keeps working. Teams chat and meetings, and file access, reportedly stayed available for most of the incident. Knowing that helps you steer users to the working tools.
  • Check Service Health. Microsoft says admins can read the preliminary report under EX1550543. A list of repair items is promised with the final report, dated 11/20/2026.
  • Don't expect tenant changes. 365fanatic reports that Microsoft said nothing needed changing in customer tenants.

The "expensive cold" question​

TheBusinessDesk's piece argues that businesses effectively rent their ability to function from a few technology giants. The accessible portion doesn't quantify any financial loss, and I have no verified dollar figure to offer. The Associated Press reported nearly 130,000 Outlook problem reports on Downdetector. That is a user-reported tracker number, not a count of affected customers or a measure of cost.

The fairer takeaway is architectural. The failure came from a latent bug, a routine failover, a capacity dip and isolation boundaries that didn't hold, and none of these would have been alarming on its own. The same combination in your own datacenter would be just as hard to anticipate. The difference with a hyperscale provider is that the blast radius is global and your recovery options are limited to waiting and communicating.

Microsoft's final report is due within days. Look there for the full repair list, especially any changes to partition isolation, recovery automation and load testing on newer hardware.

 

References

  1. Microsoft 365 outage 12 November 2026: post-incident report 365fanatic.com
  2. Microsoft pins Exchange Online outage on July code defect aiweekly.co
  3. Massive Microsoft 365 Outage Impacts Exchange Online, Teams, SharePoint, Copilot, and Defender XDR Users cybersecuritynews.com