MONDAY, NOVEMBER 16, 2026 Tech news, minus the noise.
N News Cast
LATEST
Home  /  Cloud
Cloud

Microsoft 365 Outage Lasted More Than 36 Hours, and the Cause Was a Network Upgrade, Not Hackers

Microsoft restored its Microsoft 365 services on November 14 after a global outage that began November 12, traced to an earlier network capacity upgrade exposed by a later change.

Microsoft 365 Outage Lasted More Than 36 Hours, and the Cause Was a Network Upgrade, Not Hackers
Source: BalticServers.com (CC BY-SA 3.0) via Wikimedia Commons

Microsoft 365 outage lasted more than 36 hours

Microsoft 365 outage is the story of the week in cloud infrastructure, and it is now over. Microsoft confirmed on Saturday, November 14 that it had fully restored its Microsoft 365 services after a global disruption that lasted more than 36 hours, affecting users in more than 70 countries. Outlook, Teams, Copilot, SharePoint, and other services were hit. The email backlog has been processed, and Microsoft says it will publish a full post-incident report once its review is complete.

The outage began on Thursday, November 12. Exchange Online mailbox problems started at 16:17 UTC, with users unable to open their mailboxes. A wider Microsoft 365 outage followed, and a separate Teams incident hit North America the same morning. Downdetector recorded nearly 130,000 user reports at the peak. Microsoft tracked the incidents as EX1550543 for Exchange Online and MO1550821 for Microsoft 365. The end time, according to Microsoft's final update, was 4:10 PM UTC on Friday, November 13, with service marked restored late that evening.

What caused the Microsoft 365 outage

Microsoft traced the problem to a technical fault linked to an earlier network capacity upgrade, which surfaced after a subsequent network change. Several backend servers unexpectedly became unavailable, leaving the remaining servers struggling to handle the traffic. Microsoft initially suspected unusually high traffic, then found the real cause. Engineers deployed a patch targeting an authentication and routing component, restored traffic gradually, and extended the email timeout window to 48 hours to clear the delayed messages.

Microsoft was explicit about what the cause was not. The company said the incident was a technical issue not related to cyber activity, and not related to AI or agents. That last denial is notable, because speculation about AI agents causing infrastructure failures has become a reflex this year. Microsoft ruled it out in the same sentence as it ruled out a cyberattack.

The blast radius went beyond email

Outlook was hit hardest, but the damage spread through everything that depends on Exchange Online. Teams calendars, meeting scheduling, Copilot answers that draw on Outlook, calendar, or files, SharePoint, Planner, To Do, Intune, Defender, and Purview all suffered. Teams chat and meetings mostly kept working, and files stored in SharePoint were largely unaffected, but anything that needed to read a mailbox or a calendar broke.

That dependency chain is the real lesson. A single identity and routing component sat underneath a dozen products, and when it faltered, the products faltered together. Cloudflare's double outage in October showed the same pattern from a different vendor: one shared layer, many visible failures. The cloud's promise is that failures are isolated. The cloud's reality, shown twice in two months, is that shared infrastructure fails together.

What businesses should take from this outage

The first practical step is unglamorous: check whether your Microsoft 365 service level agreement gives you credits for this outage, and file for them. Microsoft's November 2026 SLA terms cover service credits, and a 36-hour outage qualifies. Second, review your incident plan for the next one, because there will be a next one. If your team could not work without Outlook and Teams for a day and a half, that is a finding, not a complaint. Decide in advance what the fallback is: a secondary email domain, a separate chat tool, or simply a plan for who does what when the cloud goes dark.

Third, remember that cached mail kept working. Users who had Outlook in cached mode could still read old messages, even while new mail stalled. That is a small argument for keeping local caches enabled and for not treating the cloud as the only copy of anything important. The 3-2-1 backup rule exists for ransomware, but the same thinking applies to outages: one copy is not a backup, and one provider is not a plan.

Microsoft's post-incident report will tell us more about the network change and the capacity upgrade that combined to cause this. Until then, the honest summary is that a routine infrastructure change exposed a latent fault, the fault took down a shared component, and recovery took more than a day because the fix had to be applied carefully while traffic was restored in stages. Nobody was hacked. Nobody's AI agent broke the cloud. A network upgrade did what network upgrades sometimes do, and 70 countries noticed.

AS
News Cast Editorial Team

We cover AI, gadgets, startups and security — original reporting and explainers, written by humans, for humans. Got a tip? Reach us via the contact page.