Microsoft 365 Outage November 2026: Cause and Timeline
Outlook, Teams calendars and Copilot failed worldwide for about 30 hours on Nov. 12–13. Microsoft's preliminary cause, timeline and what admins should do.
Microsoft 365 suffered its longest major disruption of the year last week. Users around the world lost access to Exchange Online mailboxes, and Teams, Copilot, SharePoint and Defender were affected too. Microsoft’s incident record puts the impact window at 5:05 a.m. ET on Thursday, November 12, 2026 to 11:10 a.m. ET on Friday, November 13, about 30 hours from start to finish.
Microsoft has issued a preliminary explanation. It says a latent bug exposed by routine network changes caused a core connection-handling process on its servers to become overloaded. The company said it found no sign of malicious activity and that the outage was unrelated to cyberattacks or AI agents. A full root-cause analysis is due in Microsoft’s post-incident report.
What broke
Microsoft tracked the outage under incident MO1550821, with EX1550543 for the Exchange Online part. Exchange carried the heaviest impact, and anything that depends on it failed too:
- Outlook on the web, desktop and mobile lost mailbox access. API clients using Exchange REST endpoints failed as well.
- Teams calendar and scheduling features failed, because they read from Exchange.
- Copilot Chat lost access to mail and calendar data and returned errors on tasks that needed it.
- SharePoint and Defender saw degraded service in some regions, according to admin reports.
Earlier on November 12, a separate Teams incident, TM1550488, caused delayed and failed chat messages for users in the Americas for a little over an hour. Microsoft attributed it to “resource contention” in the underlying infrastructure and has not said whether it was connected to the larger outage.
User reports on Downdetector for Outlook peaked at about 129,000, around 18:00 UTC on Thursday, according to outage trackers. Organizations in dozens of countries reported problems. In the U.K., PublicTechnology reported that the disruption caused problems for officials across Whitehall.
Microsoft’s preliminary cause
According to Microsoft’s preliminary explanation, the problem started during an earlier network capacity upgrade. That upgrade introduced a latent defect, which stayed hidden until a more recent network change shifted traffic across the service.
Once traffic moved, connection requests to affected servers began to queue. The process that sets up secure connections became overwhelmed, and servers ran out of processing capacity and stopped handling requests. Some outside analyses describe the knock-on effect another way: servers that failed automated health checks were taken out of rotation, which pushed more load onto the remaining servers until they failed too. Microsoft has not confirmed that version, and its incident record still listed the root cause as pending as of November 14.
Recovery was slow because clients kept retrying. When servers came back, millions of queued Outlook and Teams clients reconnecting at once kept the load high. Microsoft’s status updates reported service “approaching normal levels” late on November 12 before full recovery the next morning, Eastern time.
Microsoft has apologized for the disruption, according to Forbes. It said a preliminary post-incident report would be published within two business days of the incident closing, with a fuller analysis after that.
Why a 30-hour recovery
Most large cloud outages are fixed within hours, often by rolling back the change that triggered them. This one took longer, and Microsoft’s preliminary account suggests why. The defect came from the earlier capacity upgrade, not the recent change that exposed it. Rolling back the recent change would not have removed it, and reversing a capacity upgrade across a global service is much slower and riskier.
Engineers instead had to relieve the overloaded connection-setup process while live traffic kept arriving. In practice, that usually means rerouting traffic, adding capacity and limiting reconnection rates so recovering servers are not hit with all their queued clients at once. Microsoft has not detailed which steps it took. Its status updates described recovery happening gradually, region by region, rather than all at once.
The incident also lasted through a full business day across Europe, the Americas and Asia-Pacific. Nearly every Microsoft 365 customer felt it during working hours, which helps explain the very high Downdetector numbers.
A pattern of change-driven failures
This is the second large Microsoft 365 incident of 2026 caused by Microsoft’s own maintenance work. The July outage that took down Teams, SharePoint and OneDrive for about three and a half hours was traced to a bug in a routine maintenance procedure in the West US Azure region.
Across the industry, most major cloud outages now come from configuration and change management rather than hardware failures or attacks. Recent examples include the Azure East US outage that knocked out several AI chatbots and, outside the hyperscalers, the CrowdStrike update failure in 2024. In each case, a change that passed testing behaved differently under real production traffic.
What admins should do
Guidance from Microsoft and managed-service providers after the incident centers on a few steps:
- Check the Service Health dashboard in the Microsoft 365 admin center for MO1550821 and EX1550543. Download the post-incident report when it is published for internal reviews and any SLA claims.
- Review SLA credit eligibility. Microsoft’s financially backed uptime commitment for Exchange Online is 99.9% per month. A 30-hour outage is far beyond that, but customers usually have to file claims themselves through the admin center.
- Look for stuck mail flow and sync issues. Some tenants reported delayed message delivery and offline Outlook caches that needed resyncing after service came back.
- Revisit continuity plans. Organizations that run email, chat, calendars and their AI assistant through a single vendor lost all of them at once.
What it means
The outage shows how much of corporate work now runs through a single control point. When Exchange Online failed, users lost more than email. Calendars, meeting scheduling and Copilot’s main features went down with it, because Microsoft has built most of its productivity stack, including its AI assistant, on the same mailbox and identity data. That integration is what Microsoft sells to customers, and it is also why one failure spread so widely.
The commercial damage to Microsoft is likely to be limited. Few large enterprises can replace Microsoft 365 quickly, and SLA credits are small next to subscription revenue. The pressure will show up in contract renewals instead. Customers will use the incident to push for stronger outage commitments, and vendors selling backup messaging, secondary email routing and independent collaboration tools will use it in their sales pitches.
Inside Microsoft, the question is change management. Two major incidents in four months, both traced to routine maintenance, suggest that testing before deployment is not catching how changes behave at production scale. Many in the industry expect hyperscalers to slow rollouts, test at smaller scale first and add protections against retry storms.
What to watch: the full post-incident report and whether it names specific fixes, such as slower deployment rings or retry throttling; any SLA credit announcements; whether regulators examine how much public-sector work depends on a single provider, starting in the U.K.; and whether Microsoft makes Copilot work without Exchange so the assistant does not go down when mail does.
Tagged
Keep reading
Chisato · · 4 min read Azure Outage Takes Down ChatGPT, Claude, and Grok
An Azure East US ingress failure knocked ChatGPT, Claude, and Grok offline at once on Sept 3 — the first outage to visibly take down three rival AI labs together.
Chisato · · 5 min read AWS us-west-2 Outage: Seattle Network Fault Returns
AWS hit its us-west-2 region again with a short connectivity fault traced to the same Seattle Metro network path as July's outage — the fourth reliability event since May.
Chisato · · 5 min read Microsoft 365 Outage July 2026: What Broke and Why
A maintenance bug in Microsoft's West US Azure region took down Teams, SharePoint, OneDrive and Copilot for hours on July 23, 2026. Here's the breakdown.