Global Microsoft 365 outage knocks out Outlook, Teams and Copilot for a day
Microsoft 365 went dark for millions of users on Thursday, November 12, as a network fault took down Outlook, Teams scheduling and Copilot for roughly a day. Microsoft says it was not a cyberattack — and service was fully restored by early Saturday.
- The scale: Outlook and Exchange Online were hit hardest; Downdetector logged about 130,000 Outlook problem reports at the peak, from users in 74 countries.
- What broke: Copilot Chat would not load or reach email, calendars or files; Teams calendar scheduling, invites and events failed; Defender Safe Links and Purview checks were degraded; Graph, Planner, Intune and Power BI were also affected.
- The cause: a fault from an earlier network capacity upgrade surfaced after a recent change; servers became unexpectedly unavailable and failed health checks, and a process that establishes secure connections became overwhelmed.
- The timeline: the incident opened at 16:17 UTC on Thursday, November 12; Microsoft declared it fixed at 03:51 UTC on Saturday, November 14, with the end of impact at 16:10 UTC on Friday.
- Not an attack: Microsoft said repeatedly there was no indication of malicious activity, and Copilot chief Charles Lamanna ruled out a cyberattack; a full post-incident report is promised.

Microsoft 365 went dark for millions of users on Thursday, November 12, when a fault in Microsoft's network infrastructure took down Outlook, Teams scheduling and Microsoft 365 Copilot. The disruption stretched into Friday, with mailboxes unreachable and Copilot silent for users in 74 countries — and it ended the way most big cloud outages do, with a configuration change rolled back and servers restarted one batch at a time.
Microsoft was quick on one point: this was not a cyberattack. The company traced the failure to a fault introduced during an earlier network capacity upgrade, which surfaced after a more recent change. Service was fully restored by early Saturday, November 14, and Microsoft has promised a full post-incident report once its review is complete.
What went down#
Exchange Online was the centre of the failure, and because so much of Microsoft 365 leans on it, the damage spread fast. Outlook users could not open their mailboxes on the web, the desktop app or mobile mail, and email delivery backed up behind them. Downdetector logged about 130,000 Outlook problem reports at the peak, from users in 74 countries.
| Service | What users saw |
|---|---|
| Outlook and Exchange Online | No mailbox access on web, desktop or mobile; email delivery delayed |
| Microsoft 365 Copilot | Copilot Chat would not load or answer; could not reach email, calendars or files |
| Teams | Calendar scheduling, meeting invites and events failed; calls, chats and meetings mostly kept working |
| Defender and Purview | Safe Links failed to open; portals would not load; data-loss-prevention checks delayed |
| Graph, Planner, Intune, Power BI | Listed as affected; mail, calendar and contact requests timed out |
The shape of the outage is the tell: services that depend on Exchange Online broke, and parts that do not kept running. Files stored directly in SharePoint stayed reachable, and most Teams calls and chats carried on — which is why many offices spent the day on the phone instead of in their inboxes.

A fault from an earlier upgrade#
Microsoft's own updates tell the story of a recovery that kept slipping. The incident opened at 16:17 UTC on Thursday, when users first reported they could not reach their mailboxes. Within two hours Microsoft had opened a second, wider incident covering every service that depends on Exchange Online, and by the evening it was clear the problem was not a surge in traffic: a number of servers had become unexpectedly unavailable, pushing their work onto the servers that were left.
The servers that remained started failing automated health checks and were pulled out of service. A process on the affected machines that establishes secure connections became overwhelmed, and the machines stopped processing requests. Engineers tested a workaround and restarted servers with a fix applied — and watched them go unresponsive again as soon as traffic returned. At 22:00 UTC on Thursday, Microsoft admitted its actions so far had not provided the anticipated relief.
The fix was unglamorous: revert a recent network configuration change, rebuild some servers on a previous software configuration, limit traffic to the damaged infrastructure, and bring capacity back in stages. By Friday morning the company saw early signs of stabilization and extended the email timeout window to 48 hours so queued messages would arrive instead of bouncing. Microsoft's final update, at 03:51 UTC on Saturday, November 14, declared the incident fixed and put the end of impact at 16:10 UTC on Friday — roughly a day after the first reports.
Not a cyberattack#
With email and security tooling down at the same time, the obvious question was whether someone had done this on purpose. Microsoft answered it repeatedly through the outage: there was, in its words, 'no indication this is the result of malicious activity.' Charles Lamanna, Microsoft's executive vice president for Copilot, said the same publicly. The company also warned users not to reset passwords or change security settings because of the outage — and to treat any unexpected sign-in or 'verify your account' prompt with suspicion, since outages are prime phishing season.
The outage was not an attack — but with Defender portals down, Safe Links failing and Purview checks delayed, it temporarily weakened the defences organizations rely on to spot one.

Why one fault took down so much#
The outage is also a concentration story. One shared dependency — Exchange Online — sits underneath email, calendars, Copilot's access to work data, Teams scheduling and a slice of the security stack. When it failed, everything built on it failed with it, across 74 countries at once. Governments in Singapore and Japan reported disruptions, and courts, airlines and AI providers were among those affected.
Regulators have started to notice the pattern. Australia's prudential regulator has warned that financial institutions relying on the same providers can lose access to critical services at the same time, and plans its first stress tests for AI risks in 2027, according to ABC News. The Microsoft outage is the case study in miniature: not an AI failure, but a failure of the plumbing AI tools now depend on. Copilot cannot reach your calendar if the calendar is down.
What to watch#
Three things. First, the post-incident report: Microsoft has promised a full account of the root cause once its review is complete, and the detail that matters is how a fault from an earlier upgrade survived long enough to surface later. Second, the email backlog: bounced messages are safe to resend now, and messages that never left the outbox should have sent on their own once Outlook reconnected. Third, the wider question: how many critical tools can sit on one shared dependency before a single fault becomes a company-wide outage.
Sources: 4MIPS — Microsoft 365 Outage: What's Down and What You Can Do (Nov. 12–14, 2026); Azure Blog — Microsoft 365 outage, 12 November 2026; Acorn Technology — Microsoft 365 Outage: Why It Happened & What to Do Now; Swiss Financial Times — Microsoft 365 Outage Disrupts Outlook and Teams in 74 Countries; EveryDev — Weekly AI Dev News Digest: November 8 – November 13, 2026.
Frequently asked questions
Was the Microsoft 365 outage a cyberattack?
No. Microsoft said repeatedly during the incident that there was no indication of malicious activity, and Copilot chief Charles Lamanna ruled out a cyberattack. The company traced the failure to a fault from an earlier network capacity upgrade that surfaced after a recent change.
Which services were affected?
Outlook and Exchange Online were hit hardest — no mailbox access and delayed email. Microsoft 365 Copilot could not load or reach email, calendars or files; Teams calendar scheduling, invites and events failed; and Defender, Purview, Graph, Planner, Intune and Power BI were also listed as affected. Most Teams calls and chats kept working.
Is everything back now?
Yes. Microsoft declared the incident fixed at 03:51 UTC on Saturday, November 14, 2026, with services fully recovered and the email backlog processed. A full post-incident report is promised once Microsoft’s review is complete.