Microsoft 365 had a major outage on November 12, 2026. This post covers what's confirmed from IT-Connect, Astra Security and Euronews, and ends with some general takeaways that are my own and not from Microsoft.

What happened

  • Started around 17:54 UTC on November 12 and was largely resolved around 05:33 UTC on November 13, per Astra. Reports of the duration vary between about 10 and 12 hours.
  • Affected Exchange Online (Outlook on all platforms, including REST API access), Teams, Copilot, SharePoint Online, OneDrive, Intune, Purview and Defender. On-premises Exchange and SharePoint were not affected (Astra).
  • Downdetector showed nearly 130,000 Outlook reports at the peak and users in 74 countries (IT-Connect).

Microsoft's explanation

  • First: "higher than normal load" on part of its infrastructure.
  • Later: not a traffic spike. Servers became unavailable and failed health checks, and a process that establishes secure connections became overwhelmed.
  • Astra describes the remaining servers taking the extra load and failing too. That reading is Astra's interpretation of Microsoft's updates.
  • Microsoft began rolling back a recent network configuration change that matched when impact started, and said it knew of no malicious activity (IT-Connect).

Restoration

Microsoft validated an alternative health check method and gradually reintroduced servers. It warned servers could become unresponsive again under traffic (Astra). A root cause report hadn't been published when I checked.

Downstream impact

Astra says OpenAI and Anthropic connectors that depend on Microsoft APIs reported related errors. I only found that in one source.

General takeaways (mine, not Microsoft's)

  • Health checks that remove servers can raise load on the rest, so design them with care.
  • If your app depends on Exchange or Graph APIs, plan for degraded behavior and not only for failure.
  • Subscribe to your providers' status feeds and have a fallback message for users.

Sources: IT-Connect, Astra Security, Euronews