Summary
Fireship’s Code Report breaks down the Microsoft 365 outage that began on a Thursday in mid-November 2026, when Exchange Online went down for roughly 36 hours and took Copilot, SharePoint, and other services with it. Reports came in from 74 countries, and the admin center failed too, leaving administrators without the tools to diagnose the problem.
The video explains how load balancers and health checks can create cascading failures: when many servers fail at once, the survivors are overloaded, fail their own checks, and are pulled from rotation as well. Microsoft’s attempts to restart servers failed because returning traffic overwhelmed them again. The company later said the incident stemmed from a network capacity upgrade that exposed a latent bug in the system handling secure connections.
It also covers the FAA’s nationwide halt of departures on Friday, which caused more than 9,000 cancellations, and the competing explanations offered by officials, including a router swapped out by a defense contractor. Officials and Microsoft say the two events are unrelated and that AI was not responsible, a claim the video examines in the context of wider concerns about software quality and AI. Viewers get a clear technical explanation of how a routine change can bring down critical systems.
📺 Source: Fireship · Published November 16, 2026
🏷️ Format: News Analysis







