INFONOMICON

INFO NEVER AGES

Nomicon The Scene By Skylar Lehr · Nov 19, 2026

Widespread Network Panic Leaves Many Suspecting

A person who watches the status pages and the Downdetector spikes notices the clustering. Microsoft 365 goes dark for the better part of a day. The FAA grounds departures nationwide. AWS has another authentication hiccup in Northern Virginia. Banks report delays. Apple services flicker. The calendar shows a handful of days in mid-November 2026. The same few companies sit under most of the traffic. The pattern is visible. The suspicion follows.

Control Plane’s tracker logged 19 major outages involving AWS, Microsoft Azure, Google Cloud, and Cloudflare in 2026 through mid-November. That already exceeds the full-year count for 2025. Fifty-five percent of the larger cloud incidents since 2020 began with a configuration change, a software bug, or automation acting on bad input. Power and cooling failures account for another substantial share. Physical damage appears in the long-running cases. The numbers do not require a single actor.

The notable ones

Microsoft 365, November 12–13, 2026. Outlook, Teams, SharePoint, OneDrive, and Copilot failed for users across regions. Reports peaked above 130,000. Duration ran roughly 20 hours before email queues returned to normal levels. Microsoft’s preliminary review pointed to a software bug introduced in July, exposed when traffic load hit reduced server capacity during failover. The company stated there was no evidence of an attack.

FAA ground stop, around November 13. Departures halted nationwide for more than four hours. The initial assessment linked the event to a router during a copper-to-fiber upgrade that flooded the network with connection attempts. No cybersecurity indicator was reported.

AWS, May 7–8, 2026, US-East-1. A thermal event caused power loss in one availability zone. Recovery to majority function took about 21 hours 30 minutes. Coinbase and other dependent services were affected. AWS published no 2026 post-event summary at the time of later reviews.

AWS, March 2026, UAE and Bahrain regions. Drone strikes and a related fire damaged physical infrastructure. Some services remained disrupted for months. This was external physical damage, not a network intrusion.

Google Cloud. Netherlands region, July 15–16: a voltage drop on the utility feed, followed by backup and cooling issues, disrupted services for nearly 15 hours. London region, October 19–20: a power-related incident affected Compute Engine, GKE, Cloud Run, and other products for about six hours.

Cloudflare. February 20, 2026: a change unintentionally withdrew customer IP prefixes under Bring Your Own IP, lasting several hours. October 19: Cloudflare One Gateway, Access, API, and dashboard issues. November 18, 2025 (prior year, still cited): a database permissions change doubled a Bot Management feature file past a size limit and crashed core proxy traffic for several hours. Cloudflare stated the incident was not caused by a cyber attack. Smaller regional and service-specific incidents continue into November 2026.

Additional reports in the same November window included JPMorgan Chase (software bug on on-premises systems), United Airlines (technology fault affecting pilot tools), and Apple services. Several bank and Gmail spikes on Downdetector were later clarified by the companies as not reflecting broad internal failures.

The coordinated-attack claim

The clustering invites the reading that someone is probing or disabling the shared layer. A few providers carry the majority of web, email, collaboration, and cloud traffic. When several of them stumble in the same week, the coincidence looks engineered. Officials and the providers themselves have not supported that reading.

U.S. National Cyber Director Sean Cairncross stated the failures were not a coordinated attack. Microsoft, Cloudflare, and others have published or preliminary statements ruling out malicious activity for the incidents they control. Root causes that have been documented are internal: a July bug meeting unexpected traffic, a router update that flooded links, a thermal event that dropped power, a voltage sag, a database permissions change that enlarged a file past a hard limit, physical damage from strikes. These are the ordinary failure modes of large systems that change frequently and that concentrate load.

A coordinated campaign would leave common indicators: shared malware families, overlapping command-and-control, simultaneous exploitation of the same vulnerability across independent operators, or consistent forensic artifacts. None of the public postmortems or official statements have described those. Separate failures, some years in the making (the July bug), some minutes old (a bad change), produce the same user-facing symptom: the service is unreachable. The symptom is shared. The causes are not.

What remains visible

The concentration is real. A change at Cloudflare, AWS, or Microsoft can darken large fractions of the public internet and private tools at once. Status pages lag. Postmortems arrive days or weeks later, or not at all. Power and cooling events still take regions offline for many hours. The systems were not built for the current density of dependence, and the rate of change exceeds the rate at which edge cases are tested. Those facts are observable without requiring a single directing hand.

The mid-November cluster produced the suspicion. The individual records show power, heat, configuration, and software. The records do not show a coordinated cyber attack.

Permanent link: https://nomicon.info/article.php?slug=widespread-network-panic-leaves-many-suspecting