Between Thursday 12 and Tuesday 17 November 2026, Microsoft 365, the FAA, AWS, Chase, United Airlines, Apple and the NSA’s classified network all had major outages. Every official who has spoken says it was not an attack. This post does not claim otherwise. It asks something narrower and checkable: how much has each organization actually explained about what broke and how it was fixed?
Read this first. Facts as of 19 November 2026. I tag each claim: [Official] said on record by the organization or a named official, [Reported] from press or anonymous sources, [Inferred] my own reasoning, and [Speculation] theories that circulate online without evidence. Sources are at the bottom. I have deliberately kept the speculation out of the facts.
The cluster
- Thu 12 Nov, Microsoft 365. About 24 hours of impact. See our post on the Microsoft 365 outage.
- Fri 13 Nov, FAA. Nationwide ground stop after a new router was installed. See our post on the FAA ground stop.
- Sun 15 Nov, AWS us-east-1. 3 hours 16 minutes, 11 services, sign-in and sign-up failures. [Reported]
- Mon 16 Nov, Chase, then United. A software release slowed Chase for hours; that evening United lost aircraft weight data. See our post on the Chase outage.
- Tue 17 Nov, Apple and the NSA. Apple account services failed on an “internal update”; NSANet was degraded and staff arrived in staggered shifts. [Reported]
What each organization has explained
| Organization | Stated cause | Technical detail published | How it was fixed | Still unexplained |
|---|---|---|---|---|
| Microsoft | Latent Windows lock-contention bug from July, plus a Teams failover and ~14% capacity already out [Official] | Detailed preliminary report; final due 20 Nov | Known-good code path restored and rolled out in large batches [Official] | The “separate issue” that broke regional bulkheads; why Exchange cannot fail over |
| FAA | A new router at El Dorado, Arkansas; root cause still undetermined [Official] | Partial: a flooding router, a Juniper feature switched off | Two sites were cut off from the network [Reported] | The exact traffic, the configuration, the Amarillo role |
| AWS | None. No root cause published as of 18 Nov [Reported] | Status page entries only | Not stated | Everything |
| Chase | “A software bug in on-premise hardware systems” [Official]; an FBI official linked it to a release that morning [Reported] | None beyond that sentence | Not stated; transactions normal by evening | Which component, why on a Monday morning, rollback or forward fix |
| United | None. The airline has not said what failed [Reported] | Pilots’ union described the symptoms: crew messaging and final weight data | Airline said it was resolved around 6 PM ET | The cause and which system failed |
| Apple | An unspecified “internal update” [Official] | None in the statements reviewed | Not stated; Apple said “many” systems were back | Which system, exact times, per-service recovery |
| NSA | “Temporary degradation during scheduled routine maintenance” [Official]; the DNI later tied it to patching for vulnerabilities found last month [Official] | None on which systems or how many staff | Not stated | Why it was reportedly unstable for weeks [Reported]; scope and duration |
Read the table by column. Only one of seven organizations has a real technical account, and even that one is preliminary. Two have offered no cause at all. [Inferred]
What officials have said, and what they have not
The DNI said the outages were “separate service outages associated with maintenance and technical changes” and “not a coordinated or state-sponsored attack”. The FBI’s deputy director said there was “no indication whatsoever” of malicious cyber activity, and Secretary of State Rubio said there was “zero evidence” the incidents were related. [Official] Homeland Security Secretary Mullin first said on Fox that a tech contact blamed AI speed, then told CNBC there was “no cyber threat that’s taking place”. [Official] None of this is a technical explanation. “It was not an attack” answers one question. It does not answer what broke, why it broke now, or how it was fixed. [Inferred]
Is a cluster like this new?
No. In October and November 2025, AWS, Azure and Cloudflare failed within weeks of each other. The causes were different and all internal: a latent DNS race condition at AWS, a configuration change at Azure Front Door, and an oversized feature file at Cloudflare. [Reported] One tracker of major cloud outages since 2020 finds that a new configuration, a deployment or automation acting on bad input caused 54% of them. [Reported] The same-week FAA, Microsoft and Chase events fit that pattern. What is different this time is the number of organizations in six days, the presence of a government network, and the absence of explanations. [Inferred]
Four hypotheses, side by side
1. Independent changes, bad luck (the official line)
Each organization had its own internal change and its own bug. The causes stated so far are different: a July Windows bug, a router, a bank release, a backup test at Workday, an internal update. Large systems fail through changes all the time, and counting six in a week depends on which outages you include. [Inferred] Fits: every cause published so far. Does not explain: why so many landed in one week, or why three have no cause at all.
2. A wave of emergency patching
This is the best-supported common factor. VulnCheck counts about 100,000 disclosed CVEs in 2026. [Reported] November’s Microsoft Patch Tuesday had 68 critical flaws, and F5 issued an out-of-band NGINX fix on 18 November for two critical bugs found with AI. [Reported] The PHP flaws found with AI tools in October are now tied by officials to the NSA outage, and the DNI said the patching “caused some of these systems to not come back up”. [Official + Reported] Fits: the NSA case directly, and the Chase and Apple “release” and “update” wording loosely. Does not explain: Microsoft, whose bug was shipped in July, or the FAA, whose cause was a router. Also, one source told the New York Times that NSANet had been unstable for weeks, which does not fit a patch applied this week. [Reported]
3. Shared fragility rather than a shared cause
The Cloud Security Alliance calls this correlated fragility: heavily coupled systems, common vendors and common practices that fail in similar ways without a common trigger. [Reported] Microsoft’s own report shows it: a failover plus missing capacity plus a latent bug. [Official] Fits: the shape of Microsoft, FAA and Chase, all “change meets coupling”. Does not explain: the timing, which this hypothesis treats as chance.
4. A coordinated attack or a hidden common cause
Some commentators say the cluster looks too tight to be chance; one senator called it the “greatest coincidence in the history of the world”. [Reported] [Speculation] The FBI, the DNI, the Secretary of State and the DHS Secretary all say they see no sign of an attack or a link. [Official] I found no technical indicator, no shared vendor and no common indicator of compromise that supports it. That does not prove it false, because absence of evidence is not proof of absence, and none of these organizations has published enough detail for outsiders to rule it out. [Inferred] That last point is the real issue of this post.
Why the missing detail matters
- Trust. Saying “not an attack” and “maintenance” without a mechanism asks the public to take it on faith, which fuels exactly the theories the officials want to stop. [Inferred]
- Everyone else’s defense. The Microsoft report is valuable because other engineers can check their own fleets for lock contention on many-core hardware. An unexplained AWS outage teaches no one anything. [Inferred]
- The standard already exists. Microsoft published a timeline with factors and fixes; Meta, Cloudflare and AWS have done the same for past incidents, including the AWS DNS race condition in 2025. [Reported] Detailed post-mortems are normal practice for infrastructure at this scale.
- Ruling things out. A technical account is how an outsider can separate “unrelated change” from “common cause”. Without it, nobody can.
Open questions, organization by organization
- AWS: What failed in us-east-1 on 15 November, in sign-in and network load balancers, and was it a change? When will a summary appear?
- United: Which system produced final weights and crew messaging, was it internal or a vendor, and what was the fix?
- Apple: What was the internal update, which identity or account service did it touch, and why did recovery take most of a day?
- NSA: Which systems were affected and for how long? If the cause was patching for the PHP flaws, why were reports of instability going back weeks?
- Chase: What did the release change, and was there a canary?
- FAA: What traffic was the router sending, and what is the Juniper feature?
- Everyone: Was any common vendor, library, certificate, patch or cloud region involved in more than one of these?
What would change my mind
- A common vendor, software component or patch identified in more than one incident.
- Technical indicators of compromise shared by independent investigators, not only official denials.
- Post-mortems from AWS, United, Apple or the NSA that contradict the stated “internal change” explanations.
Conversely, detailed post-mortems that show independent, ordinary causes would settle the question the right way: with evidence. Microsoft’s final report is due on 20 November.
Sources
- Cloud Security Alliance: The November Outage Cluster, Correlated Fragility and Attribution
- MLQ.ai: A week of outages and what is known about the causes
- StrongMocha: Why US tech outages became a concern
- The Record: NSA Net faces unspecified disruptions
- CNN: Highly classified US national security communications network goes down
- HNGN: NSA network still down for a second day
- BNN Bloomberg: Surge in software patches likely behind recent tech outages
- The Register: AI found more critical bugs in NGINX
- SANS ISC: Microsoft Patch Tuesday November 2026
- DEV: AWS us-east-1 outage on 15 November 2026
- NBC News: United says it has resolved a widespread system outage
- MacRumors: Apple blames internal update for outage
- SnowRock: What Apple has said
- Applescoop: Every major outage of November 2026
- Cybernews: The pattern behind the 2025 outage cluster
- Control Plane: Cloud outage tracker since 2020
- Our earlier posts: Microsoft 365, FAA, Chase
Facts as of 19 November 2026. Several investigations are ongoing, Microsoft’s final report is due on 20 November, and some details here may change.
