> ## Content Index
> Fetch the complete content index at: https://www.lacaribenanews.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Twice in Five Days, the Thing That Spread the Failure Was the Backup. Nobody in the Caribbean Has Said What the Week Cost.
- URL: https://www.lacaribenanews.com/microsoft-faa-chase-united-outages-caribbean-latin-america-exposure-2026/
- Published: 2026-11-17T00:33:19.000Z
- Updated: 2026-11-19T21:12:41.000Z
- Description: Microsoft's servers failed on 12 November. The FAA stopped every US departure on the 13th. Chase and United followed on the 16th. In the two cases we can see inside, the fault spread through the systems built to stop it. Argentina's banks published an all-clear. No Caribbean body said anything.
- Author: Theon Alleyne
- Tags: Business, Caribbean, Latin America, technology, Aviation, Banking, Beat: economics, MSME: indirect

**Quick summary:** Microsoft's servers failed on 12 November. The FAA stopped every US departure on the 13th. Chase and United followed on the 16th. In the two cases we can see inside, the fault spread through the systems built to stop it. Argentina's banks published an all-clear. No Caribbean body said anything.

By LCN Business Desk

On 12 November 2026, Microsoft's traffic managers did exactly what they are built to do. They saw servers failing and moved users onto healthy ones. Within two hours more than 80 per cent of the fleet had stopped responding.

"The redirected traffic increased the likelihood for the same problem to occur on the remaining endpoints," the company's [preliminary report](https://www.365fanatic.com/wp-content/uploads/2026/11/PostIncidentReport.pdf?ref=lacaribenanews.com) says. Every healthy server it moved users to met the same defect and stopped answering in turn.

The safeguard carried the failure. So did the one before it: Teams had been moved onto the gateway that carries Exchange about fifteen hours earlier, which the report calls "part of normal operations and integral to standard Microsoft resiliency efforts."

## Did the same thing happen at the FAA?

The flood started at new equipment, installed during routine work.

The Federal Aviation Administration's [questions and answers](https://www.faa.gov/newsroom/nationwide-ground-stop-questions-and-answers?ref=lacaribenanews.com) of 15 November describe data hitting its network at 1,000 connection attempts every four to five seconds. Seventeen of its 22 control centres became unstable. The FAA names its provider, L3Harris, and a new router at a radio site in El Dorado, Arkansas.

The work was replacing ageing copper with fibre. The remedy was to switch off the same function on about 60 similar routers. About 3,500 aircraft landed safely. Some 12,462 flights were cancelled that Friday, 9,082 of them by major airlines.

The two faults are unrelated and the shape is identical. The upgrade and the failover were the way in. In both, resilience work carried the fault.

## So were the four outages one event?

No. That is what makes the shape worth noticing.

Microsoft's defect sat in a fix "prepared specifically for Microsoft 365 networking gateway servers," and the report says "neither the underlying defect nor the fix applies to broader Windows deployments." It closed three days before Chase's fault began. Chase blames "a software bug in on-premise hardware systems." United Airlines says there is "no indication that this was a cybersecurity-related issue," and none that its outage was linked to the FAA's. No operator names a cyberattack. No two name a shared supplier, though in a separate incident [one script reached the pages of 178 organisations](https://www.lacaribenanews.com/optimizely-script-178-organisations-caribbean-cirt-silence-2026/).

Mike Chapple teaches information technology at the University of Notre Dame. He told [The New York Times](https://www.nytimes.com/2026/11/16/business/chase-bank-outage-software-bug.html?ref=lacaribenanews.com): "each one has an explanation that makes sense on its own, but they all point to a common theme: the complexity and fragility of the systems that we depend on every day."

## Why did one company's fault go global?

Microsoft says its barriers failed, and will not yet say why.

Its gateway is built in units grouped into regions. "Due to a separate issue, the normal bulkheads between these different partitions did not stop the cascading failure," the report says. "This is what allowed the impact to spread across our network gateway infrastructure globally."

The separate issue is never described. A regional fault became a worldwide one because the barriers gave way, and the explanation runs to four words. **The final report is due on 20 November.**

Chase has named no vendor, no platform and no component, unlike the French bank where [hackers used a supplier platform to send fake transfers](https://www.lacaribenanews.com/credit-mutuel-arkea-100-million-euro-supplier-breach-2026/). Transport Secretary Sean Duffy said the FAA was still working out whether the cause was "incompetence" by the contractor or "a faulty router," [The Guardian reported](https://www.theguardian.com/business/2026/nov/16/us-bank-outage-chase-wells-fargo?ref=lacaribenanews.com).

## Who told businesses in this region anything?

One banking chamber in Argentina.

Argentina's banking chamber, the Asociación de Bancos Argentinos (ADEBA), speaks for Argentine-owned banks. It told [La Nación](https://www.lanacion.com.ar/estados-unidos/reportan-fallas-en-los-principales-bancos-de-estados-unidos-miles-de-usuarios-denuncian-problemas-nid16112026/?ref=lacaribenanews.com) on the Monday that *"las entidades nucleadas en la cámara no tuvieron problemas en sus sistemas ni en la prestación de servicios"*. The banks in the chamber had no trouble with their systems or their services, it said. It checked its members and published while the American figures were still rising.

User reports against Citi climbed past 100 that afternoon. Chase's passed 4,800\. Shares in Chase, Citigroup, Bank of America and Wells Fargo each fell about 2 per cent. Citigroup said nothing all day.

No Caribbean bank has published a statement on any of the four events. Neither has any bankers' association, airline, airport authority or central bank in the region. [LCN reported on 14 November](https://www.lacaribenanews.com/faa-ground-stop-caribbean-canada-flights-us-airspace-transit-2026/) that Air Canada warned flights between Canada, the Caribbean and South America might be cut if they crossed US airspace. Nobody published which ones were.

## What is a regional firm actually exposed to?

Airspace, a correspondent bank in New York, and a gateway with no local version.

In aviation the test was the airspace a flight crossed, not the airport it used. [Twenty-five flights went at Dominican airports](https://www.lacaribenanews.com/dominican-republic-25-cancelled-flights-air-transat-quebec-faa-ground-stop-2026/) on the Friday and Saturday. Two were Air Transat services between Puerto Plata and Quebec, which touch no American airport. What a cancellation costs depends on how many carriers fly the route: [United is the only airline flying Georgetown to Houston](https://www.lacaribenanews.com/united-georgetown-houston-sole-carrier-expansion-no-americas-2026/). FlightAware counted 418 United delays by late afternoon, and nothing published shows that route among them.

In payments it runs through correspondents. Scotiabank Bahamas and Scotiabank Cayman tell customers to route incoming US dollars through JPMorgan Chase in New York. Republic Bank in Trinidad and Tobago lists Chase among its US-dollar correspondents, as does FirstCaribbean International Bank and Trust Company in Cayman. Chase says other banks were hit "only to the extent that they were a party to a transaction with us," which is the position a regional bank's customer occupies.

A Guyanese or Trinidadian firm on Microsoft 365 sits behind the same global gateway as everyone else. The bulkheads meant to hold one region's fault inside that region are the ones that gave way.

## What should a business do this week?

Reconcile the week, and change the question it asks a supplier.

Check any payment made or expected between 12 and 16 November. Confirm it as pending, rejected or completed before repeating the instruction, keep the transfer reference, and ask which correspondent carried it. A local bank's app can work perfectly while a dollar payment sits inside a correspondent in New York.

For travel, price the airspace and the number of carriers, not the destination. A passenger on a route served by three airlines loses a flight. On a route served by one, they lose the route.

All four of these organisations had failover. Asking a supplier whether it has redundancy no longer separates anyone. Ask what happened the last time the redundancy fired, and who was told.

### The entities

\- **Microsoft**: published a preliminary report on 15 November and owes a final one on 20 November. - **Federal Aviation Administration**: the United States air regulator, which stopped all departures on 13 November. - **L3Harris**: the FAA's network provider, named by the agency, with fault not yet settled. - **Sean Duffy**: United States Secretary of Transportation. - **JPMorgan Chase**: the largest American bank, with more than 86 million customers and about 5,000 branches. - **United Airlines**: held its own departures on 16 November, and is the only carrier between Georgetown and Houston. - **Asociación de Bancos Argentinos (ADEBA)**: the Argentine banking chamber, which said publicly that its members were unaffected. - **Mike Chapple**: professor of information technology at the University of Notre Dame.

### What the title card says

\- Microsoft at its worst: more than 80 per cent of gateway servers unresponsive. - How the fault travelled: onto the healthy servers its own traffic managers moved users to. - The FAA's flood: 1,000 connection attempts every four to five seconds, 17 of 22 centres unstable. - Where it started: a new router installed during routine work to replace ageing copper. - Microsoft's reason its regional barriers failed: "a separate issue," never described. - Public all-clears reaching this region: one, from ADEBA in Argentina.

## Frequently asked questions

Were these four outages connected?

No shared supplier or common cause has been found. Microsoft traces its outage to a defect in a fix written for its own gateway servers, and closed it three days before the Chase fault. The FAA blames data flooding its own network after a router was replaced. Chase names a bug in its own on-premise hardware. United says there is no sign its outage was linked to the FAA's.

What does it mean that the backup spread the failure?

Microsoft's systems move users off failing servers onto working ones, which is normally what keeps a service up. Here the moved traffic carried the same fault onto each server it reached. At the FAA, the flood began at equipment newly installed during routine upgrade work. In both, the safeguard was the path.

Was any of it a cyberattack?

Every operator says no. The FAA states it has "no indication that this was a cyberattack or an outside nefarious actor." Chase says its problem was not cyber-related. United says there is no indication of a cybersecurity issue.

Were Caribbean banks or airlines affected?

No Caribbean institution has been named as affected, and none has published a statement either way. In payments the exposure runs through correspondent banks in New York rather than shared software. In aviation it ran through US airspace rather than US airports.

What is still outstanding?

Microsoft's final report, due 20 November, which should say why the barriers between its regions failed. Chase has not named the vendor or the component. The FAA has not settled whether the cause was contractor error or faulty equipment.