Travel tech: The computer failure
that grounded Britain

A massive air-traffic-control failure in the United Kingdom this week offered cold comfort to South Africans who have watched their own air navigation services stumble.  

A similar breakdown in the UK three years ago cost up to £100-million and produced 34 recommendations. This week’s sequel was a concise assessment of their effectiveness.

The latest fault hit the flight-processing system run by National Air Traffic Services (NATS), the company that manages most aircraft moving through British airspace. The system was unavailable for about three hours, helping to wipe more than 1,750 flights from airline schedules.

NATS restored the system within several hours, but aircraft were scattered across the wrong airports and crews ran out of permitted working hours.

For South Africans, this was a first-world failure with a familiar outcome. Our own Air Traffic and Navigation Services has spent the past two years recovering from unreliable navigation systems and a shortage of controllers.

ATNS had also failed to maintain procedures used by pilots when approaching airports in poor visibility. At Richards Bay, Upington and Bloemfontein’s Bram Fischer International airports, it was air-traffic management by prayer for clear skies.

A committee appointed by transport minister Barbara Creecy found gaps in safety governance and weaknesses in communications systems in March 2025. ATNS later approved R1-billion for modernisation, but by October controller staffing remained at 70% of requirements at OR Tambo and 82% at Cape Town International.

It should have been no surprise, then, that South Africa managed a close imitation of Britain’s crisis on 1 August 2025, when the Flight Plan Management system at OR Tambo International Airport failed and delayed departures. ATNS restored it later that day.

The British version came with a more expensive pedigree. In August 2023, a flight plan from Los Angeles to Paris contained two waypoints with the same identification code, one in Mexico and the other in Japan. The computer dealt with the ambiguity by shutting down. Its backup received the same information and joined it.

Investigators called this a “critical exception error”, which is technical language for an exception that proved exceptionally critical. More than 700,000 passengers were affected, with an estimated cost of up to £100-million.

To make matters sillier, the specialist needed to restart the system was working remotely and had to travel to the control centre for a manual restart. Before Britain could get its aircraft moving again, it first had to get an engineer through the traffic.

Nor is the issue confined to one or two countries.

The country that pioneered scheduled aviation, the United States, discovered this in January 2023, when the Federal Aviation Administration stopped every domestic departure for nearly two hours.  

A system that distributed Notices to Air Missions, which warn pilots about runway closures, hazards and other information required before a flight, had failed. It turned out that contract workers had accidentally deleted files while fixing a synchronisation problem between the main database and its backup. The corrupted information affected both. By the time departures resumed, about 1,300 flights had been cancelled and 11,000 delayed.

Then came CrowdStrike in July 2024. A faulty update from the cybersecurity company crashed Windows computers around the world, taking airline check-in and dispatch systems down with them. Airports lost access to some baggage and passenger-processing services.

Delta struggled for several days after its main US competitors had recovered because the outage also hit software used to match crews with aircraft. Computers had helped airlines build vast networks of tightly connected flights, but the same connections spread the disruption with uncanny efficiency for a system that barely gets meal orders right.

Alaska Airlines offered another demonstration in July 2025 when hardware failed at a data centre. It halted the airline’s entire operation for about three hours and caused more than 150 cancellations. Alaska later said the failed component had multiple layers of redundancy – an excellent feature until all the layers fail together.

The underlying problem is that airlines cannot simply resume a timetable when a computer system returns to service. Aircraft have to be brought back into position, airport slots have to be renegotiated and replacement crews found.

For South Africa, the UK and the USA, aviation’s dependence on ageing systems and scarce technical skills reveals itself in very public failures, but these in turn also expose an underlying rot.

* Arthur Goldstuck is CEO of World Wide Worx, editor-in-chief of Gadget.co.za and GadgetWings, and author of “The Hitchhiker’s Guide to AI – The African Edge”.