One Software Fault Grounded a Thousand Flights, and Private Jets Were in the Same Queue

At 13:46 UK time on Tuesday 8 September, the system that processes filed flight plans for British airspace stopped working properly. By the time NATS applied a repair at around 16:40, controllers had spent nearly three hours restricting how many aircraft could depart, and the knock-on delays ran into a second day. Aviation analytics firm Cirium counted roughly 2,000 cancelled flights across the Tuesday and Wednesday.

Almost every account of that afternoon was written about airline passengers. That framing misses something structural. Business aviation files into the same flight data processing system as British Airways and Ryanair, through the same channels, subject to the same flow restrictions. When that processor degrades, a Luton departure sits in the queue on identical terms to a Heathrow one. There is no private lane.

What NATS said, and when

NATS published updates through the afternoon, still the only first-hand account of the fault.

At 13:46 it said it was investigating a technical issue "causing some disruption to flight departures." At 15:25, nearly two hours later, it narrowed the location: the issue "was in our flight processing system." A fix went in at 16:40, with NATS saying the system was "starting to recover." At 19:30 it declared the issue resolved. At 22:10 it described what had happened as "a very complex recovery" that "has created difficulties for the whole aviation network." Operations were not called stable until 12:55 the following day.

NATS has not published a technical explanation of what failed. It has ruled out a cyber attack. Beyond that, the cause is still unstated.

The counts, and where they came from

The figures vary by who was counting and when, so they are worth attributing rather than rounding into one headline.

Flightradar24 logged at least 625 cancelled services to and from UK airports on the day itself. FlightAware, measuring by mid-afternoon, counted around 490 delayed departures plus 525 delayed or cancelled arrivals across six airports: Heathrow, Gatwick, London City, Luton, Edinburgh and Bristol. Cirium's larger figure covers both days including the Wednesday recovery.

Ryanair said more than 65,000 of its passengers were delayed by up to eight hours and that it cancelled more than 50 flights. British Airways cancelled at least 16 departures from Heathrow. Two of the six airports on FlightAware's list, Luton and London City, are core business aviation gateways. They appear there for the same reason Heathrow does.

Why a single processor reaches every operator

The 2023 precedent explains the mechanism better than anything said last week.

The Civil Aviation Authority's independent review into the August 2023 bank holiday failure found the cause was one flight plan. It was technically valid. Filed by French Bee, it contained two waypoints carrying identical identifiers roughly 4,000 nautical miles apart, one near the start of the route and one near the end. The software had no rule for resolving that, so it entered a fail-safe mode and shut down automated processing rather than act on data it could not interpret. The backup ran the same software logic, so failing over reproduced the same fault. That incident produced almost 1,600 cancellations, affected around 700,000 passengers and cost the industry close to £100m.

The detail that matters for owners and charter customers: a single operator's filed flight plan, entering a shared processor, took the system down for everyone in it. Nothing about the size of an aircraft or the price of a seat changes its position in that architecture. An instrument flight plan out of Farnborough is data in the same pipeline as a plan for a 737, and when departure rates are cut, they are cut against the flow, not against the passenger list.

What a private flyer can and cannot do

Cannot: route around it. A national flight data processing fault is not a local problem to be solved with a different handler or a better FBO. While restrictions were in force, UK airspace stayed open and aircraft kept departing, but at a metered rate set centrally.

Can: recover faster than the airlines, for structural rather than preferential reasons. NATS itself identified the hard part as the recovery, not the fault, and the recovery problem was aircraft and crew ending the day out of position. A scheduled network unwinds that over days because every rotation depends on the one before it. A single aircraft with one crew has far less to unwind. The lever available is flexibility on departure time and airfield, not exemption from the restriction.

That exposure grows as utilisation does. The fleet is being flown harder than at any point since 2019, which means more movements filed into the same infrastructure. For anyone routing through British airspace on a transatlantic leg, the North Atlantic track structure makes that infrastructure hard to avoid.

What happens next

Transport Secretary Heidi Alexander called the disruption "totally unacceptable" and said "air passengers do deserve better." She has asked NATS for a report on what went wrong and separately asked the CAA, as NATS' independent regulator, to conduct its own review. Her stated reason was blunt: "we can't have NATS marking their own homework here." Reporting indicates NATS was given roughly a week for its internal account, while the CAA's full independent review is not expected until March 2027.

Ryanair's chief operating officer Neal McMahon publicly called for NATS chief executive Martin Rolfe to resign, saying "following the 2023 collapse, we were told lessons would be learned. Yet here we are again." Wizz Air said the organisation was "not fit for purpose." Airlines characterised this as the third significant failure in three years. Rolfe has said he accepts full responsibility, and Downing Street has said the government retains confidence in him. Those are attributed positions in a live dispute, not settled findings.

For business aviation, the useful outcome is not a resignation. It is whether the CAA review addresses the single point of failure its own 2023 report already identified, where a primary system and its fallback share the logic that broke. Until that is answered, the honest planning assumption is that this recurs, and that private flights will be in the queue when it does.