Rock-a-bye baby, on the tree top,
When the wind blows, the cradle will rock.
When the bough breaks, the cradle will fall,
And down will come baby, cradle and all.
Traditional nursery rhyme
On Friday afternoon, every departing flight in the United States was held on the ground. For about five hours the FAA kept aircraft parked because it could not trust its own systems. The cause, as officials first described it, was not old equipment failing on its own. It was new technology added to a system still running on 1980s copper lines.
It would be unfair to pick on America. In September, NATS, the UK’s air traffic service, lost its flight processing system for three hours, and more than 1,750 flights were cancelled. In March last year, a fire at the North Hyde substation shut Heathrow, Europe’s busiest airport, for most of a day. The fault behind it had been found almost seven years earlier and never fixed.
A month later, Spain and Portugal went dark in Europe’s worst grid incident in more than twenty years. That September, a cyber attack stopped Jaguar Land Rover’s production lines for over a month. And behind all of them sits the Post Office Horizon scandal, where an institution’s blind faith in its own computer outlasted years of evidence that it was wrong.
The week of the ground stop brought a different kind of failure. The founder of Nvidia, the company at the centre of the AI boom, became embroiled in a scandal and resigned, and the company lost 13 per cent of its value in five days. The chips were not the problem. The company rested on things the story left out: customers borrowing to buy, voters who had just approved measures that could slow data centre building, and one leader who had run the company for 33 years with no successor in waiting. A structure built around a single person, and a critical point of failure.
These events have different causes: a failed component in a substation, a timing clash measured in milliseconds, stolen credentials, a stretched valuation, and wilful blindness.
What they share is a gap between the story told about a system and the load its foundations can carry. Attention follows the story, not the foundations it is built on. Markets reward the story, and under pressure we tend to make it bigger rather than check what it stands on.
The Germans have a word for the feeling this produces in us, if we take note: Fingerspitzengefühl, a feeling in the fingertips, the sense of a situation before the reports arrive. Mine says something is shifting.
The organisations that measure outages find they are not becoming more frequent. What has changed is how far they reach. Lightning at Luton in August stopped Luton. A fault at Swanwick stopped the country. A ground stop in America held flights bound for it from across the world. Each failure now travels further, through more shared systems, and costs more.
Seismologists know that the first waves from an earthquake, the P waves, arrive before the destructive ones. They are small and fast, and they are a warning. Almost every case above had its P wave on the record. The FAA had a nationwide ground stop in 2023 and a government report in September. NATS failed in 2023 and again in 2025. The North Hyde fault was found in 2018. DeepSeek wiped $593 billion off Nvidia in a single day in January 2025. The signals were there. Mostly they were filed as lessons to be learned, and the organisations moved on.
Azeem Azhar has a name for the underlying condition: the exponential gap, the distance between how fast technology moves and how fast our institutions adapt. This month looks like that gap showing up in the foundations of our economy.
Hubris is usually treated as a flaw of character. I think it is better understood as a structural gap: a story that has run ahead of what holds it up. Humility is not modesty; it is an accurate knowledge of our dependencies. The word has its roots in humus, earth: being grounded.
Ambition has humility: it knows where the edge is. Hubris has stopped looking.
And the earth is the deepest dependency of all. Many of the stories being told are a logical extension of an industrial-era view of the worker as an element in a machine. To an investor who sees only the machine, AI is an enormously attractive story. It is more complex than that. Everything we do is supported and permitted by a planet that has little regard for machines. If we ignore that, and fail to learn the lessons now in front of us, if we treat ourselves and the planet without respect, we will face a stop of an entirely different order.
Every collapse looks like hubris in hindsight, and I have chosen examples that fit. Nvidia may report strong results on Tuesday, and the lessons may be glossed over, for now. Some of what I call hubris is plain neglect. Stories always have to run ahead of reality, or nothing gets built. The question is whether anyone is minding the gap.
So the questions for anyone leading an organisation are practical. What is the oldest thing your newest promise rests on? Which change are you making at the same time as three others? Where is your single point of failure, and is it a person? And who has the authority to call a ground stop?
And for ourselves, a simple question: where have we come from, and where are we going?
If we sit quietly, we can hear the bough creaking. The cradle does not fall because of the wind. It falls because no one checked the branch.
Have we not always failed to learn the lesson? And what might it take for us to be able?