Back in 2014, 2015, 2016 I was one of those guys going around to conferences and meetups trying to get my name out there, carving out space as a professional. I would tell people: “Folks, there is this thing called machine learning, machines learn from data, so you can automate a whole set of tasks, especially the simple, repetitive ones that get done over and over, where it is easy to collect data and easy to build statistical models.” And back then it actually made sense to say something like that, because that really was what you could do. Nobody paid me any attention except a small, tight circle of slightly sharper technicians and a few enlightened entrepreneurs.
Then I remember it well: machine learning, then deep learning, neural networks, then generative AI, then ChatGPT, then agents, then Claude Code. And now the situation is very different.
Automation now hits cognition, not repetition
It is different because, first of all, automation is no longer about repetitive tasks, the way everyone expected. Automation is about cognition. In 2026 you are standing in front of machines that replace you, augment you, whatever you want to call it - machines that fundamentally think, or behave like an entity that thinks. Who cares whether they really think. The fact is they make decisions, and what they automate is a long way from being simple and repetitive, because they automate writing reports, scientific research, resource management. They can genuinely do anything.
And this changes the game completely. When automation touches cognition in general, you do not get what you got in past years. Think of the old highway toll booths: the little machine arrives, it takes your credit card, it does the calculation, it handles the transaction, it gives you your change, and one person keeps an eye on five booths where before there were five people, one per booth, slowly handing people their tickets. That kind of automation was tied to something simple and repetitive, and it was automation for that one thing, then another, then another.
The automation we are witnessing today is not only automation of cognition in general, it is also across the board. It is not that agents now automate a single job or one specific task. They take a huge pile of everyday activities off the table, at 360 degrees, and often these are automations that are more intelligent than the people themselves - even if many struggle to admit it. They are far more intelligent than any person at a whole range of things.
This is not another iPhone
I keep hammering on this theme because I have trouble getting it across even to the people I love most, the ones I have practically thrown myself at, going “look at this, look at this.” Listen, this is not like before. This is not a simple new gadget. This stuff is disruptive. It changes habits, it changes the economy. In my opinion it changes the very way we see ourselves as human beings and the meaning of being on this planet. This thing is enormous.
Look at the numbers on infrastructure investment. Why all these data centers? Why the whole world? Why so big? Why so energy hungry? They are basically rebuilding the equivalent of the railways, the water grid. The network of data centers is the infrastructure for this stuff. It concerns hundreds of millions of people, at least. And I cannot get people to understand how big this is. Why is it talked about so superficially? Why is politics asleep? Why is everything treated as if, well, another iPhone showed up. No, the iPhone did not show up. A beast showed up. Intelligence as a commodity showed up. A new mode showed up in which whoever has infrastructure, and therefore compute, takes it all.
The winner-takes-all reality
The nice story goes: we finally reach full automation, we all work less, some form of universal basic income or similar, we all go part time, I guess, because the machine does most of the things. All very beautiful, and maybe in the long run it even happens, but I would not bet a single cent on it.
What actually happens is that whoever has the infrastructure, whoever owns energy, data, and the power tied to deciding who can do what and where - because that becomes a bureaucracy too - takes it all. Here it is winner takes all, folks. And the rest of us are left scrambling.
