The shifts aren't shifts

I sat down last week to map my last working decade. Ten years isn’t a thought experiment for me. It’s the distance to my pension, if the rules don’t change, and everything I want to have when I stop working has to be earned before then. So “how will people make money in ten years” stopped being futurology and became planning. And almost everything written about it answers a different question: which jobs AI will destroy.

I don’t think the job-loss lists are wrong. They’re at the wrong altitude. They describe the wave and say nothing about the water.

Look at the water. Agriculture gave way to industry, industry to services, services to knowledge work. Each wave changed which labor paid; none of them changed that labor paid. People kept earning from work, and the work moved. Same on the other side of the ledger: people have always also earned from owning things, and only the things changed. Farmland, then factories, then shares, then software. Two categories of income, labor and assets, and every wave so far reshuffled the contents without touching the structure.

I’ve watched a slice of this myself. I’ve built systems in telecom, banking, retail and agriculture, and in each of those sectors software ate a layer of labor. The people didn’t stop earning. They moved, mostly upward, toward work that needed more thinking.

Which leaves one thing unexplained. If the structure survived every wave, why does this one feel different, and why does knowledge work, the best-paid labor of my whole career, suddenly look exposed?

Here’s the mechanism. Knowledge work paid best for decades because cognition was the scarce input, and it could only be bought one way: by the hour, attached to a person. Every earlier machine automated muscle, so people climbed toward thinking work, and the thinking rung had endless headroom. That’s the ladder my generation was told to climb, and it worked.

AI is the first technology that turns cognition itself from labor into capital. Thinking stops being something you hire by the hour and becomes something you buy, amortize, and own. That one move explains the picture. The knowledge-work premium falls not because the work disappears but because its scarce input became a purchasable asset. The wage premium moves to whatever cheap thinking can’t replace: hands, presence, trust, a person who can be held accountable. It’s why the nurse and the plumber look better positioned for the coming decade than the analyst.

I spent my career selling thinking by the hour. Which means I spent it on the exact side of this trade that’s being repriced.

There’s a harder edge underneath. The old advice, move up into what machines can’t do, assumed there’s always a higher rung. After muscle and after cognition, what remains is thinner: presence, responsibility, being someone rather than something. The ladder strategy that carried every previous generation has less room this time.

And one boundary is dissolving. A one-person business running on AI is labor behaving like an asset. A published body of thinking is working hours turned into something that keeps producing. It has never been cheaper to convert labor into assets, and that conversion rate, not any list of safe jobs, is what actually changed.

So the question for my remaining decade is simple: which of my working hours convert into something that keeps earning after I stop.