Mark Zuckerberg just published a long essay laying out his vision for what he calls personal superintelligence. He is not the first. In the past two years, Sam Altman, Dario Amodei, Marc Andreessen, and Leopold Aschenbrenner have all written similar essays, each running to thousands of words, each making a similar case: a massive AI transformation of the economy is coming, and it cannot be stopped.
That pattern is worth pausing on. These are not outsiders. They run some of the richest companies on the planet, control huge amounts of money, and have direct lines to world governments. Historically, long persuasive essays like these have been written by people trying to convince others of something they do not yet believe. Powerful people do not usually need to convince anyone of anything.
So why are they doing it? Because there is one thing they still cannot control: whether businesses actually buy what they are selling. Read the essays closely and they all lean on the same idea, that a huge AI buildout across the whole economy is simply inevitable. That framing matters, because if something is inevitable, questioning it looks pointless. If it is a choice, it can be negotiated, slowed down, or rejected.
There is a real difference worth noticing here. It is true that AI tools already in use, like writing assistants or customer service systems, are not going away. Companies using them now will keep using them. But that is a small claim. The bigger claim in these essays, that trillions of dollars in infrastructure and a fully AI-run economy are unavoidable, is not the same thing. That version needs money, electricity, computer chips, favorable rules, and paying customers to actually happen.
Nuclear power offers a useful comparison. In the 1960s, US regulators expected over a thousand nuclear reactors to be running in the country by the year 2000. The real number came in at a small fraction of that, and stands at roughly 96 today. Nuclear reactors work fine as technology. What limited their spread was cost, public worry, and political decisions, not physics.
This is not just an academic point. The biggest AI companies have committed hundreds of billions of dollars a year to building data centers, a bet that only pays off if businesses keep buying more AI, on schedule. That means the real decision about how fast and how far AI spreads through the economy is not being made by these essay writers, or even by regulators. It is being made inside ordinary companies, one procurement decision at a time.
None of this is an argument against using AI. Plenty of it already works well and is worth paying for. But business leaders should notice that adopting it on a vendor's timeline, under a vendor's terms, is itself a choice with consequences, not something owed to fate. The pace belongs to the people signing the contracts, not the people writing the essays.