The Invisible Trillion

Who Decides

Where the Decisions Actually Get Made

If you drew an organization chart of the people who decide what the chip and AI industry does — what gets built, where, powered by what, sold to whom, under what restrictions — it would fit on a single page. That is one of the strangest facts about the modern world. The infrastructure that carries most of the global economy, runs most scientific research, and touches the daily life of nearly everyone alive is shaped by the decisions of maybe a few hundred people, most of whom have never been elected to anything.

The sentence is true, and on its own it’s misleading, because it suggests that naming those people is the same as understanding how the decisions get made. It isn’t. Knowing the CEOs is the civics-class version of the story. The version you need — if you ever want to predict an outcome, or change one — is about the mechanisms the decisions travel through: the legal instruments, the financial structures, the approval processes that determine which choices reach which people, and which people have any say at all.

It’s worth having the cast in your head first, though, because they’re real and they recur.

A handful of companies make the chips. TSMC in Taiwan makes most of the world’s advanced ones; Samsung in South Korea is the nearest rival; Intel in the United States is trying to climb back after falling behind. Between them they produce essentially every leading-edge logic chip on Earth, which means their chief executives have more influence over what computing looks like in 2030 than all but a few elected officials anywhere.

None of them can operate without ASML, the Dutch company that makes the extreme-ultraviolet lithography machines the advanced fabs depend on — there’s a waiting list, and who gets the next machines is its own quiet decision. Nvidia, under Jensen Huang, designs the GPUs most AI training runs on, one of the most powerful commercial positions in any industry on the planet.

Above the silicon sit the buyers and builders. The hyperscalers — Microsoft, Google, Amazon, Meta — own most of the world’s cloud and AI infrastructure, and each has committed to capital-spending programs in the tens or hundreds of billions a year, a scale that used to belong to governments. The frontier labs — OpenAI, Anthropic, Google DeepMind, xAI, a few others — actually train the models; they’re small in headcount and enormous in influence. And a newer camp, described earlier in the book, is betting on a different architecture entirely: Fei-Fei Li’s World Labs, Yann LeCun’s AMI Labs, Bezos and Bajaj’s Project Prometheus.

Then the governments. The U.S. Commerce Department’s Bureau of Industry and Security writes the export controls. The FTC and DOJ handle antitrust. The EU’s AI Act is the world’s most detailed AI regulation, with reach beyond Europe’s borders. China is running a multi-year, multi-hundred-billion-dollar effort to build a chip industry free of American and Dutch technology. The Dutch, Taiwanese, South Korean, Japanese, Emirati, and Saudi governments all hold instruments in play.

And — easy to miss, and the most important for most of what this book is about — the local governments. County commissions approve zoning. Utilities negotiate power contracts. Municipalities grant or withhold water. State utility commissions decide rate cases. A county commissioner in rural Virginia has more immediate say over whether a data center gets built there than the President of the United States does, and that decision gets made in a meeting ordinary people can attend and almost never do.

All of these actors hold exactly the formal roles the textbook assigns them. None of them are villains; most are thoughtful people who’ll discuss the implications of their work at length, in public. The point isn’t that they’re bad. It’s that the way their decisions actually become outcomes runs on mechanisms the textbook doesn’t name — and that the actors themselves tend to discuss only in carefully chosen rooms.

All 19 chapters of The Invisible Trillion · Library · All courses