Fifty years ago, critics called the idea of a single integrated-circuit chip for a computer a fantasy. As the creator of that chip, I called it the future. I was right.
The critics are back, this time casting AI data centers as a crisis for American infrastructure, water and local communities. They’re wrong again. The data center companies are the heroes, not the villains. Yes, AI is hungry for electricity and water. But data centers are feeding it. The companies building these data centers are pouring billions of dollars into new power sources and new water sources.
McKinsey estimates $7 trillion in global data-center capital expenditure by 2030, with more than 40% flowing to the U.S. Having committed that kind of money, these companies are now building their own dedicated sources of electricity and water. Those investments will make energy less expensive and more readily available for everyone, not just for the tech companies footing the bill. And within the next 20 years, fusion reactors and “super-hot rock” geothermal sites will likely deliver electrical energy and clean water costing not much more than sunshine and rainwater.
These solutions will help support America’s all-of-the-above energy portfolio, while reducing and hopefully eliminating dependence on Arab oil.
The leading technology is enhanced geothermal, with Meta signing a geothermal power deal last year to supply its data centers. Next come small “modular” fission reactors—compact nuclear plants that can be factory-built and deployed almost anywhere. And eventually, enormous fusion reactors will harness the same power that generates sunlight, fueled by deuterium, a form of heavy hydrogen found in virtually limitless supply in seawater.
The technology is well understood for advanced geothermal and modular fission; the challenge is commercialization at scale, and data-center capital is significantly accelerating that timeline.
AI data centers are essential to American dominance of the worldwide AI race. Opponents are stirring up opposition to the AI data centers, miscasting them as a major problem to infrastructure. If they succeed, the effect will be to drive data centers into foreign countries, hampering America in the AI race. Local opposition has already delayed or denied dozens of U.S. data center developments. That friction has only intensified the industry investment in dedicated power sources that bypass the existing Grid entirely.
Moreover, the infrastructure issue is only temporary and will soon be resolved as advanced geothermal and modular fission reactors are commercialized. In the near term, companies are deploying modular gas turbines that can power data centers in weeks rather than years — and can supply the Grid during peak demand.
The activists hide these critical points—“temporary” and “soon”—and miscast it as an endless problem. But here is the irony: the activists are actually helping because focusing attention on a problem in America focuses investment and produces much quicker solutions. Data centers accounted for roughly 78% of all built-environment venture investment in 2025. The Data Center companies have the wealth and the motivation to make it happen sooner rather than later.
The world became addicted to cheap oil, and that addiction continues to this day with expensive oil. But, metaphorically speaking, future generations may wonder what the Strait of Hormuz had to do with the world’s energy supply because the Strait did not have any geothermal sites or fission reactors—all it had was seawater. And by then, geothermal sites and modular reactors will have been dispersed around the world, rendering the oil chokepoints irrelevant.
I’ve seen this before. When I set out nearly sixty years ago to put a whole computer on a single chip, the critics tried to make it sound like a fantasy. I founded Micro Computer, Inc. in the 60s, got it financed, and that “fantasy” went on to run the world. The same skepticism now aimed at AI and data centers that power AI will look just as shortsighted in hindsight.
My work on the microprocessor came out of my efforts to help America win the cold war against the Soviet Union. Today, I’m writing this for the same reason: keeping America first. As America celebrates its 250th birthday, let’s remember that the data centers powering AI are financing America’s future, not draining it.
Gilbert Hyatt is an American engineer, scientist and inventor known for his pioneering work in microcomputer technology.
This article was originally published by RealClearEnergy and made available via RealClearWire.
