Every major advance in technology, in human capability, has come with an increase in energy density of the primary fuel used in that technology. From biomass (wood, dung) to charcoal, from charcoal to coal, thence to oil, natural gas, and now nuclear power: Every major step forward, accompanied by an increase in energy density. We are now in the age of nuclear power, although it's a lot slower in the adoption and development than would be ideal.
This is also the age of artificial intelligence, or AI. AI is still in its infancy; we haven't yet understood all the uses to which it may be put or the problems it may cause. The AI genie, however, is not going back in the bottle. Which brings up another problem: AI is very energy-hungry, relying on massive data centers to run smoothly.
So, the future of this new technology isn't in chips or software, so much as it is in electricity.
AI is not just a software race. It is an infrastructure race. The countries that can most quickly permit, build, connect, and operate reliable power will have a decisive advantage in the next phase of digital growth. Every breakthrough in artificial intelligence still depends on physical systems: generation, substations, transmission, cooling, backup architecture, and the industrial work required to make it all run.
The scale of demand is no longer hypothetical. EPRI estimates U.S. data center electricity use in 2024 at roughly 177 to 192 terawatt-hours, with the potential to climb significantly by the end of the decade. New AI-oriented campuses are no longer 20- or 40-megawatt facilities, but increasingly in the 100-megawatt to 1-gigawatt range. That is no longer a conventional commercial load. It is grid-shaping infrastructure.
In other words, this is a world-changing technology, not just in what it can do, but in what it requires to do it. But there's a serious question, right here in these United States, as to whether our current infrastructure is up to the challenge. At least one recent poll has shown that American voters aren't exactly brimming with confidence where our electrical grid is concerned.
The Center Square Voters’ Voice Poll found that 55% of American voters said they completely or somewhat trust the American power grid. When that figure is broken down, 42% of voters said they only “somewhat trust” the power grid and 13% of voters said they “completely trust” the power grid. But 37% of voters say they don't trust the grid, including 13% who "completely distrust" it.
That's not really encouraging - nor is it completely unwarranted. Our electrical grid isn't exactly new, and there are some serious areas of vulnerability.
Read More: $510M AI Smuggling Case Blows Hole in U.S. Export Controls on China
White House AI Plan Signals Government No Longer Fully in Control of the Technology
So, what's to be done? Interestingly, the solution for the energy-hungry AI data centers may serve as an example for the provision of electrical power as a whole.
And, as you might suspect if you've been reading my work on energy issues, there's one solution that I see as obvious: Nuclear power. We are already seeing companies that run and maintain the big new data centers are increasingly turning to dedicated nuclear power plants. Small modular reactors (SMR) seem tailor-made for this application.
Data centers are increasingly turning to small modular nuclear reactors (SMRs) to power their AI-driven workloads, as the U.S. power grid struggles to keep up with the surging electricity demands of the tech industry. SMRs can be deployed on-site, eliminating the need to wait years for grid upgrades, and providing a reliable, carbon-free source of power. While regulatory hurdles and public perception issues remain, major tech companies are already investing in SMR projects, signaling that the technology is viable and bankable.
Nuclear power is everything we need for this next inevitable (now) step in human technology. Despite the howling cries of outrage from climate scolds, nuclear power is everything "green" energy sources are not. Nuclear power is clean. It's virtually emissions-free. It's reliable. Most of all, nuclear fuel represents a level of energy density that is orders of magnitude greater than coal or natural gas.
Here's the thing: This new energy tech can and should go well beyond dedicated power for a fixed installation like a data center - or a factory. Nuclear power, especially modular reactors, is the key to decentralizing our electrical grid. There are a couple of reasons why this should be done:
Reliability. There are too many obvious weaknesses in a system that relies, in many locations, on miles and miles of transmission lines to extend electricity to small rural communities. A small modular reactor could provide local, reliable power to those communities without the necessary time and expense of maintaining those miles of lines.
Economy. When small modular reactors are an established commodity, there will be waves of innovation. More companies will enter the market, and there will be more developments; technology, over time, always grows cheaper and more reliable. If you, like me, are old enough to remember the first pocket calculators, you may remember that the first generation of those cost well into three figures for very basic mathematical functions. But by the time I went to college, I could buy a Texas Instruments TI-35 scientific calculator that did scientific notation and advanced calculations for about $35.
Security. Decentralizing the grid makes it harder for a bad actor, either a nation state or just some random troublemaker, to knock a part of the grid offline. More generation points, fewer interconnections.
We solve today's problems with tomorrow's technology. Both AI and nuclear power are part of that technology.
