Artificial intelligence (AI) has been heralded as a technological revolution that will transform our world. From curing diseases to automating dangerous jobs to discovering new inventions, the possibilities are tantalizing. We’re told that AI could bring unprecedented good—if only we continue to invest in its development and allow labs to seize precious, finite natural resources.
Yet, despite these grand promises, most Americans haven’t experienced any meaningful benefits from AI. It’s yet to meaningfully address most health issues, and for many, It’s not significantly improving our everyday lives, excluding drafting emails and making bad memes. In fact, AI usage is still largely confined to a narrow segment of the population: highly educated professionals in tech hubs and urban centers. An August 2024 survey by the Federal Reserve and Harvard Kennedy School found that while 39.4% of U.S. adults aged 18-64 reported using generative AI, adoption rates vary significantly. Workers with a bachelor's degree or higher are twice as likely to use AI at work compared to those without a college degree (40% vs. 20%), and usage is highest in computer/mathematical occupations (49.6%) and management roles (49.0%).
For the majority of Americans, especially those in personal services (12.5% adoption) and blue-collar occupations (22.1% adoption), AI remains an abstraction, something that exists in the future rather than their present.
While the rewards of AI are still speculative, the costs are becoming increasingly tangible. And the people paying those costs are not the ones benefiting from AI today. In fact, much of the burden of AI’s development is falling squarely on the shoulders of the American West—both its people and its land. According to recent research, data centers in the United States are consuming an increasing share of the country's total electricity. These facilities, which are crucial for AI deployment, used about 3% of all U.S. electricity in 2022. By 2030, their share is estimated to grow to 9% of total U.S. electricity consumption.
This surge in energy demand is particularly significant for the Western United States, with its concentration of tech hubs and data centers. Moreover, the carbon dioxide emissions from data centers may more than double between 2022 and 2030, further intensifying the environmental impact on these regions.
Here’s why: developing and deploying AI requires enormous amounts of energy. Advanced machine learning models demand computing power on a scale that most people can barely comprehend. Recent International Energy Agency projections highlight the magnitude of this demand: global electricity consumption from data centers, cryptocurrencies, and AI is expected to reach between 620-1050 trillion watt hours (TWh) by 2026. To put that in perspective, 1,000 TWh could provide electricity to about 94.3 million American homes for an entire year.
All that energy has to come from somewhere. Increasingly, it’s coming from the West —the part of the country that has long been tapped to fuel the nation’s ambitions, from oil and gas to solar, wind, and hydropower.
This energy extraction is putting immense pressure on the West’s already strained resources. Land is being consumed, water is being diverted, and communities are being disrupted, all to keep the lights on in tech labs far removed from the realities of life on the ground. The irony is that the very regions making AI possible are the least likely to benefit from it.
The rush to ramp up energy production for AI feels eerily familiar. We’ve seen these “get rich quick” schemes before—industries that swoop into rural areas, extract valuable resources, and leave environmental and social destruction in their wake. The West has been exploited before by out-of-state interests with big promises and shallow commitments, and AI risks becoming the latest chapter in that story.
We need to have an honest conversation about the true costs of AI development—particularly when it comes to energy consumption. AI labs may talk about curing diseases and inventing new technologies, but until those breakthroughs become reality, the rest of us—especially those in the West—are left footing the bill. And right now, that bill is being paid in the form of depleted resources and communities that are being squeezed for the sake of a future that remains distant and uncertain.
The truth is, we can’t continue to deplete our resources in the hope that AI’s promises will eventually materialize. We must demand accountability and transparency from those developing AI. Where is the energy coming from? Who is being impacted? And most importantly, who will benefit?
AI’s future may hold incredible potential, but we must make sure that we’re not sacrificing the West’s present for a future that may never arrive. If AI is going to reshape our world, it must do so in a way that lifts up all Americans, not just a select few. Until then, we need to be clear-eyed about the costs—and demand better.
Frazier is an adjunct professor of Delaware Law and an affiliated scholar of emerging technology and constitutional law at St. Thomas University College of Law.
image of U.S. President Donald Trump is displayed on a digital billboard in Times Square in New York on April 8, 2026.
Trump is stuck between two realities. Neither serves the American people
Normally, I worry that events may overtake a column. But not so with the Iran war.
I don’t worry about running afoul of a headline or Truth Social post from the president because what is said about the situation is no longer very relevant to the reality.
On April 8, Nick Catoggio, my Dispatch colleague, dubbed an earlier stoppage with Iran “Schrödinger’s ceasefire.” This was a reference to the famous thought experiment by the physicist Erwin Schrödinger, who was trying to explain the weirdness of “superpositionality” in quantum physics. A cat in a box is both dead and alive at the same time until you open the box. Schrödinger meant to illustrate the absurdity of the idea that particles aren’t any one thing, but a “cloud of probabilities.”
The Trump administration is stuck in a word cloud of probabilities of his own making. The war is over. The war is on. The war isn’t a war. We have a deal, but we don’t have a deal, but we’re about to have a deal. We destroyed Iran’s military. No, we left it intact. We want regime change. No we don’t. We already accomplished it. We “obliterated” Iran’s nuclear program a year ago. We had to go to war in February to prevent nuclear war. The Strait of Hormuz is open, closed, or something in-between. No deal without “unconditional surrender.” Let’s make a deal!
This everything-all-at-once vibe can be disorienting, particularly since most Americans didn’t have a war with Iran on their bingo cards until the shooting had already started. President Trump didn’t prepare the country or consult with Congress beforehand because he thought it would all be a smashing success in a matter of weeks.
The miscalculation that started it all: killing Iran’s Supreme Leader, Ayatollah Ali Khamenei, and much of Iran’s senior leadership, on the first day of the war. To “the great proud people of Iran, I say tonight that the hour of your freedom is at hand,” Trump announced on Feb. 28. “When we are finished, take over your government. It will be yours to take. This will be probably your only chance for generations.”
I support regime change in Iran and shed no tears for Khamenei or his goons. But when you start a war by killing the regime’s top leaders, it’s not unreasonable for the remaining ones to conclude that you really intend regime change.
Khamenei was a murderous fanatic, but he was a fairly cautious one. He liked to threaten closing the Strait of Hormuz or attacking our regional allies, but he was reluctant to actually do it, fearing it would invite a regime change war. The mullahs and IRGC goons believed, not unreasonably, that if they lost their grip on power, they’d be lynched by the Iranian people they’ve brutalized for decades.
By starting with a regime change war, Trump removed any reason for the regime not to go for broke. When you have nothing to lose — particularly when you are a millenarian religious fanatic — a Persian Alamo strategy makes a lot of sense.
So Iran closed the Strait of Hormuz and attacked its neighbors.
But it turns out this wasn’t the Alamo. In the contest of wills, Trump blinked. The Iranian regime’s tolerance for punishment proved — so far — to be greater than Trump’s and that of our gulf allies. Militarily we could finish the job, but that would require ground troops and much greater economic turmoil. In a conflict Trump launched unilaterally without the prior support of Congress, NATO or the American people, Trump doesn’t have the political capital for that.
But that’s only half the problem. Trump wants the war over, but he doesn’t want to pay — militarily, economically, politically — what that would cost. So he wants to make a deal that ends it. But there is no deal available that wouldn’t come at an equally undesirable cost. Any deal that looks like what President Obama struck with the Iranians would be too embarrassing to bear. But the Iranians are convinced that they can get just such a deal, and they’re willing to drag things out as long as it takes.
The result: Trump’s in a box of his own making. He thinks he can talk his way out by simply asserting a reality that doesn’t exist. When the financial markets get nervous, he announces a breakthrough that is, at best, a possibility. When the Iranians agree to a deal that looks similar to one Obama might negotiate, Trump goes back to his threats.
It can’t go on forever. But I’m sure it’ll last until long after this column is forgotten.
Jonah Goldberg is editor-in-chief of The Dispatch and the host of The Remnant podcast. His Twitter handle is @JonahDispatch.