Iran is a masterstroke to stop the AI backlash
Who can blame data centers for energy prices now?
We’ve gone from the “shock and awe” of my generation to the “shock and uhh” of Gen Alpha. Since February 28, Iran has been the overwhelming news story, bringing thirteen days of seesawing energy prices, expeditionary open-source intelligence work on the Strait of Hormuz, and a deeply polarizing debate centered on whether Iran’ll go all out or the lights will go out first. Even narrowing to the nexus of Iran and AI, analysis has highlighted Iranian attacks on data center infrastructure, leading Bloomberg to describe these facilities as the new “casualty of war.”
Maybe. I hold by my comments in May 2025:
Now, artificial intelligence and its voracious appetite for all forms of energy is animating the dreams of a younger generation of [Middle East] political leaders who can envision themselves at the fulcrum of global power through the ones and zeros of training and inference rather than the dollars and cents of barrels.
Actually, it looks like the Middle East is getting both the ones and zeroes of AI and the dollars and cents of barrels. Touché.
Even so, there is an under-explored angle in the frenzy of coverage, and it is America’s sordid domestic politics on artificial intelligence.
A backlash against AI has been gathering force for at least a year, and what was once a trickle against the logjam of politics has suddenly become a flood. New York is a perfect example. The Senate Environmental Conservation committee is debating a bill that would place an immediate moratorium on data center construction for three years. Meanwhile, the Senate is considering a bill that would add civil liability to any business that uses a chatbot to offer advice that, were humans to offer it, would be covered under occupational licensing, including health and law (1.2 million people are licensed in New York).
Such legislation is being mirrored across the country and in DC, where just yesterday, Senator Bernie Sanders called for a nationwide moratorium on AI data centers as part of a progressive push to stop AI.
America is among the most negative countries in the world on AI, and a variety of polls show that the AI backlash is deeply connected to consumer frustration with rising utility bills. Energy inflation for home delivery is one of the darkest spots in consumer price surveys. Costs rose almost 6.3% over the last year, and that was before Iran shut down the most important shipping channel for oil and gas. While extreme weather and a brutally cold and snowy New England winter is one of the larger causes for the rise, AI has ultimately become the target for much of the vitriol.
Those widespread pocketbook concerns have dovetailed with liberal environmentalists and their concerns around water conservation; religious conservatives over generative AI and its usage in sexual imagery; and broad middle class angst over the future of employment and the nature of work. That’s a hell of a convergence among otherwise political enemies, which is why the backlash has been gaining so much strength.
Unsurprisingly, in private conversations with lawmakers and a few recent off-the-record policy events, the general view has been that AI is going to be a major electoral issue in 2026, and the dominant one in 2028.
Yet all of that is now bunk, at least for the time being. America’s attacks on Iran have completely upended the electoral math, and particularly the connection between data centers and energy usage in the minds of voters. As Fortune bluntly headlined an article earlier this week, “People really hate AI but not as much as Iran—or Democrats.” Any rise in utility costs from data centers pales in perception to the battlefield in the Strait of Hormuz that’s streaming on every cable channel and social media account 24/7.
As with much of the Trump administration’s decisions, launching a war with Iran was an unintentional masterstroke for the tech industry. Tech’s golden star status has fallen in Washington, from the apogee of the Obama internet freedom agenda and the Arab Spring in the early 2010s to the perigee of AI, social media and more. The long-term implications of rising suspicion of the tech industry have been profound: an increasingly tough fight against China in the race for frontier tech, extraordinary regulations around social media usage and adult verification, and a breakthrough level of support for antitrust, just to name three. As one Hill staffer described it recently to me, there is an increasingly bipartisan basis for just “blowing up” the largest tech companies.
Distraction is a useful political strategy; it may not ensure victory, but it can help ensure delay.
Now? Iran is seared into the American psyche for the midterms. A new configuration of antiwar Democrats and isolationist MAGA Republicans will focus their efforts on war and peace, or narrowly on Anthropic’s ongoing fight with the Department of War/Defense. Even in the best-case scenario, I expect there to be large, Covid-like global disruptions to the oil and gas trade as companies rebuild their supply chains and get everything exporting smoothly again. The International Energy Agency launched the largest-ever strategic reserve release this week, but that’s merely a stopgap. AI is now an afterthought.
Make no mistake: at the federal level, the AI backlash will forcefully return by 2028. Employment will be the dominant conversation, either the casualization of once professional labor or the widespread unemployment of the managerial class (and likely both). As I critiqued in my brief note on that viral Citrini Research report:
The report doesn’t address any political blowback to the upper-middle class suffering from massive unemployment. Congress and the executive may not be the most responsive political bodies around, but truly, they are responsive to this voting bloc. If there were widespread job fears by elite Americans, these institutions would act, and act posthaste.
The bigger and more immediate challenge for AI, then, is the state-by-state fight now underway. Data center construction is running into America’s entrenched NIMBYism, and despite successes like xAI’s approval by Mississippi this week for a massive power plant expansion including 41 natural gas turbines, AI’s compute demands look to be insatiable. American hunger for ugly box buildings is not.
Meanwhile, the political organizational power of professions means that whole categories of work — including extremely lucrative and important functions like healthcare and law — will be increasingly declared off-limits to AI models. As I noted in “Professional Prerogatives” back in 2023 and continue to believe:
The upshot is that the United States has a decentralized governance system around licensing that ensures that at least some states will be the first to jump into an AI-driven world. The downside is that AI’s progress in the decade ahead will stall far earlier and last far longer than many analyses predict.
It’s stalling right now in New York, as it will in other states. The fight against AI progress is multiplying, and Big Tech will struggle to fight back (in fact, their notoriety at this point might make them the weakest fighters despite their resources).
Distraction is a useful political strategy; it may not ensure victory, but it can help ensure delay. There is only so much time legislators and voters can spend on individual issues. Increasing the visibility and salience of other problems like Iran will slow the momentum of the backlash. I am still hopeful that the aftermath of our intervention in the Middle East is going to redound quite positively to the United States, but the immediate economic effects will be dire. AI thus gets a little breather.
What we really need, though, is a much more positive and broad-based motion in support of AI. What’s so frustrating about the backlash is that at the same time politicians and citizens are lobbying to ban the infrastructure and applications of this technology, everyone is using it. Anthropic’s and OpenAI’s revenues are staggering, with never-before-seen growth rates. Their upcoming IPOs (along with that for the combined SpaceX–xAI) will likely be the largest of all time (current record holder: Saudi Aramco. It all comes full circle!) Usage rates globally are unlike anything we have seen for technology that was released to the general public less than four years ago.
Everyone is using it, but no one wants the externalities or apparently upgraded power infrastructure. That’s the illogical madness of AI today in America, as it is with so many other issues, from housing to reindustrialization. Trump’s attack on Iran was perhaps just as mad, but at least it’s got a logic.
