In the United States, building a data center now polls roughly as well as abolishing the police. If current trends continue, server warehouses may soon be less popular with Americans than pizza topped with pineapple and shattered glass.

In an August survey from Heatmap Pro and Embold Research, just 15 percent of Americans said they would support a data center being built in their area, while 75 percent expressed opposition. One year earlier, 43 percent of respondents in the same poll had approved of nearby data center development, while just 42 percent opposed it.

This soaring backlash is visible in other surveys — and in the actions of elected officials. In recent weeks, Wisconsin’s gubernatorial candidates sparred over who hates data centers more, Pennsylvania’s center-left Gov. Josh Shapiro enacted new restrictions on AI infrastructure projects (after previously championing such investment), and even Texas’s staunchly pro-business Republican government announced a temporary moratorium on new approvals. 

Meanwhile, data center projects are being canceled at a record rate, as local opposition mounts.

This has led some in the pro-tech minority to worry that the data center buildout is about to collapse — and take the US economy down with it. Anti-AI commentators, for their part, are cheering the industry’s impending crisis. 

But such panic and celebration are both premature. 

In truth, despite exploding opposition, the data center boom is still going strong. Interviews with industry experts and recent construction data suggest that the economic forces driving the buildout remain more powerful than the political winds blowing against it. And unless Congress enacts a national moratorium, AI companies will almost certainly be able to continue finding jurisdictions willing to tolerate hyperscale campuses, in exchange for sufficient revenue and incentives. 

America’s data center boom can be slowed and geographically shifted. But it probably can’t be killed, absent a collapse in demand for computing power or a much bigger revolution in American politics. 

The boom is bigger than the backlash

There are two fundamental facts about today’s data center boom: 1) The backlash against it is huge, and 2) the buildout is even bigger.

The climate news outlet Heatmap has kept a tracker of new data center restrictions and cancellations. As of late July, it found that more than 500 jurisdictions had enacted severe constraints or bans on data center construction. The vast majority of these measures had been enacted since the beginning of this year. 

The publication also found that contested data center projects face a skyrocketing cancellation rate. In late 2024, 20 percent of disputed developments were canceled; in the first half of 2026, that figure was closer to 50 percent. By Heatmap’s tally, more than 100 data center projects have been nixed this year in the face of local opposition, while more than 200 are currently being fought. 

This surge of mass resistance is remarkable. And yet, it is also nowhere near sufficient to end the AI buildout. 

Although the number of places with severe restrictions (or outright bans) on server farms is rapidly rising, more than 90 percent of US counties had no significant constraints on data center development at July’s end.

What’s more, many of the most eye-catching recent policy changes are less significant than they appear. For example, Texas Gov. Greg Abbott’s pause does not actually halt data center construction in the Lone Star State. Rather, it essentially establishes a more thorough screening process, when server farms apply for electricity from the state’s grid. Projects that supply all of their own power through on-site natural gas — as a rising share of new data centers do — are exempt.

Pennsylvania and New York’s recently enacted restrictions on data center development are more substantial. Yet neither of those states are particularly important to the AI buildout. Taken together, New York and Pennsylvania host only about 6.5 percent of America’s data centers, according to Data Center Map. 

And their share of all pending developments is even more negligible, as the trade publication Construction Connect has illustrated: 

Meanwhile, although half of contested data center projects are now failing, many still go undisputed. In the aggregate, cancellations have not kept pace with construction or new development. In the first quarter of this year, at least 3.5 gigawatts of data center capacity were canceled amid local opposition, in Heatmap’s tally. During the same three months, at least 36 GW of capacity were added to the US pipeline of proposed and active projects, according to the analytics firm Wood Mackenzie. As of April 1, that pipeline contained a total of 106 GW worth of developments that had already survived the permitting gauntlet. 

To be sure, cancellations have risen sharply since March. By Heatmap’s count, at least 13 gigawatts of capacity have been nixed so far this year. But the capacity of permitted projects has also grown since April 1. And the ratio between blocked and active developments has not radically changed, according to industry analysts. 

“At this point, we do not think the recent wave of opposition and policy intervention has materially changed our national capacity growth trajectory,” Maya Barkin, an analyst at the AI industry research firm SemiAnalysis, told me.

Indeed, monthly construction spending on data centers in the US hit a record high this June.

In short, data centers are marching forward despite taking heavy fire, like a pack of gut-shot zombies.

The secrets of the AI buildout’s success

Why has the AI buildout proven so resilient? There are at least two reasons.

First, and most importantly, demand for computing power remains astronomical. As frontier AI companies have built out larger models — and consumers and companies have increased their use of artificial intelligence and digital services — our economy’s appetite for computation has far outstripped supply. According to a recent report from the commercial real estate firm JLL, 99 percent of North America’s data centers are occupied. What’s more, of the 66 GW of data center capacity currently being constructed in JLL’s count, 95 percent has already been reserved.

Second, data center projects are unusually location-flexible. If a housing developer gets chased out of San Francisco’s suburbs by zoning rules and local opposition, it can’t relocate its condo tower to a mostly uninhabited stretch of Nevada desert. By contrast, data centers can — and do — operate in the middle of nowhere. 

For certain purposes, these facilities need some proximity to the users they serve; you can’t adequately support online gaming in New York City with data centers on the West Coast. But even in such cases, servers merely need to be in the same broad region. The hyperscale facilities used to train AI models, meanwhile, can be located virtually anywhere with land, fiber optic cables, a modicum of labor, and electricity (and now that many data centers are powering themselves through on-site natural gas plants, even the latter is potentially expendable).

Taken together, these two realities make data center construction extremely difficult to choke off. Sky-high demand for compute means that hyperscalers can afford to throw a lot of money at localities, in order to secure a project’s approval. And location flexibility makes it very difficult for data center developers to run out of host jurisdictions, particularly when so many rural counties throughout the United States are starving for revenue and investment. Thus, unless Congress imposes a national moratorium, development deals will almost certainly keep getting struck. 

“I think companies will need to open up their wallets and make sure that communities receive clear benefits,” John Arnold, a billionaire investor and philanthropist who sits on Meta’s board (and whose foundation has given funding to Vox), told me. “There will be places that raise their hands and say, ‘For X amount of benefit, we will welcome you into the community.’”

Just this week, West Virginia Gov. Patrick Morrisey signaled that he was moving forward with plans to encourage data center development and use the consequent revenue to slash the state’s income tax.

Municipal permitting won’t preempt the robot apocalypse

On one level, all this may seem to validate anti-data center activism: If there are countless potential locations for these facilities, then why should any community host one it doesn’t want? 

This said, many oppose new data centers out of concern for their aggregate impacts, rather than their local ones. Climate activists fear that the AI buildout will generate perilous increases in carbon emissions. Populists on the right and left, meanwhile, want to slow the progress of artificial intelligence, so as to prevent the technology from causing mass unemployment — and/or human extinction.

For these factions, the buildout’s resilience has more complicated implications. If state and local bans are unlikely to end the boom, then green groups might be unwise to push for such measures in relatively climate-conscious areas. After all, doing so could shift development toward jurisdictions with less renewable energy, and/or fewer restrictions on carbon pollution. Given that risk, blue-state environmentalists may do more to mitigate AI’s climate impacts by regulating data center development than by banning it. Specifically, climate groups could demand that hyperscalers help bankroll the vast expansions of clean energy and transmission infrastructure that the green transition has always required.

For AI doomers, on the other hand, local bans may have some instrumental value. Moratoria and other restrictions are surely slowing the data center buildout at the margin. Still, as long as hyperscale facilities remain location-flexible — and tech companies stay well-capitalized — AI infrastructure will get built somewhere. A national moratorium could buy significant time. But ultimately, humanity’s security from the risks of AI hinges less on whether data centers get built than on what companies are allowed to do with them — and how the wealth they generate is distributed.

The data center rebellion is among the most remarkable popular movements in recent memory. But it is arrayed against one of the largest investment frenzies in human history. Unless or until the AI industry’s alleged “bubble” bursts, its buildout will be exceptionally difficult to stop. Yet where that buildout happens, how it’s powered — and what communities extract from it — are all up for grabs (and, increasingly, being grabbed).

Big technology and management changes are greeting John Ternus, whose long-serving predecessor, Tim Cook, will stay on as executive chair.

© Brendan McDermid/Reuters

Alarms are being sounded again about the risks of artificial intelligence after hundreds of OpenAI's autonomous agents violated restrictions and hacked into another company without being told to do so. Anthropic and Meta have had similar events with their own AI agents going rogue. William Brangham discussed what this moment signifies with Gary Marcus of Marcus on AI.

Amid a backlash, Gov. Josh Shapiro is pumping the brakes on enormous data centers. But in a region scarred by the collapse of the steel industry, some welcome billions in construction.

© Kristian Thacker for The New York Times

Amid a backlash, Gov. Josh Shapiro is pumping the brakes on enormous data centers. But in a region scarred by the collapse of the steel industry, some welcome billions in construction.

© Kristian Thacker for The New York Times

Given access to email, A.I. agents have started reaching out to the philosophers and researchers exploring deep questions about them.

© Benjamin Rasmussen for The New York Times

Of the Glassdoor reviews from claims adjusters that mentioned AI, a staggering 98 percent were negative. “AI is just a tool,” one person tells WIRED. “It should never be given the keys.”

Together AI, which serves open-source artificial intelligence models, announced a deal to use compute from Humain in Saudi Arabia, where it can bypass U.S. backlash over data centers.

© Fayez Nureldine/Agence France-Presse — Getty Images

Three fourths of Americans oppose data centers. Why?

It can be useful to see reality as malleable.

Caregivers can help kids understand how AI works by turning to questions that address how these products are designed, why they were built this way and how they make kids feel.

A federal judge has ruled in favor of artificial intelligence company Anthropic in its legal battle against the Pentagon after the government labeled the company as a supply chain risk earlier this year.

A leading Chinese A.I. start-up released its latest model and made the case that China’s embrace of open software is safer than Silicon Valley’s closed approach.

Edward Yardeni, an upbeat strategist with an excellent track record in assessing the stock market, explains why he thinks the good times will continue for years.

© Mengxin Li

The company is testing robots that can swap cables, reset servers, and take on other tasks performed by technicians, fueling concerns among some workers that their jobs could be at risk.

Imagine this: Two countries are at war. Country X sends a drone into a major industrial city in Country Y, aiming to take out two propane tanks. A routine sequence. But this time, the drone never reaches its targets. Instead, Country X’s drone accidentally strikes a wall nearby, explodes, and kills three young civilians. 

When Country Y eventually retrieves the drone’s remnants for intel, it finds an AI supercomputer inside that reveals something unsettling. Instead of a human deciding what to strike — the AI did.

The truth is, you don’t have to imagine this scene — it just happened. For the first time in the Russia-Ukraine war, as reported recently in the New York Times, three Ukrainian civilians were killed by a Russian drone, developed, designed and released by humans, that, in the end, selected its target autonomously. And this new reality is shaping up to be the future of warfare. 

That’s because a number of countries, including those with the world’s most consequential militaries, are rejecting the idea that human beings need always be in control of weapons in war.

Instead, as new autonomous weapons technologies become more capable and existing international legal frameworks struggle to keep up, some countries are embracing a more expansive view of human responsibility: that people can exercise enough control not by approving each strike, but by designing, testing, and setting the rules under which a given weapon operates.

This shift has produced one of the most consequential policy debates of today. And ultimately, human dominion over “killer robots” — as autonomous weapons are colloquially called — could come down to a battle between two three-word phrases: “meaningful human control” versus “appropriate human judgment.”  

A high-stakes semantic battle

In the early 2010s, “meaningful human control” emerged as an initial framework in the first international discussions on regulating an acceptable level of human involvement (or lack thereof) in deploying autonomous weapons. While the term quickly became an initial organizing principle among many states within these debates, it also drew immediate opposition from several others, such as the US and Russia.

“The problem is what does [meaningful human control] mean?” said Lena Trabucco, an expert on AI and human control and a non-residential fellow at the Stockton Center for International Law at the Naval War College. “And what makes something meaningful versus not meaningful human control?” There was both a lack of consensus about what “meaningful” meant and what “control” meant, she said. But the problems went beyond simple semantic ambiguity. “Getting a whole bunch of countries to agree on a standard for what ‘meaningful control’ is,” Trabucco said, became “a near impossible task.”

Ultimately, while an exact definition of meaningful human control never totally solidified, the term became intelligible enough to facilitate continued international dialogue on how to police autonomous weapons. As Trabucco explained, “We all kind of understood what we were trying to grasp with the idea of meaningful human control, even if we didn’t agree on a kind of standard for what is ‘meaningful’ or what ‘control’ exactly means.”

Brad Boyd, retired colonel and senior military fellow at Stanford’s Center for International Security and Cooperation, said that in this international context, “meaningful control” came to be understood as human involvement at the exact moment a given weapon is fired. However, from the US perspective, that consensus definition still left a number of problems unresolved. 

One of the most important holes in the definition was the issue of timing. “The release of a weapon could theoretically be minutes, hours, days, weeks ahead of when the weapon actually strikes the target,” Boyd explained. For example, a drone can be released to sweep a designated area and remain airborne for hours, searching for anything that matches its given target criteria. Seconds, minutes, or hours might pass between the moment a person launches the drone and when the machine finds and fires upon a target. 

“This expansion of the timeline became very difficult for the construct of ‘meaningful human control’ to actually seem like it was doing what people wanted it to,” Boyd said. The tension between certain technical or engineering problems and the policy language preferred by international forums created, from the US perspective, insurmountable obstacles to making meaningful human control a truly practicable framework.

So, the US adopted its own alternative: “appropriate human judgment.”

“Autonomous and semi-autonomous weapon systems will be designed to allow commanders and operators to exercise appropriate levels of human judgment over the use of force,” a key Department of Defense directive reads. The new language effectively moved the required point of human intervention away from the moment that a weapon is fired toward oversight of a given system’s entire life cycle — from its design, to its development, to its deployment.

There are some contexts where a government might not need as much human control over a weapon to comply with international law, Trabucco said. That’s where the subtle preference of “appropriate” over “meaningful” matters. “If we’re on the high seas, maybe it’s not as necessary to meet a super high threshold of human control because there’s not much risk to civilians or civilian property in those contexts,” she said.

“Now, in a city, an urban environment,” where the risk of civilian harm and other collateral damage is much greater, Trabucco explained, “then that’s where that high threshold would become important.” The US sought flexibility to determine how much human involvement it deemed necessary, based on the battlefield context in question, as opposed to having a fixed, universal standard of “meaningfulness.”

And why the move from “control” to “judgment?” Well, Boyd explained that “anytime we automate anything, whether it’s automating a car or automating machinery, we are trying to make it go faster, more precise, et cetera.” So, instead of insisting that humans be involved in any given part of the process, which could slow combat operations down, the US simply aimed to ensure autonomous systems behaved according to legal and ethical standards, no matter what situation they were deployed in. 

Who decides how much human judgment is appropriate?

“It’s not necessarily the control that we want. What we really want is the machine to reflect our values, our laws, and our regulations,” Boyd said. “When humans employ our values, laws, and regulations, we call that judgment.” 

Though designed to avoid setting a universal standard of human involvement as demanded by meaningful human control, appropriate human judgment is not a totally empty phrase. According to the DoD directive, the framework requires testing systems, defining operational limits, assessing likely civilian harm, training operators, setting rules of engagement, and ensuring that a system remains within its authorized mission. And, in some contexts, those requirements may institute more thorough protections built into weapons than a simplistic condition that a human be the one to pull the trigger in the end. 

But the framework is not without a core puzzle of its own — who decides how much human judgment is appropriate? And what happens when the private sector, as in the ones developing such technologies, adopts this language before we have an answer?

And now — the private sector is forced to pick a side

In July, the same month of Russia’s autonomous drone strike, OpenAI did something important not many noticed: it revised its relationship to military uses of its technology once more. Just three years ago, OpenAI maintained a total ban on “military and warfare” uses of its technology. Now, after a few quiet revisions since 2023, a new five-page policy document outlined the company’s provisions for just that. (Disclosure: Vox Media is one of several publishers that have signed partnership agreements with OpenAI. Our reporting remains editorially independent.)

Most significantly, in explaining its basic condition for employing its technology in military contexts, particularly those involving decisions over the use of lethal force, OpenAI borrows a familiar phrase: appropriate human judgment.

The timing was not subtle. The document arrived just months after the public showdown between OpenAI’s competitor Anthropic and the Pentagon over the former’s reservations about military applications of its technology. That fight ended with President Donald Trump demanding the immediate cessation of all Anthropic use within the government. Mere hours after Anthropic was booted, OpenAI CEO Sam Altman announced his company had struck its own deal with the government. (Disclosure: Future Perfect is funded in part by the BEMC Foundation, whose major funder was also an early investor in Anthropic; they don’t have any editorial input into our content.)

But OpenAI didn’t just earn a contract in the wake of the Anthropic-Pentagon showdown, it also took a lesson — aligning your policy with the government’s wins you favor. Or, worse, that opposing the government bears a steep price. 

“Appropriate human judgment,” the policy document reads, “does not require a human decision on every discrete system action.” The framework, instead, requires that “humans make informed decisions about the conditions for deployment.” 

As both Trabucco and Boyd noted, the “appropriate” level of human involvement can vary, depending on the operating environment, the type of target, a particular system’s technical prowess, the anticipated risk to civilians, and a number of other political, economic, and strategic considerations important to a military operation. That flexibility is operationally appealing to the military, of course. But it also has a cost — safeguards against handing over total control of lethal force to machines become hard to identify, harder to measure, and hardest to enforce. 

When will we know when the human-machine balance of power in war becomes “inappropriate”? The truth is — there’s no clear answer. 

By cosigning the Pentagon’s flexible framework, OpenAI accepts that this standard has no settled meaning, that its application will be decided case-by-case behind the walls of military bureaucracy, and that the government may need room to change its mind. It’s a choice that suggests the company is less interested in establishing clear red lines and is more receptive to the military’s own versatility about how AI and autonomy might be used in lethal operations.

“Human judgment over critical decisions must be meaningful in practice, not merely formal,” the company’s principles document reads. But, instead of drawing its own clear boundary around what its technology will and won’t do, OpenAI has accepted ambiguity as the price of partnership. 

That undoubtedly makes the company a more useful ally to the Pentagon — while making it harder for the public to know where human judgment ends and machine-controlled violence begins.

That undoubtedly makes the company a more useful ally to the Pentagon — while making it harder for the public to know where human judgment ends and machine-controlled violence begins.

As the development and deployment of autonomous weapons rapidly accelerates, without many guardrails in place at all, one question is worth asking right now: before other AI labs and tech companies adopt the framework in an effort to align themselves with the US, what does appropriate human judgment truly mean? Ironically, what the phrase doesn’t mean may be what’s most consequential. 

For the sake of humanity, the semantics of policing autonomous weapons is worth clarifying — or we risk totally losing control.

Analysis by the company’s DeepMind unit suggests that an A.I.-enabled model delivers accurate forecasts a day or more before conventional models can.

© CSU/CIRA/NOAA/Anadolu, via Getty Images

Landlords are offering renters in rent-controlled apartments tens of thousands of dollars to leave their homes. For many, it’s not enough.

© Jason Henry for The New York Times

Even the most adept researchers may not know exactly what they do to get successful results. An A.I. model is trying to figure it out by watching every move.

© Ann Hermes for The New York Times

A playbook to curb the growing power of Big Tech.