Earlier this month, Moderna announced successful late-stage results for a melanoma vaccine that significantly reduces the risk that patients’ cancer will return or spread. The company says it can now sequence a tumor and then use artificial intelligence to identify which mutations to target in a personalized vaccine. Moderna calls the process “a testament to the power of AI.”
That same week, Anthropic reported that its Claude models designed new proteins that successfully latched onto 14 of 15 biological targets—work that can form the basis of new medicines. The breakthrough could dramatically shorten one of the most laborious steps in developing new drugs and therapies.
Finally, a reason to check your email.
Sign up for our free newsletter today.
These are glimpses of what the AI revolution is already making possible. While chatbots and AI “slop” may get more attention, the computational power and increasingly sophisticated capabilities behind AI are already giving us extraordinary new tools to fight disease, design medicines, analyze enormous biological datasets, improve manufacturing, and accelerate scientific discovery, with enormous benefits for humanity.
Yet the physical infrastructure on which these advances depend is becoming politically toxic. Americans increasingly oppose AI data centers in their own communities. Concerns about energy and water use and other local impacts are resulting in moratoriums and rejected projects, even in rural areas. Several pro-growth governors who previously touted data centers are starting to crack down on the industry. The debate over data centers is increasingly shifting from where we should build them to whether we should build them at all.
Halting or slowing down data-center construction would mean deliberately constraining the infrastructure behind one of the most important innovations of the twenty-first century. If the growing panic succeeds in blocking that buildout, it will mean less economy-wide investment, slower scientific and technological progress, more expensive computing power, and fewer productivity gains and breakthroughs—to the detriment of Americans and of humanity more broadly.
AI is rapidly becoming a general-purpose technology in today’s modern economy. Much as electricity transformed the twentieth century, computing capacity is becoming a basic input to innovation in the twenty-first. Data centers are simply the physical infrastructure that supplies it.
The investment boom alone is already contributing significantly to economic growth, but the biggest benefits come from what all that computing power enables: researchers developing drugs faster, doctors catching diseases earlier, manufacturers boosting production, and enhanced productivity.
The political challenge is that many of these benefits are spread across the economy, while the costs of new data centers are more concentrated—and thus more apparent. In mid-August, the National Republican Senatorial Committee warned that public concerns over data centers could become an electoral liability in November—and that if some politicians lose because of the issue, others “will not go near the next one.”
Several political leaders understand the stakes. Pennsylvania Senator Dave McCormick has called the current debate a “false choice” between letting developers run roughshod over communities and shutting down the industry. He argues instead for rules that protect ratepayers and local land and water resources while still allowing America to build. Montana Senator Tim Sheehy recently warned that opposition to data centers could become another chapter in a familiar pattern of lost American industries. “First they killed our timber mills, then our mines, then our factories—now data centers,” he wrote on X.
Sheehy’s comparison of data centers to lost American industries is on point. For decades, domestic mining and mineral processing became harder and less competitive. Meantime, China built up its capacity. Today, the United States depends heavily on Chinese supplies of rare earths and other minerals essential to modern supply chains. The same could happen with AI infrastructure. If data centers become too hard to build here, American companies could grow dependent on foreign computing capacity to train proprietary models or conduct cutting-edge research. Government agencies could find themselves renting computing capacity from geopolitical adversaries. It’s easy to imagine Washington later spending enormous sums to “reshore” that capacity, just as it is doing now with critical minerals.
Fortunately, the public may be more persuadable of the benefits of data centers than the headline polling suggests. Echelon Insights recently found that voters favored banning nearby data centers. But every local benefit it tested—such as developers paying for grid upgrades, new tax revenue reducing property taxes, or more school funding—flipped the result toward allowing them.
Communities can look to numerous examples of how data centers can benefit them. In Quincy, Washington, data centers vastly expanded the local tax base; the city’s property-tax levy rate is about 70 percent lower than before they arrived, even as Quincy built new schools, streets, sidewalks, a hospital, and other infrastructure. In rural Louisiana, increased tax revenues from construction of Meta’s massive campus helped finance individual bonuses of up to $50,000 for local teachers. Last week, a Georgia county announced a data-center deal with Google that will provide it with $40 million per year for seven years, allowing it to virtually eliminate its share of property taxes and raise salaries for firefighters, police officers, and other county employees.
The buildout could also help finance investment in aging electric grids, while also cutting residential electricity rates. Utilities shoulder enormous fixed costs, but when large users pay for the costs they impose, their steady demand can support new generation and transmission while spreading those fixed costs across more customers. That can fund grid upgrades and cut prices for residential consumers. Utilities in Georgia and Indiana are pointing to data-center growth to explain why they are now reducing residential electricity prices.
But the AI industry and its supporters need to do a better job explaining that data centers have much more to offer than just these local benefits. AI’s biggest promise lies in much broader, more universal applications—cancer treatments, personalized medicine, faster scientific discovery, better manufacturing, and economy-wide productivity gains, to name a few. Chatbots and AI slop are just a distraction.
America must not make the computing power fueling these innovations artificially scarce. The challenge is to build enough of it while addressing legitimate local costs and making sure that host communities share the gains.