Data centers: The factories of the intelligence age

By Keith Ammon
Real Clear Wire

In 1956, the first hard drive weighed a ton and held about as much data as a single photo on your phone. Today, one AI data center processes more data in an afternoon than humanity wrote in its first five thousand years. We are living through the fastest infrastructure buildout in history, and few understand what these buildings do.

Everyone has a smartphone. Every photo you upload, every question you ask an AI, every package tracked to your doorstep passes through a data center. These buildings house the servers that power modern life. But the newest AI data centers are something more: the factories of the intelligence age.

As a software developer, I see them the way I see manufacturing facilities. AI data centers take in mountains of raw data and, through intensive computation, manufacture useful patterns: models that can predict, reason, and create. That is what model training is: the production of computed intelligence from raw information.

Like electricity, computed intelligence will soon be a utility we notice only when it fails, powering hospitals and schools with insight that once required human experts. And unlike any factory before it, there are no smokestacks, no rail cars, no fleets of trucks. Raw material flows in through fiber optic cables, and the product flows out the same way.

Whether America builds these factories or watches them rise overseas will shape our prosperity for a century. Offshoring the factories of the intelligence age would be outsourcing on a scale that makes the last generation of lost manufacturing look small.

America’s AI Action Plan, the White House’s official strategy, puts it plainly: “The United States is in a race to achieve global dominance in artificial intelligence. Whoever has the largest AI ecosystem will set global AI standards and reap broad economic and military benefits. Just like we won the space race, it is imperative that the United States and its allies win this race.”

Our rivals, with China first among them, understand the stakes. If we fail to build here, they gain the advantage. And just as the race accelerates, organized opposition to data center construction is spreading, and not all of it is homegrown.

OpenAI banned China-linked accounts caught using ChatGPT to manufacture fake grassroots content stoking opposition to U.S. data centers. The Bitcoin Policy Institute traced the broader pattern: Chinese state media, a nonprofit network funded by pro-Beijing billionaire Neville Roy Singham, and foreign dark money, linked to 21 coordinated campaigns in 14 states that stalled or blocked $23.6 billion in AI infrastructure investment. When foreign adversaries secretly amplify opposition, Americans deserve to know who is pulling the strings.

Foreign amplification does not mean local concerns are fake. Neighbors asking about jobs, taxes, water, and electricity deserve honest answers, and the answers hold up.

Critics note that a finished data center employs fewer people than a factory of similar cost, and dismiss the construction jobs as “temporary,” which is true in the sense that every construction job is temporary. But each hyperscale campus supports hundreds of high-paying construction jobs lasting years, and those workers’ spending supports local businesses.

According to PwC, each direct data center job supports 4.5 additional jobs elsewhere, in an industry that directly employed just over one million Americans in 2024. Brookings research confirms it: counties that land their first large data center see private employment rise 4 to 5%, information-sector jobs climb 22%, and wages rise 3 to 4%. 

In Loudoun County, Virginia, the heart of the world’s largest data center market, data centers generate almost half of the county’s property tax revenue. For every $1 in services the county provides, it receives $26 back. Imagine what that would do for your school district.

On water, the newest facilities recycle cooling water in closed loops rather than draining local supplies.

Energy is the hardest question, and the one New Hampshire has answered. With data centers projected to consume as much as 9% of America’s electricity by 2030, the Department of Energy expects data center power demand to double or even triple by 2028. It is not fair to burden existing customers with those costs.

That is why I co-sponsored HB 672 with Republicans and Democrats alike, and Gov. Kelly Ayotte signed it into law. This first-in-the-nation law lets a large energy customer like a data center develop its own power source with no costs on anyone else’s bill, because the entity is not connected to the existing grid. To keep that status, providers must operate independently, cannot cross state lines, and cannot use public rights-of-way.

For data centers, that is a feature, not a burden: onsite generation means power on the builder’s timeline, with no waiting years in interconnection queues for a grid hookup. For everyone else, it means new demand builds new supply instead of competing with families and businesses for the power we already have.

History offers one technology per era that rewires civilization: steam power, electricity, the internet. Computed intelligence is ours. The question is not whether this future arrives, but where it is built and whom it serves. If we build it here, power it wisely, and keep it in American hands, the result will be a wave of prosperity unlike any we have known.

With HB 672, New Hampshire drew the blueprint. Now we should build on it: welcome the factories of the intelligence age and have them bring their own on-site power.

Keith Ammon is a five-term representative in the New Hampshire State Legislature and majority floor leader.