Data center construction represents a major opportunity to create millions of skilled, high-paying jobs that could rebuild America's middle class, but local opposition threatens to derail this economic pathway and push infrastructure investment and capability building overseas.
- America needs roughly 500,000 new construction workers annually through the rest of the decade to meet data center and AI infrastructure demand.
- Craft workers on data-center projects earn 32% more than comparable construction workers while developing valuable, transferable skills.
- PwC estimates $25 trillion in AI infrastructure investment through 2050—comparable to the entire world's railroad, electrical grid, and highway spending over 50 years.
- 71% of Americans oppose local data center construction despite fearing AI job losses, creating a paradox that prioritizes short-term local concerns over long-term economic opportunity.
- If America blocks data center construction domestically, companies will build abroad and sell compute back to Americans, allowing other countries to build industrial capability and middle-class opportunities.
According to a recent Pew survey, 71% of Americans have a fear (real or imaginary) that AI will eliminate millions of jobs over the next twenty years.
Naturally you’d think that same group would be doing everything it can to support the creation of more American jobs.
But a recent Gallup poll found that 71% of Americans also oppose data-center construction in their local area, even though building them will require millions of workers over the coming decades. So, millions of Americans simultaneously fear AI will take away jobs while opposing the infrastructure buildout creating millions of future-proof ones.
The resource concerns around data centers are real, and the tech companies profiting from AI should bear the lion’s share of the responsibility for solving them. But the data center issue isn’t really a fight between communities and corporations. It’s more lopsided than that. It’s really a fight between short-term local interests (aka NIMBYism) and one of the greatest opportunities to rebuild America’s middle class for generations to come.
During the industrial buildout of the twentieth century, craft work was one of the primary ways young Americans entered the middle class. In 1950 alone, more than 40% of employed American men under 35 worked as craftsmen or industrial operatives.
That generation built much of modern America, including tens of thousands of miles of highways, millions of homes, and the power, communications and transportation infrastructure that connects all Americans today.
I think all of us alive today can agree we’re grateful for that infrastructure. But much of it was fiercely controversial when it was built. Railroads faced angry landowners who refused them rights of way. Highways triggered organized revolts across many American cities as homes and neighborhoods were demolished and divided. Electrification, an invisible force capable of electrocution, frightened the hell out of people and gave rise to passionate debates over safety and land use.
Many of those concerns were clearly legitimate. But imagine what America would have lost if we had caved to those concerns. Instead, we solved the problems and kept building, giving millions of people debt-free access to good wages, homeownership, retirement savings and the expectation that their children would be better off than they were.
We’re now standing at the beginning of an even greater physical buildout capable of creating the same kind of economic pathway for a new generation.
PwC estimates that investment in AI infrastructure in the United States will approach $25 trillion through 2050. For perspective, the entire world spent roughly $15 trillion in today’s dollars over a 50-year period building railroads, electrical grids and highways and employed some 30 million craft workers in the process.
America has roughly eight million construction workers today. Meeting existing demand and building the data centers and power infrastructure ahead of us will require roughly a million more by 2030. Factoring in workers who retire or leave the industry means America needs to attract roughly 500,000 new construction workers every year for the rest of the decade.
Given about 3.8 million young Americans enter the workforce each year, that represents a new pathway into the middle class for roughly one in seven of them, every year, potentially for decades.
And before you say, “Yeah, but Bill Gates says the robots are coming”: no, they’re not. I talk constantly to construction executives, union leaders and VCs investing in physical AI, and not one of them believes robots are replacing skilled craft workers anytime soon, if ever.
These are also very good jobs. Our data at Skillit shows craft pros on data-center projects earn on average 32% more than comparable construction workers, while developing scarce, transferable skills that will remain valuable long after a project is complete. They also achieve this all while avoiding the pesky debt hangover and increasingly shaky educational ROI of a four-year college degree.
Critics point out that data centers employ far fewer people once construction is complete, but that’s a silly argument. Every construction project ends. The Interstate Highway System wasn’t a temporary economic opportunity because workers finished one section and moved to the next. Craft professionals build careers precisely by carrying increasingly valuable skills from one project to the next.
So when America invests in AI infrastructure, we aren’t only building giant structures and generating compute. We’re building America’s next middle class and generating American capability. And when we lose that capability, entire industries and communities suffer the consequences for generations.
After World War II, America emerged with an enormous concentration of industrial capability. We had millions of skilled workers, factories, suppliers and businesses that knew how to build things. That capability became one of the foundations of the postwar middle class.
We then had the bright idea to move electronics manufacturing to China, textiles and apparel to Asia and Latin America, and large parts of automotive production and its supply chain out of Detroit and in return it became the poster child for industrial decline.
Now, for the first time in decades, America has almost unlimited demand for physical infrastructure. But instead of focusing on the massive employment upside for millions of Americans, we’re missing the forest for the trees and arguing about the short-term costs of building it.
The irony is that even if the NIMBY movement succeeds in stopping data center construction locally, it won’t stop AI. The data centers will simply get built somewhere else. Amazon, Google, Microsoft and others can build enormous amounts of compute abroad and sell the output right back to Americans.
The NIMBY crowd think that’s a win of course, but we’d be cutting off our nose to spite our face by repeating the same mistake we made with manufacturing. Once again, we’d become mere consumers of compute while other countries accumulate the skills, supply chains and industrial knowledge required to build it, a capability that endures long after the projects are finished and forms the foundation of a middle class.
Blocking data-center construction might win the local resource battle, but it loses America the capability war.
Perhaps fittingly for a fight over who wins that war, Napoleon once said (and Sam Altman later tweeted), “A revolution can be neither made, nor be stopped. The only thing that can be done is for one or several of its children to give it direction by dint of victories.”
The AI revolution is here and it cannot be stopped. America, however, has a singular opportunity to be its favorite child and steer it in a direction that gives future generations the shot they deserve at achieving the American Dream.
