Blog

Remember all the excitement over “coding for kindergartners?” The belief that coding was a surefire path to a future job inspired everything from coding camps for high schoolers to coding classes for seniors.

Consider some recent headlines. “Goodbye, $165,000 Tech Jobs. Student Coders Seek Work at Chipotle,” read an August one in the New York Times. “The Computer-Science Bubble is Bursting,” warned a June headline in The Atlantic.

Coding was never going to become a dominant employment opportunity. The number of Americans working as farmers or ranchers is seven times greater than those toiling as coders or computer scientists. Both farming and coding constitute a tiny sliver of the total workforce, and while both excite certain demographics, neither is a primary vector for future employment. That’s because such specializations, while important, of course, will continue to be niche jobs.

In fact, coders and computer programmers, including those with AI skills, account for less than 1 percent of all STEM jobs. But the surprise for our times is how those coders and computer scientists are driving a massive increase in demand for construction workers.

In July, the White House released its manifesto, “Winning the Race: America’s AI Action Plan,” at a high-profile event that drew a who’s who of tech and industrial titans. It was a call-to-arms to “build and maintain vast AI infrastructure and the energy to power it,” and in that report we find the signature line: “Build, Baby, Build!” To do that, America will need construction workers, a cohort that constitutes about one-fifth of all skilled trade workers.

It’s hard to ignore the irony that ever-more clever code, especially AI, has resulted in a frenzied buildout of data centers. The rate of private sector spending on data center construction has surpassed construction spending on all other U.S. commercial buildings. It’s also more than current total spending on manufacturing facilities or power plants.

This surge carries big implications for jobs. All major construction projects draw from the same pool of skilled workers. Every data center that goes up means fewer workers available to build what many policymakers see as equal or higher priorities: reshoring factories, mines, shipyards, repairing roads and bridges, and new homes.

How did the world of bits and bytes become the second-largest source of demand for construction workers, trailing only residential housing? Two sets of facts hold the answer.

First, code runs on machines. Some are tiny, like smartphones and earbuds. But the cloud—thousands of data centers linked together across the globe—is where the real computing power lives. It enables everything from streaming sports to navigating cars, from banking to scientific research. Global sales of cloud services, the code in the Cloud, are nearing $1 trillion annually and climbing fast.

Those cloud data centers—enormous buildings packed with silicon processors—are what Google calls “warehouse-scale computers.” Each one costs as much as a skyscraper to build and consumes even more steel and concrete. It should come as no surprise, then, that erecting a warehouse-scale computer takes about as much labor as putting up a skyscraper or laying down a highway.

That brings us to the second set of facts. Most big projects—ships, bridges, power plants, or data centers—draw on the same kinds of skills: equipment operators, electricians, welders, carpenters, and the like. A rule of thumb: every $1 billion in construction spending requires 5,000 to 10,000 workers.

But, as Mike Rowe of Dirty Jobs fame points out, our culture “has glorified the ‘corner office job’ while unintentionally belittling the jobs that helped build the corner office.” Given the stampede to build data centers, we can revise Rowe’s aphorism: our culture has glorified coders while unintentionally belittling the jobs that helped build the machines that make coding possible.

The United States faces a shortage of skilled-trade workers. Those in the skilled trades sector are, on average, older than the rest of the workforce, and fewer young people are entering the pipeline. To Build, Baby, Build! as the White House exhorts, we’ll need lots more construction workers. What’s to be done?

Policymakers have only three tools at their disposal. First, they can encourage many in the aging workforce to delay retirement. Second, they can fill the pipeline with more people learning the skilled trades. Third, and probably most politically challenging, they can import skilled labor. Fortunately, technology itself plays a big role in the first two.

First, the aging workforce. Many on the front edge of the so-called silver tsunami are leaving the trades for a simple reason: construction work is hard. These are Mike Rowe’s dirty jobs—physically demanding and sometimes dangerous.

Innovators have perfected and are commercializing semiautomated and fully automated machines that augment human strength and reduce the heavy physical lifting. Consider, for example, the wearable exoskeleton that’s no longer the stuff of science fiction. Commercially available systems can minimize stress on specific joints and muscles or even the whole body.

Of course, with automation, the go-to focus has been on robo-trucks—many such have long been deployed at mine sites, with more becoming commercially available for public roads—which can free up drivers to be re-skilled for other construction tasks. Industrial automation is a long-standing solution for amplifying hard labor. Ask farmers. However, there’s far less automation across non-farm industries than most imagine, especially when it comes to robots. While nearly every construction-equipment maker is developing semiautonomous or fully autonomous machines, the barrier to adoption is always the need to realign processes and train workers. For the foreseeable future, most automation will still depend on skilled human operators.

AI can help—not only by making machines more capable, but by delivering faster, easier, and better training. Commentators who warn that AI will eliminate jobs often have it backward. True, AI can sometimes cut in half the labor needed for a task. But just as often, AI can make existing workers twice as productive—a particularly powerful advantage given the shortages in the skilled trades.

That brings us to the second tool for expanding the skilled workforce: convincing more people to pursue a career in the trades. Here, policymakers should tap into the vast potential workforce among young men released from prison for nonviolent offenses by expanding inmates’ access to vocational education. Only a small fraction of this group currently receives such training.

And to train more would-be tradesmen in general, we need to make training more effective—and more interesting. Technology can help here, too. Leading construction-equipment makers already use virtual reality and augmented-reality systems for their training simulators. Tests show VR training significantly improves users’ training-completion and employment outcomes.

Virtual-reality flight simulators have been the mainstay of the skills-centric aviation industry for a long time. While the heavy equipment “operator training simulator” market is already north of $10 billion annually, the technology remains greatly underutilized. Now, AI is taking simulators to the next level.

The third option, importing skilled labor, is a political third rail. But it shouldn’t be dismissed. One could imagine, for example, a visa program aimed at the trades—a “Steel Visa”—to fast-track qualified foreign workers into the U.S. Selectively expanding the labor pool this way could help avoid delays or cancellations of major construction projects and also mitigate inflationary wage spikes.

Expanding and enhancing the skilled labor pool will require both technology and capital. Instead of launching a massive federal program, Washington should expect deep-pocketed tech firms to provide both.

The American economy will continue to need students to pursue both STEM and skilled trade jobs, each of which represents about one-fifth of the U.S. workforce. That’s likely to be the case for a long time yet. And that’s a good thing.

Photo by Eli Hiller/For The Washington Post via Getty Images