The technology promising to automate everything cannot build its own buildings. And the industry tasked with building them is running out of the people who know how.
The data centers powering the artificial intelligence revolution are being built right now, across the country, on tight schedules and tighter labor budgets. Every one of them depends on electricians, ironworkers, pipefitters, and concrete crews. And the people who do that work, who carry 30 years of job site knowledge in their heads, are walking out the door faster than the industry can replace them.
That is the irony nobody in Silicon Valley is talking about: the technology promising to automate everything cannot build its own buildings.
"It's a little bit ironic in a way, right?" said Kris Lengieza, Vice President and Global Tech Evangelist at Procore Technologies. "All of this demand is being put on us by something that is going to automate a lot of things, but we can't actually build it as fast as we want to, because we don't have the people to go build it."
Lengieza is not speculating. Procore's construction management platform processes more than $1 trillion in annual construction volume. What he sees across that data is an industry staring down a demographic cliff with no clear plan for what happens when it goes over.
As of late 2025, approximately 2,788 data centers had been announced or were under construction across the United States, according to the American Edge Project. Total spending on U.S. data center construction starts hit an estimated $77.7 billion in 2025, a 190% year-over-year increase, according to ConstructConnect. The five largest hyperscalers combined planned to spend roughly $660 billion to $690 billion on infrastructure in 2026 alone.
All of that is waiting on a workforce the industry cannot staff fast enough.
According to the National Center for Construction Education and Research, roughly 41% of the U.S. construction workforce is projected to retire by 2031. The Associated Builders and Contractors projects the industry will need 349,000 net new workers in 2026, rising to 456,000 in 2027. A Deloitte analysis estimates the shortage of skilled craft professionals could exceed 2 million by 2028.
The Associated General Contractors of America reports that 45% of firms are already experiencing project delays directly caused by labor shortages, and 63% say a project has been postponed, scaled back, or canceled in the past six months.
But the statistics, as stark as they are, do not fully capture what is actually being lost.
A superintendent or project manager with 30 years of experience does not check notes when a problem surfaces. They pause, look around, and draw on a catalog of situations accumulated across decades of hard days. That pattern recognition is what makes a veteran worth three of anyone else on the site. It is also what vanishes when they hand in their badge.
"There is so much information stored in their brains about situations they've come into contact with before, problems they've had to solve before," Lengieza said. "That's what makes some of those most senior people extremely valuable. The judgment that they have, and the pattern recognition."
What makes the knowledge transfer problem harder to solve is timing. Construction has been relentlessly busy for a decade. The traditional apprenticeship model, that slow passing of expertise from veteran to newcomer, has been squeezed out by the pace of the work itself.
Workforce shortage discussions in construction tend to focus on headcount. What receives less attention is what the shortage is doing to the workers still on the job.
Male construction workers are 75% more likely to die by suicide than the general population, according to OSHA. The AGC has documented that 56 out of every 100,000 male construction workers died by suicide in 2021, a rate roughly four times the national average and nearly six times the rate of all construction fatalities combined.
A 2025 national survey of more than 2,000 construction workers and executives commissioned by design-build firm Clayco found that 64% reported experiencing anxiety or depression in the past year, up from 54% in 2024 and nearly three times the rate in the general population. Nearly half said they would feel ashamed to discuss mental health issues on the job.
"We will put more stress and strain on the people in this industry than we need to or than we ever should," Lengieza said. "We already have a mental health crisis. It will get worse if we don't figure out how to solve some of this workforce challenge."
There is a practical consequence beyond the human cost. Data centers built under the strain of an understaffed, burned-out workforce do not perform at full capacity. The workforce shortage does not just delay the AI buildout. It quietly degrades it.
Procore's answer to the knowledge transfer problem begins with a simple observation: the workers with the most to share are the least likely to stop and document it.
A worker who just finished a 12-hour day, handed a form and asked to fill it out, will provide whatever gets them out the door. The same worker, asked to sit down for five minutes and talk through what happened that day, will not stop talking.
"If you go and ask a superintendent to do a daily log and you give them a form, you're going to get probably the minimal amount of information required," Lengieza said. "But if you go and sit down with that same superintendent for five minutes and ask them what happened on the job site that day, you will get more information than you can even imagine."
Procore is building toward that gap. AI-powered voice tools allow workers to debrief naturally, talking through the day with a digital assistant that structures and logs everything said. The institutional memory that would otherwise retire with its owner starts accumulating somewhere it can be retrieved.
Every record in Procore is, in Lengieza's framing, a decision preserved. The platform processes more than $1 trillion in construction volume annually, a repository of how the industry actually solves problems growing in real time.
The fear on job sites is that AI is coming for the jobs. Lengieza describes how automation reaches a workforce in three stages, and the framework is less threatening than the headlines suggest.
The first stage is the most immediate. The construction industry runs on 60- to 70-hour weeks. Automating the administrative load, the paperwork, the repetitive documentation that buries workers, gets people closer to 40 hours. That change alone would meaningfully reduce the mental health toll driving experienced workers out of the industry ahead of schedule.
The second stage is about quality. Freed from administrative burden, workers can do what actually improves projects: more time walking the site, more schedule analysis, more inspections that catch problems before they become disasters.
The third stage is the one generating anxiety, and where Lengieza pushes back hardest. Some efficiencies may allow projects to get done with leaner teams. But the math is not zero-sum.
"Even that third bucket, that seems like 'we're actually taking away workers.' We're actually just affording us to keep up with the demand that the industry and the global economy is putting on us right now," he said.
Data centers. Energy infrastructure. Semiconductor fabs. Manufacturing coming back onshore. Housing. The construction industry has never been asked to build this much simultaneously. Every efficiency that automation creates is not a worker displaced. It is a project that can finally get staffed.
Beyond retention, construction faces a harder problem: attracting workers who have never considered it.
According to a March 2026 National Association of Home Builders study, only 6% of young adults want a career in the construction trades, though that figure has doubled since 2016. At a median wage of $90,000 or above, 32% of those not currently interested said they would reconsider. The compensation is competitive. The industry's story about itself has not caught up.
Bridgit data shows Gen Z's share of the construction workforce grew from 6.4% in 2019 to 14.1% in 2023. The trend is moving in the right direction. The pace is not.
Lengieza sees technology as the pitch to accelerate it. The generation entering the workforce grew up with technology and is drawn to careers that are stable, intellectually demanding, and consequential. A job site that uses AI tools to solve complex problems in real time looks very different from the one the previous generation described at the dinner table.
Lengieza has carried a particular vision for about five years. It is less about the technology and more about what a worker's day feels like.
Workers get a briefing on the way to the job. The materials are staged. The trades are sequenced. There is nothing in the way when they arrive. Instead of one step forward and two steps back, the whole system moves forward together.
"When people can show up and know that they're going to be effective, the whole system moves together faster," he said. "The workforce is healthier mentally, physically. You'll have less accidents, less physical and mental strain. And then they'll get home and they won't talk about the things that went wrong that day. They'll talk about what they built."
The irony at the center of this story has not gone away. The most powerful artificial intelligence ever developed needs someone to wire the building it lives in. What happens over the next five years, whether institutional knowledge gets captured before it retires, whether younger workers step in to learn it, whether the industry finds a way to slow the attrition long enough to survive the transition, is not a technology problem.
It is a human one. And right now, it is the only bottleneck that actually matters.
If you or someone you know in the construction industry is struggling, the 988 Suicide and Crisis Lifeline is available 24 hours a day, 7 days a week by call or text. Additional resources are available through the Construction Industry Alliance for Suicide Prevention at preventconstructionsuicide.com.
