Artificial intelligence has sparked one of the largest infrastructure races in recent memory. Data center construction is accelerating across North America as technology companies, investors, and governments compete to expand computing capacity. Research suggests that global data center capacity could nearly double by 2030, while construction costs and demand for specialized labor continue to climb.
Much of the public discourse tends to focus on investment figures and technological breakthroughs. Michael Hulst, founder of Harbor Project Management, believes a far more integral conversation is being overlooked: how this infrastructure will actually be built.
Harbor Project Management helps industrial and commercial clients navigate complex engineering and construction projects through disciplined execution and coordination. Drawing on decades of experience in power generation, construction, and project management, Hulst sees warning signs emerging as enthusiasm around AI fuels increasingly ambitious construction plans.
He says, "We're charging ahead 100 miles an hour. We're all excited about what can get done, but I don't know if we can really sit back and look at the complexity and the real risks." His concern is not with AI itself. Hulst regularly uses the technology and views it as an important advancement. His concern lies in a phenomenon he calls 'positivity bias,' which is the tendency to focus on potential outcomes while underestimating the practical challenges required to achieve them.
According to him, large infrastructure projects offer plenty of examples. Hulst points to California's long-running high-speed rail project as a cautionary tale in how optimism can outpace execution. He highlights that cost overruns, shifting timelines, regulatory hurdles, and changing political priorities can transform promising plans into expansive unfinished work.
"It drives me crazy from an engineering and project management perspective to see projects that don't get finished and billions of dollars are spent with nothing to show for it," he says.
Data centers, he acknowledges, face their own set of challenges. Hulst points to factors like power generation capacity, site selection, permitting requirements, equipment availability, and community considerations that influence project viability. Yet he believes the most pressing issue is talent.
Across the United States, the construction industry faces a significant shortage of skilled workers, with positions remaining unfilled as demand continues to rise. Electricians, boilermakers, pipefitters, commissioning specialists, and other trades are becoming critical constraints on project schedules. "Building AI requires trades, and it requires people. Those people don't exist. There aren't enough skilled trades to do it," he says.
His observations are reinforced by what he sees across the industry. According to Hulst, companies are competing for the same labor pools, while some major contractors have secured critical equipment years in advance simply to ensure availability. At the same time, developers are announcing large-scale projects that require enormous amounts of specialized expertise.
Project discipline becomes especially important in that environment, he notes. Hulst believes many delays originate long before construction begins, often during planning and budgeting stages. He explains, "I think it's a lack of transparency, or a lack of honest assessment of risks and potential delays. You can't build faster, more complex, bigger, and cheaper at the same time. Something has to give."
Managing projects for years taught him that successful execution depends on realistic schedules and labor assumptions, and a willingness to communicate difficult truths early. Investors and owners may not always welcome those conversations, but Hulst argues they are essential.
"If we aren't really honest about what can get done and what's the real time frame, there will be a lot of money spent on unfinished projects," he says.
Solving the labor challenges, in Hulst's view, ultimately requires looking much earlier in the pipeline. He believes Western education systems have spent decades presenting college as the primary path to success while undervaluing skilled trades.
"We need to seriously invest in our youth and young people getting into the trades," he says. "Shop classes aren't for kids who can't go to college. They're for people with skills."
Electricians, boilermakers, welders, pipefitters, and other tradespeople, Hulst notes, will play a direct role in shaping the AI economy. He argues that apprenticeships, trade education, and workforce development deserve the same attention currently being given to data center investment announcements. "You can't have AI without welders," he says.
As AI infrastructure expands, Hulst believes success will depend on viewing the challenge through a wider lens. The lens, he adds, encompasses mindful financing, technology, power, skilled labor, and education. Progress, he posits, only happens when every piece of that equation receives an equal level of attention.
