The AI boom is about to meet America's old-fashioned infrastructure bust

Bob Hellman's Washington Examiner op-ed argues that America's artificial-intelligence boom is on a collision course with the country's inability to build physical infrastructure on schedule. He opens with a striking figure. The top four tech companies have committed roughly $700 billion to AI data-center construction in 2026 alone. That money, he notes, only becomes AI infrastructure if it can be paired with traditional infrastructure such as power, water, and permits, and Big Tech is now discovering the hard way that the United States cannot deliver.

Hellman cites Gartner's projection that 40 percent of AI data centers will be constrained by power availability by 2027. He also observes that between 30 and 50 percent of the facilities scheduled to open this year are already stalling. Grid-connection waits are now measured in years rather than months, strained water systems are being asked to cool thousands of server racks, and permitting timelines stretch so far past capital-deployment schedules that investors watch their cost of carry climb while shovels sit idle.

None of this, he argues, is new. Hellman describes what he calls the oldest pattern in American infrastructure. A need is identified, commitments are announced, and then projects die slowly inside environmental reviews, permitting fights, funding paralysis, and political indecision while costs compound. He backs the claim with McKinsey's estimate that roughly $1.5 trillion in proposed infrastructure is trapped in permitting bottlenecks, producing $100 billion to $150 billion in annual economic losses. He notes that construction costs have risen 70 percent since 2020, so every year of delay costs more than the last. Even the American Society of Civil Engineers' 2025 report card, a grade of C and its best since 1998, still projects a $3.7 trillion investment gap that is growing rather than shrinking. He points to the $42 billion BEAD rural-broadband program, which was allocated in 2021 but is only now beginning implementation after years of delays and rule changes, as a preview of the trajectory AI data centers are on.

Hellman is careful about the diagnosis. The instinct is to blame regulation, but he argues that the deeper problem is political fear. Officials at every level, in red states and blue, have learned that ordering another study carries less risk than delivering a result, because saying yes to a data center means owning the trade-offs on water, power, noise, and land, while calling for more review costs no one their job.

The consequences, he warns, extend far beyond delayed data centers. If power, water, and permits cannot be secured on predictable timelines, AI investment, and ultimately AI leadership, will migrate to countries willing to decide faster, leaving the United States at risk of becoming a designer of AI systems that are trained, operated, and monetized elsewhere. His prescribed fix is accountability: treating timelines as binding commitments, attaching decision-making authority to decision-making responsibility, and imposing real consequences for delay without cause. As a model, he cites Dallas's LBJ Express, the largest private infrastructure project in Texas history, which opened three months ahead of schedule because private investors owned the deadline and bore the cost of missing it. When delay stops being an abstraction and becomes a number on a balance sheet with a name attached, Hellman argues, projects get built, and the United States can only keep its technological edge by delivering functional public infrastructure and clearing the way for private AI infrastructure alike.