Forbes examines the scale of the AI data center buildout, which the article says could cost as much as $10 trillion in the U.S. This would surpass the railroad boom of the late 1800s as a share of the economy as early as next year. The piece tracks where the money is going across land, power, water, and cement.

DataBank CEO Raul Martynek is quoted on the timing realities facing developers. Because data centers follow multi-year development cycles, near-term power is already spoken for. DataBank is acquiring land today for facilities that will not be ready for service until 2030 or later.

“We’re planning for 2030 or beyond, literally acquiring land now where we think the ready-for-service date is 2030, 2031, 2032.”
Raul Martynek, CEO, DataBank

Martynek also points to skilled labor as a pressure point, noting that electricians cannot be produced overnight. He adds that delivery times for key electrical equipment such as generators have stretched from about 12 weeks to nearly two years.

The article sets those constraints against rising electricity demand, growing scrutiny of water use, heavy debt financing, and an ongoing debate over how much of the planned pipeline will actually get built.

Read the full article at Forbes.