Data center construction is moving in a very different direction than the rest of U.S. construction.
In July 2026, U.S. data center construction spending hit an annualized rate of $75.2 billion, up 57% from a year earlier.
That is a huge move on its own. It looks even bigger next to the broader construction market.
Total U.S. construction spending was down 3.8% year over year. Private construction was down 5.5%. Office construction outside of data centers was down 11.3%.
But because data centers are now such a large part of the office category, total office construction was still up 21.3%.
That tells you how much the AI buildout is changing the shape of the physical economy.
The market talks about AI through chips, cloud revenue, model performance, and software adoption. But the buildout also shows up in steel, concrete, land, grid connections, cooling systems, backup power, electrical equipment, and construction labor.
That makes data centers one of the cleaner ways to see AI demand outside of Big Tech earnings calls.
The investor question is how long this can keep running.
If AI infrastructure spending keeps growing, the beneficiaries may extend well beyond semiconductors. Power equipment, grid services, cooling, engineering firms, utilities, landowners, and construction suppliers all become part of the story.
But the same chart also raises the risk side. A boom this vertical usually brings bottlenecks, higher costs, permitting fights, power constraints, and eventually the question of whether demand can justify the amount of capacity being built.
For now, the data is pretty clear: while much of construction is slowing, data centers are still in expansion mode.
https://www.census.gov/construction/c30/current/index.html https://www.federalreserve.gov/econres/notes/feds-notes/the-ai-buildout-and-the-economy-publicly-available-data-to-assess-ais-impact-20260717.html