Data Centers Are Carrying US Construction. What Happens When the Boom Slows?

Contractors working on data centers are having the best year of their careers. Backlogs are longer, crews are bigger, and announced projects keep growing past a trillion dollars. If your firm is plugged into that pipeline, 2026 looks like a boom year, full stop.

Pull data centers out of the picture and the story flips. Private nonresidential spending excluding data centers fell 7.9 percent, according to Associated Builders and Contractors. Only a narrow slice of the industry, around 13 percent of contractors, actually has data center work booked. Everyone else is reading a much softer market.

That gap raises a question worth sitting with if you build, fund, or sell into construction: what happens to the rest of the industry if this one source of demand slows down?

The Two Economies

One narrow slice of construction is doing the heavy lifting

The numbers back up what contractors are feeling on site. Associated Builders and Contractors found data center construction spending up 46 percent year over year in June, while private nonresidential spending excluding data centers fell 7.9 percent over the same period. Total nonresidential spending still crossed 1.2 trillion dollars for the month, but that headline figure hides how uneven the growth actually is.

The backlog data tells the same story from a different angle. Firms with data center contracts reported roughly 11 months of work on the books, compared with 8.5 months for firms without that work. And only 13 percent of contractors surveyed had data center projects at all. A narrow group is carrying a disproportionate share of the industry’s momentum, and that concentration is exactly what makes the moment worth watching closely.

Building Around the Boom

Tech and finance are now funding the workforce themselves

The labor market has not kept pace with the pipeline, so some of the biggest names in tech and finance have started building the workforce directly. Meta launched America’s Workforce Academy in June with a 115 million dollar first year commitment, offering free training and a guaranteed job to graduates.

BlackRock made a similar bet. Its Future Builders program is committing 100 million dollars to help train 50,000 workers in skilled trades over the next five years. We covered the mechanics of that shift, and what it means for contractors competing for the same crews, in our piece on why big tech is suddenly paying to train electricians.

When two of the largest companies on the planet start funding trade schools instead of just writing checks for buildings, that says something about how long they expect this cycle to run. It also means the boom is starting to reshape the workforce around the assumption that demand keeps climbing.

The Fed Is Watching This Show Up in GDP

Business investment just posted its strongest quarter in years

This is no longer just a construction story. The Federal Reserve’s July Monetary Policy Report found business fixed investment grew at an 11 percent annualized rate in the first quarter of 2026, and the Fed said most of that strength was connected to building the infrastructure that supports AI services.

Fed Governor Lisa Cook has been tracking the scale of what is still coming. She noted that companies have announced more than 1.5 trillion dollars in data center plans, with only a small share actually built so far. Cook pointed to rising prices for chips, specialty construction labor, and other high tech equipment as an early sign of the pressure that pipeline is already creating.

Vice Chair Philip Jefferson has raised a related concern about timing. If the demand created by AI investment shows up faster than the productivity gains AI is supposed to deliver, that mismatch can push inflation higher, which in turn keeps interest rates higher for longer. That matters well beyond data centers, since higher rates make every other type of construction more expensive to finance.

The Price Paradox

AI keeps getting cheaper, so why is everyone still building

Here is the part that does not add up at first glance. Enterprise AI costs are falling fast. Blended token prices, the per use cost of running AI models, dropped roughly 67 percent between Q1 2025 and Q1 2026, based on an analysis of 2.4 billion enterprise API calls. If AI keeps getting cheaper, why keep building infrastructure at this pace?

Falling prices can cut both ways. Cheaper AI could mean more businesses can afford to use it, which pushes total usage up even as the price per use goes down. The real question is not whether AI gets used more. It is whether that additional usage generates enough revenue to justify the money already committed to building the infrastructure behind it.

What Happens If It Just Slows

Lessons from two earlier booms

AI does not need to fail for this to become a problem. It just needs to grow more slowly than the current pace of construction assumes. A slowdown could look small at first, a few delayed projects, some paused orders, hiring that cools off. But the debt, leases, and long term contracts already signed do not pause with it, and that is where a construction slowdown can turn into a financial one.

The internet and the railways both show why that distinction matters. Both technologies were real and both eventually transformed the economy, but both also attracted more capital than could immediately be put to productive use. The dot com crash destroyed enormous amounts of investor money without the internet itself failing. Being right about the technology and being right about the investment are two different bets, and the data center boom is currently asking construction firms, lenders, and investors to make both at once. We looked at the broader softness building underneath this boom in our earlier piece on what is keeping US construction afloat right now, and that gap has not closed since.

Related Articles
Data Centers Are Keeping US Construction Afloat, But the Rest of the Market Is Struggling
Why Big Tech Is Suddenly Paying to Train Electricians