US nonresidential construction spending fell 0.2% in March 2026, the latest in a run of soft months. The headline understates what is actually happening: data center work is up 34.3% year over year, while almost everything else, especially manufacturing, is shrinking. Contractors are operating in a two-speed market where one segment is paying for the weakness of every other. Where your backlog sits now matters more than the topline number.
The latest spending data from the US Census Bureau looks dull at first glance. National nonresidential construction spending dropped 0.2% in March 2026 on a seasonally adjusted annualized basis, settling at $1.244 trillion. A small number, easily ignored.
That is the wrong takeaway. Underneath the headline, the industry is splitting in two. On one side, data centers are running hot, fuelled by hyperscaler capex that touches three quarters of a trillion dollars globally this year. On the other side, manufacturing, public projects, and most commercial categories are quietly grinding lower. The aggregate looks flat because one boom is masking a broader slowdown.
For contractors, developers, and tech operators selling into the AEC industry, the implication is awkward but useful. The market you are reading about is not the market you are bidding into, unless your pipeline happens to sit on top of an AI campus.
What the March numbers actually said
A small drop, a wider weakness
According to the Associated Builders and Contractors analysis of Census data, nonresidential spending was down on a monthly basis in 9 of the 16 tracked subcategories. Both public and private nonresidential spending fell 0.2% in March. Private sector activity is now down more than 2% year over year, which is the more telling figure.
ABC Chief Economist Anirban Basu put it bluntly. A large portion of the drop is steady manufacturing weakness, but the softness is becoming more widespread. With one obvious exception, there are few sources of momentum across the market.
That exception is doing a lot of heavy lifting.
The data center distortion
One sector is keeping the topline alive
Data center construction is up 34.3% year over year, the only major category showing strong growth. To put a number around what is happening, US data center construction starts hit roughly $77.7 billion in 2025, a 190% jump year over year, according to ConstructConnect tracking cited by Equipment World. Average monthly spending on starts climbed from around $500 million in mid-2021 to $6.5 billion by December 2025.
The Associated General Contractors of America’s 2026 Outlook Survey backs this up. A net 57% of contractors expect data center spending to rise this year, the most bullish reading in the entire survey and well ahead of power construction at 34%. McKinsey projects $7 trillion in global data center capex by 2030, with over 40% of that landing in the US.
Strip the data center category out of the national numbers and the contraction looks far less ambiguous. The 0.2% headline becomes the polite version of a more uncomfortable reading.
“With the exception of the ongoing boom in data center construction, there are few sources of momentum.” – Anirban Basu, ABC Chief Economist
Manufacturing was the engine, now it is the drag
From boom to ballast
Manufacturing construction was the surprise growth story of the post-pandemic cycle, climbing from around $75 billion in 2020 to roughly $225 billion by 2025, on the back of CHIPS Act incentives, EV plants, and reshoring projects. That tailwind has turned. FMI expects manufacturing put-in-place construction to finish 2025 down 4.7% on the year and decline another 3.7% in 2026. Dodge Construction Network’s starts data shows manufacturing fell more than 16% in the 12 months ending October.
When the category that was carrying the market loses its grip, the rest of the floor starts looking thin. Public spending is soft. Commercial and institutional backlogs have been weaker for months. ABC’s January reading put overall construction backlog at a four-year low, before partially recovering to 8.6 months by April.
The pattern is consistent across the survey data. Backlog strength is concentrated in data centers, heavy industrial work, and infrastructure, while traditional commercial and institutional segments are thinning out.
Why contractors still feel okay about it
Personal optimism vs. peer pessimism
Here is the curious part. ABC’s Construction Confidence Index readings for sales, profit margins, and staffing all sit above 50, the threshold for growth expectations over the next six months. Contractors as a group remain optimistic about their own books, even as they look at the wider market and see weakness.
Basu has flagged the gap before. Only around 13% of contractors expect their own sales to decrease in the next six months, yet 46% expect other contractors’ sales to fall. It is a striking split, the kind of disconnect that usually resolves one way or the other within a couple of quarters.
The optimism is not irrational. Backlog at firms with over $50 million in annual revenues has been rising, while smaller contractors have been hit harder. If you are at the top of the food chain, working on a hyperscale campus or a multi-billion dollar industrial site, the macro looks fine from where you are sitting. If you are a mid-market commercial contractor in a softening metro, it looks quite different.
Segment-by-segment view
Where the work is, and where it is not
Below is a snapshot of how the major nonresidential categories are tracking, pulling from ABC, AGC, FMI, and Dodge data. This is the picture the topline averages away.
| Segment | Direction | Key data point |
|---|---|---|
| Data centers | Strongly positive | +34.3% YoY spending; net 57% of contractors expect growth in 2026 (AGC) |
| Power | Positive | Net 34% of contractors expect growth; lifted by data center grid demand |
| Infrastructure / heavy industrial | Holding up | Backlog holding firm per ABC; supported by IIJA pipeline |
| Manufacturing | Negative | Down 4.7% in 2025; another 3.7% drop forecast for 2026 (FMI); starts down 16%+ (Dodge) |
| Commercial | Mixed to soft | Backlog weakening; demand uneven outside data-adjacent work |
| Institutional | Soft | Backlog has not recovered; flat starts over the past year |
| Public | Negative | Down 0.2% in March; momentum easing across federal and state projects |
The split is not subtle. Two categories are doing the work of carrying the headline, and almost everything traditional is grinding sideways or down.
What this means if you are running a business
Practical takeaways for contractors and tech operators
For contractors, the strategic question is whether to chase the boom or insulate against it. Data center work is technically demanding, has aggressive timelines, and rewards firms with strong self-perform capability and tight trade partnerships. The opportunity is real, but the field is already crowded with Turner, DPR, Mortenson, Kiewit, and Crusoe, plus a long list of regional specialists scaling into the niche. Late entry is harder than it looks.
For ConTech operators, the segmentation matters for go-to-market. A buyer running data center delivery has a fundamentally different problem profile, faster cycles, higher margins, more tolerance for new tooling, than a mid-market commercial contractor protecting 2-3% margins through a slow patch. Pitching the same tool the same way to both is leaving money on the table.
For anyone modelling 2026, the prudent read is to assume the data center boom continues to obscure broader weakness for several more quarters, then plan for the possibility that the rest of the market does not snap back quickly. If tariff pressure, immigration enforcement, and elevated financing costs persist, the funding hangover ENR flagged becomes a base case, not a tail risk.
FAQ
How much did nonresidential construction spending fall in March 2026?
Why is data center construction such an outlier?
Is the manufacturing construction boom over?
If the market is weakening, why are contractors still optimistic?
What share of construction confidence depends on data centers?
Are tariffs affecting the spending picture?
What should mid-market contractors do if they cannot easily pivot into data centers?
Trusted Sources
