Construction Spending Explained

• New Starts + Existing Backlog generate Spending
• Spending = Revenue
• Revenue includes inflation which adds nothing to volume
• Revenue – Inflation = Business Volume

Construction spending fell slightly in April, down 0.4% from March. Spending has fallen slightly each of the last 3 months, but total spending is still at/near an all-time high with the seasonal rate at $2,200 billion. The forecast predicts spending will increase to 3% growth by year end.

Construction Forecast Update – Data Centers shows the largest % growth for 2025, forecast +33%. Manufacturing is still the largest $ contributor ($223bil/yr) to nonresidential bldgs total spending ($772bil/yr), but has fallen 6% in the last 5 months.

Peak manufacturing construction spending was posted from Sep thru Dec 2024. The avg of 1st 4mo of 2025 is down 4% from that peak. By Q4’25, avg spending will be down 10% from peak.

BTW, this is totally normal. We are beginning the tail end of an above normal huge influx of new manufacturing projects that started over the last 3 years, and the spending curve is beginning the downhill slope. Spending will continue to fall for the next 3yrs.

Data Center construction spending has not yet hit peak. Data Centers are continuing on a phenomenal streak of +45% growth in 2023 and +56% in 2024 and now 33% in 2025. The avg of 1st 4mo of 2025 is up 39% from same 4mo 2024. Spending will finish the year almost 20% higher than today. 2025 forecast +33% over 2024. Currently projecting peak spending end of 2027, or later.

Headwinds could slow new starts growth. Many economists predict current trade impacts will slow overall economic growth. That in turn could slow capital expenditures, which, in this case, is new construction starts.

Any capex pause could reduce all Data Center numbers. However, starts are up 400% since 2020 and could finish 2025 up 500%. Would take a lot of canceling or delaying to collapse these numbers. (This is going to first appear in construction starts, “firms pausing or delaying capex.” It’s already started with Data Center).

What’s propping up construction spending growth? Here’s the top growth markets.

• Market——1yr/%/$ growth ——3yr growth
• Educational 1yr/+8%/+$10bil +3yr/+30%/+$30bil
• Data Centers +56%/+$10bil +286%/+$18bil
• Public Utilities +11%/+$9bil +58%/+$38bil
• Power +10%/+$14bil +24%/+$29bil
• Highway +4%/+$6bil +40%/+$40bil

Data Centers far and away takes the prize for highest % growth, but Data Centers is only 1.7% of all construction spending. Power is 7%, Highway is 6.6%, Educ is 6.3%, Pub Util is 4.4%

Manufacturing is notably absent from the above list, because after 3 outstanding years, Mnfg is no longer contributing growth. Mnfg spending is beginning to taper off. Mnfg is 9.3% of construction spending. The only market over the last year, or 2 or 3 years, with more $ spending than Manufacturing is Residential.

Mnfg 1yr/+20%/+$39bil 3yr/+284%/+$150bil

Mnfg 2025 forecast -10%/-$24bil

What’s holding spending growth back?

6-17-25 When May construction starts are reported later this month, I’m expecting an overall decline and a lower forecast. There are already reports of pauses in manufacturing facilities and data centers. Hiway and Public Utilities are probably immune from cuts but Power may see some reductions. Education and Healthcare are questions. Residential construction expected down slightly. Housing permits continued a downhill trend in April for the fourth month in a row. KB and Lennar report market pricing is down slightly. # of homes on the market is increasing.

This next plot shows the number of workers required to put-in-place $1 billion of construction in 1 year. Except for Nonbldg Infra, which has remained relatively flat over time, it requires more jobs to put-in-place $1bil today than it did 10 years or 20 years ago. Total construction workforce (8,300,000) divided by # of billions$ put-in-place (2,200 billions$/yr) is the simplest way to show the decline in construction productivity. Results here broken out for major sectors.

Construction Jobs Total hours worked peaked in March, now down 0.4%. For May, jobs increased by 4,000 (<0.1%), but unemployment dropped from 5.6% to 3.5% (175,000). That would mean that 171,000 workers dropped out of the workforce.

Steel Tariffs 50%

50% of nonres bldgs are structural steel. Str Stl is 10% of total bldg final cost. Nonres bldgs construction spending = $770bil/yr., $385bil on SS nonres bldgs., $38.5bil on str stl

Struct Steel material is only 25% of steel total contract cost, so 38.5 x 25% = only $9.6 bil is mtrl used in SS nonres bldgs. So at 50% tariff = $4.8bil added cost to nonres bldgs total spending. (This assumes ALL steel increases in cost).

Above is structural steel only. When including all other steel used in a building, (rebar, studs, frames, etc), steel is 15% of total bldg cost. So added cost would be $7.7bil.

And that is just nonres bldgs. MF Rsdn uses a little steel, but Nonbldg markets, Power, Highway, Transport, Pub Util adds about another $5bil.

So, steel tariffs, IF ALL STEEL WERE TO INCREASE, adds inflation to total Nonres Bldgs and Nonbldg. Half of all Nonres Bldgs use structural steel, so inflation to SS bldgs is 4.8/385, or 1.25% on structural steel bldgs. If looking at the macro view, inflation over the nonres bldg sector, then the $4.8bil increase would be divided by the total nonres spending, or 4.8/770 = 0.625%. All other bldg steel, applied to all nonres bldgs, added another 2.9/770 = 0.4%

Nonbldg Infrastructure markets, if it is a SS building, add the same 1.25% + 0.4%. But Public Works and especially Bridge construction can add significantly more. If domestic producers also raise pricing to follow suit with tariffs, as expected, this is what happens to total Nonres inflation.

US imports 30% of steel it uses. 40% of all steel is used in construction. If 30% is balanced across all types of steel, then 30% of constr steel is imported. (It would take some concentrated effort to determine % imports for each of the individual steel uses.) IF ONLY IMPORTED STEEL WERE TO INCREASE and no domestic manufacturers raise prices, 30% of building steel increases in cost. That is not likely at all. But if so, tariffs would add only 0.4% to nonres bldgs and about 0.1% for all other steel.