The US AI data center boom is consuming capital, power, and people that the rest of the economy needs, and is crowding out Trump's manufacturing revival plans
Spending on factory construction is down 2.5% so far this year, while that for data centers is up almost 18%.
Context & Ripple Effects
U.S. data-center construction had already more than doubled from late 2022 levels by 2024, while builders reported shortages of parts, suitable property and power in the earlier race to add AI-serving capacity. This report puts a cross-sector cost on that buildout: factory construction is weakening as data-center spending rises.
The pressure is not limited to financing. Subsequent coverage identifies a shortage of skilled trades, including electricians, in the data-center construction surge, reinforcing the link between AI infrastructure demand and constrained industrial capacity.
First-order effects
- Capital, construction labor, power access and equipment are being allocated more heavily toward data centers as their construction spending rises, leaving less capacity for factory projects.
- Manufacturing-revival plans face a more difficult execution environment: factory builders must compete with AI projects for the same scarce inputs while factory construction spending is down.
Second-order effects
- Industrial developers and manufacturers may face higher costs or longer project timelines where data-center projects bid for land, grid connections and skilled labor.
- Power policy becomes a direct competitiveness issue for both sectors; operators had already warned that constraints on renewable power could weaken the AI buildout, while the same supply constraints also affect new factories.
Third-order effects
- If this allocation persists, U.S. industrial policy will have to treat data centers as a competing form of strategic infrastructure rather than assume AI expansion and manufacturing expansion can scale independently.
- The buildout's durability will increasingly depend on whether new power, labor and equipment capacity can be added fast enough; coverage of capacity limits and ROI concerns suggests the investment cycle has real physical and financial constraints.
The trend: AI infrastructure is becoming a broad industrial-demand shock, forcing trade-offs among compute capacity, power systems and domestic manufacturing investment.