Thanks to AI: US data center construction is accelerating while other sectors cool
Office, hotel, residential, and warehouse builds are slowing, but data center spending in the US is set to climb 23% in 2026, according to FMI Corp. That push lifts data centers to more than 6% of total non-residential construction, up from about 2% in 2023.
For real estate and construction teams, this is where demand is heading. The driver: a surge in computing needs for AI from companies like OpenAI, Google, and Meta.
Why data centers are taking the lead
These projects are bigger and heavier than typical large-scale builds. Andy Galik of Skender notes they're more than double the size, require massive electrical infrastructure, and bring thousands of workers on site-often backed by budgets that can top a billion dollars.
That scale changes the playbook. Power, schedule, and procurement decide winners.
Execution pressure: labor and materials
- Labor strain: Data centers run on tight schedules. Per Michael Gackes at ConstructConnect, labor is a primary challenge. Over the last six months, a third of US construction firms were affected by immigration laws, and a quarter dealt with contractor-driven layoffs.
- Materials inflation: Tariffs introduced under the Trump administration continue to lift costs on aluminum, steel, copper, and imported lumber-core inputs for frames and electrical systems. Skanska USA Building reports notable price increases, and 40% of construction firms have raised bid prices as a result.
Practical moves for developers, GCs, and subs
- Power-first site selection: Prioritize sites with available high-capacity electrical service and near-term utility upgrades. No power, no project.
- Front-load procurement: Lock in long-lead gear (switchgear, transformers, cabling) early. Build escalation clauses and alternates into contracts to manage volatility.
- Phase for speed: Plan shells and electrical rooms to enable staged commissioning. Deliver capacity in increments to meet AI buildout schedules.
- Industrialized delivery: Use prefab skids, modular electrical rooms, and repeatable MEP layouts to compress timelines and reduce onsite labor risk.
- Workforce strategy: Pair apprenticeship pipelines with targeted recruiting in high-demand trades. Validate subcontractor staffing plans against the compressed schedule.
- Cost control: Add material indices to bids, carry contingencies for metals, and diversify suppliers. Clarify tariff exposure upfront.
- Owner alignment: Align early with hyperscaler standards and uptime targets. Bring MEP-heavy partners and an owner's rep into design before schematic freezes.
What to expect through 2026
With spending projected to rise 23% and the category exceeding a 6% share of non-residential work, data centers are set to anchor pipelines while other asset classes cool. The constraints won't be demand-they'll be power, people, and procurement.
Teams that secure utility capacity, control schedules, and de-risk materials will capture the best margins.
Sources
If your team needs a quick primer on AI concepts driving this demand, explore curated learning paths here: Latest AI Courses.
Your membership also unlocks:
