AI Data Centers Now Drawing City-Scale Power, Forcing a New Playbook for Contractors
According to Engineering News-Record, AI data centers have crossed into territory that no longer fits the traditional “building” category. Industry groups are now treating them as city-scale infrastructure, and the construction and field services sectors are being forced to adapt fast.
The trigger is a framework launched June 10 by the National Electrical Manufacturers Association (NEMA), ASHRAE, and Pacific Northwest National Laboratory (PNNL): the AI Data Center Energy Performance Framework. The document is designed to guide planning, siting, design, commissioning, operations, retrofit strategies, and grid-interactive design for AI facilities. It also addresses energy storage, microgrids, demand flexibility, and liquid cooling, technologies that are becoming standard on large-scale data center builds.
Background
The scale numbers behind this shift are difficult to ignore. As ENR reports, NEMA’s Grid Reliability Study, prepared by PA Consulting, projects that data center electricity consumption will increase by roughly 300% over the next decade, accounting for 38% of net U.S. electricity demand growth through 2037.
To put the power demand in human terms: Patrick Hughes, senior vice president of technology at NEMA, pointed to Buffalo, New York, whose peak electricity demand sits at roughly 500 MW. Emerging AI data centers are increasingly approaching or exceeding 1 GW of demand. That’s a single facility drawing more power than a mid-sized American city.
“Data centers are fundamentally different than any other building type,” Hughes told ENR. “You have to think of them more as cities than a building.”
Meta’s hyperscale campus in Mesa, Arizona, is cited as one visible example of this broader wave of AI-related infrastructure investment. Five buildings on a single campus, driving demand for power, transmission, and skilled labor simultaneously.
Analysis
The launch of a formal framework by NEMA, ASHRAE, and PNNL is a signal that the industry recognizes the current moment as a structural shift, not a cyclical boom. When major standards bodies feel compelled to write new rules from scratch, it usually means the old ones no longer fit. That’s exactly what’s happening here.
For the construction and field services world, the implications cascade outward quickly. A 1 GW data center isn’t just a large electrical job. It requires a coordinated buildout of power generation, transmission infrastructure, cooling systems, backup generation, energy storage, and the civil and structural work to hold it all together. Each of those scopes represents a subcontract opportunity, but each also represents a scheduling dependency that can break a project if one piece falls behind.
The inclusion of microgrids, demand flexibility systems, and liquid cooling in the framework points to where the specialty trade demand is concentrating. These aren’t standard HVAC and electrical scopes. They require crews with specific certifications and experience, and that supply is not currently abundant. The 300% growth projection in data center electricity consumption over the next decade means competition for those crews isn’t going away.
The geographic dimension matters too. Hyperscale campuses are landing in places like Arizona’s East Valley, Texas, Virginia, and the Pacific Northwest, often outside of major metro labor markets. That forces general contractors and owners to either import skilled labor, which raises costs and logistics complexity, or build local pipelines fast, which takes time that aggressive project schedules don’t always allow.
The grid interconnection challenge compounds the workforce issue. A facility approaching 1 GW can’t just plug into existing transmission infrastructure. Utilities need lead time, and that lead time is often measured in years. Projects that break ground before interconnection is secured create scope gaps and schedule risk that flow directly downstream to subcontractors.
The NEMA/ASHRAE/PNNL framework is an attempt to get everyone working from the same playbook earlier in the process, so that power, cooling, commissioning, and grid-interactive design are planned together rather than sequentially. For subcontractors, that means earlier engagement in design coordination will increasingly be a requirement, not a nice-to-have.
What It Means for Subcontractors
- Crew specialization is now a competitive differentiator. Data center scopes increasingly require expertise in liquid cooling, microgrid integration, and energy storage. Subcontractors who can demonstrate certified experience in these areas will have a shorter path to preferred vendor status on hyperscale projects.
- Plan for labor mobilization well outside your home market. Large campuses are being built in areas that don’t have deep local labor pools. Subcontractors need to build their remote mobilization capacity now, including housing, per diem structures, and travel crew management.
- Get into design-phase conversations earlier. The new NEMA/ASHRAE/PNNL framework is explicitly designed to integrate planning across power, cooling, and commissioning from the start. Subcontractors who wait for issued-for-construction drawings on these jobs will miss the window where scope and schedule are still being shaped.
- Understand the grid interconnection timeline before pricing. A project that is delayed because utility interconnection isn’t ready will compress your execution window without reducing your scope. Build schedule contingency into bids on new-site data center work.
- The 300% demand growth projection means this market has duration. This isn’t a one-cycle opportunity. Subcontractors who invest in data center-specific capabilities now are positioning for a decade-plus of demand, not just the current wave.
