Industry press reported in mid-2025 that Applied Digital Corporation (Nasdaq: APLD) crossed 1 gigawatt of contracted capacity across its U.S. AI campus portfolio, anchored by a reported $7.5 billion, 15-year lease at its Polaris Forge 3 campus. Applied Digital's current SEC filings on EDGAR are the authoritative reference for confirmed figures. The broader pattern, however, is not in question: multiple hyperscaler-grade AI campuses are under construction or in contract across the Dallas-Fort Worth, San Antonio, and Abilene corridors simultaneously.
One gigawatt of contracted capacity is roughly equivalent to the output of one large nuclear power plant. The U.S. Energy Information Administration places the average U.S. nuclear unit at approximately 1,000 megawatts. Applied Digital is one company. Google, Microsoft, Meta, and several other hyperscalers are building concurrently in Texas. The aggregate load is substantially larger than any single announcement.
This is not a technology industry story. It is an operations story. Power, land, and skilled labor are shared resources — and mid-market manufacturers are on the losing side of the capital asymmetry.
Pressure Point One: Your Power Contract May Already Be Stale
ERCOT operates the grid serving roughly 90 percent of Texas. Unlike every other major U.S. grid, it has minimal interconnection with neighboring systems — PJM, SPP, and WECC — which means large in-state load additions cannot be absorbed by drawing power from outside the state. When a gigawatt of new demand comes online in Texas, Texas has to generate it or conserve it.
ERCOT's own load history shows record peak demand in consecutive recent summers, with grid operators issuing conservation appeals during heat events. The grid was already operating near its margin before the current AI buildout began adding load.
The mechanism of harm for industrial manufacturers is specific. ERCOT uses locational marginal pricing (LMP): the price of power at any given point on the grid reflects both generation cost and congestion on the transmission lines serving that node. When large data center loads come online near a transmission node, congestion rises and LMP prices increase for all industrial users on that node. A manufacturer in Denton or Garland may be paying for congestion caused by a 300-megawatt AI campus two substations away.
Practical risk exposure by contract type:
- - Fixed-rate contracts signed before 2022–2023 predate the current AI infrastructure demand wave and expire into a materially different market.
- - Variable-rate or indexed contracts already expose operators to real-time LMP spikes during peak events.
- - Demand charge structures can increase substantially when grid stress events push utilities to revise tariff assumptions.
Concrete action: Pull the settlement point pricing history for your ERCOT node — your utility or retail electric provider can supply the node designation — and compare the 12-month average to your contracted rate. A gap there needs to be understood before your contract comes up for renewal.
Pressure Point Two: The Industrial Site You're Eyeing May Go to a Data Center
Data center developers have site requirements that overlap directly with industrial manufacturers: large contiguous acreage, proximity to high-voltage transmission infrastructure, fiber access, flat topography, and sites already zoned or permitted for heavy electrical load. Those characteristics describe the same power-ready industrial parks that mid-market manufacturers target for expansion or relocation.
CBRE and JLL have both tracked tightening industrial vacancy across major Texas metros since 2020. Competition from data center developers for power-ready parcels has intensified that pressure. A site that might have supported a 200,000-square-foot manufacturing facility three years ago is now worth more — in absolute dollar terms — to a data center developer who can commit to a 20-year lease on 100 acres at transmission-level power.
The DFW industrial corridor from Alliance in north Fort Worth through Denton, Frisco, and McKinney has seen particular pressure. Several large data center projects in that zone absorbed sites previously marketed to industrial users. San Antonio's southeast industrial corridors, near existing utility infrastructure, face the same dynamic.
For a manufacturer evaluating expansion or relocation: site availability and power availability are now the same decision. A site without confirmed substation capacity and a clear ERCOT interconnection queue position is a liability — regardless of what the listing says about zoning.
Pressure Point Three: You're Hiring Electricians in the Same Market as a $7.5 Billion Project
Construction of a single large AI campus at the scale reportedly planned for Polaris Forge 3 requires hundreds of licensed electricians, ironworkers, concrete formers, and low-voltage controls technicians over a two-to-four-year build cycle. Multiple campuses building simultaneously in the same metro create a sustained wage premium across those trades.
The Associated General Contractors of America has documented construction trade shortages nationally, with Texas among the states reporting the most acute skilled trades gaps. Data center construction — combining high-voltage electrical work, precision cooling systems, and controlled fiber infrastructure — draws from the same licensed electrician and controls technician pool that manufacturers rely on for maintenance, capital projects, and automation upgrades.
A Texas manufacturer trying to hire a licensed electrician in 2025 DFW is competing against project labor agreements and construction wages from hyperscaler-funded builds. The same applies to any manufacturer planning a plant expansion, automation retrofit, or new substation installation over the next 24 months.
What to watch:
- - If your facility expansion or capital project requires electrical contracting, get bids now — before construction pipelines for the next wave of AI campus projects fill the available labor pool.
- - Check whether your current maintenance contractors are being pulled toward larger projects; turnover in that segment has accelerated.
- - Apprenticeship and in-house technician development programs carry a longer payback horizon but reduce ongoing exposure to external labor market pressure.
What Applied Digital's Pattern Actually Confirms
Applied Digital's publicly disclosed business model is clear from its SEC filings: the company builds large-scale AI infrastructure campuses, signs long-term leases with investment-grade hyperscaler tenants, and monetizes power delivery as its core product. A 15-year lease structure is not a cyclical bet. It is a structural commitment. Infrastructure built to that duration does not come offline when AI spending cycles. The grid load, the land use, and the labor demand are long-duration.
That is the material fact for Texas manufacturers. Capital commitments are reportedly already signed. Campuses are under construction or in planning. ERCOT is managing a tighter grid than it managed in 2020. The window for easy power contract renegotiation, straightforward site selection, and competitive trades labor is narrowing.
A manufacturer who hasn't reviewed power contract terms, assessed transmission node exposure, or benchmarked local electrician and controls technician wages against current market rates in the past 12 months is working from stale assumptions. The hyperscalers moving into Texas have 15-year visibility on what they need. Mid-market manufacturers should have at least a 24-month view on the same inputs.