How Data Center Development is Shaping Infrastructure

Data centers rarely make the front page — until they do. And when they do, it's rarely about servers.

It's about water rights, property tax exemptions, utility rate structures, and — increasingly — who sits on the local utility board. The story of modern data center development is really a story about infrastructure power: who controls it, who pays for it, and who benefits from it. The recent entanglement of Salt River Project (SRP) board elections with data center developer interests isn't an anomaly; it's a window into how deeply this industry has embedded itself into the machinery of local governance and infrastructure planning.

Understanding that dynamic is essential for anyone buying, selling, or developing land and infrastructure assets in markets where hyperscale demand is reshaping the terrain.

Data Centers Are Infrastructure — Full Stop

Strip away the tech branding, and a data center is a heavy industrial facility. It consumes enormous amounts of power (a single hyperscale campus can draw 100–500 MW, enough to serve tens of thousands of homes), requires redundant fiber connectivity, generates significant heat, and demands reliable water access for cooling. That profile puts data centers squarely in the same infrastructure category as manufacturing plants, logistics hubs, and energy generation facilities.

The industry isn't building data centers anymore — it's building campuses, and the scale distinction matters enormously for how communities and utilities have to respond.

Global data center construction spending surpassed $49 billion in 2023 and is projected to climb steeply through the decade, driven by AI compute demand that is orders of magnitude more power-intensive than traditional cloud workloads. A single AI training cluster can require 30–50 MW of dedicated capacity — a figure that would have described a mid-sized data center campus just five years ago. The infrastructure implications of that shift are only beginning to register with utilities, municipalities, and grid operators.

Site Selection: Where Geography Meets Grid Politics

Ask any data center developer what drives site selection, and they'll give you a list: power availability, fiber connectivity, climate, seismic risk, proximity to population centers. All true. But the real filtering mechanism — the one that actually determines whether a project breaks ground — is the relationship between the developer and the local utility.

Power costs can vary by a factor of two or three between markets. Arizona, for example, offers relatively low utility rates, a dry climate that reduces mechanical cooling loads, and abundant land. That combination has made the Phoenix metro one of the fastest-growing data center markets in the country. But cheap power only stays cheap if the utility's rate structure accommodates large industrial customers — and that rate structure is, ultimately, a political decision.

The Hidden Variable: Interconnection

Beyond headline electricity rates, interconnection timelines are becoming a critical differentiator. In markets where grid capacity is constrained, a data center project can sit in an interconnection queue for three to five years before taking a single watt of load. Developers who can negotiate directly with utilities — or influence the boards that set interconnection policy — gain a measurable competitive advantage. This is precisely why data center development and utility governance have become so intertwined.

Site selection factors also include:

• Tax incentives: Many states and counties offer property tax abatements ranging from 10 to 20 years for qualifying data center investments. Virginia, home to the world's largest concentration of data center capacity, pioneered this model.
• Regulatory environment: Permitting timelines, zoning flexibility, and environmental review processes vary enormously and can add or subtract 12–18 months from a project schedule.
• Labor market depth: While hyperscale facilities aren't labor-intensive to operate, they require skilled trades during construction — electricians, ironworkers, HVAC technicians — and access to ongoing technical talent.

When Local Elections Become Infrastructure Policy

The SRP situation crystallizes something the broader market has been slow to acknowledge: local elections are now a material risk factor in data center development timelines and economics.

SRP is a member-owned utility, meaning its board is elected by agricultural landowners and residential customers within its service territory. Those elections have historically attracted little attention — turnout is often minuscule. But as data center developers recognized that SRP board composition directly influences rate structures, interconnection policy, and load growth planning, the dynamic shifted. Developer interests began engaging in elections that most residents didn't know existed.

This isn't unique to Arizona. Across the Sun Belt and mid-Atlantic — the two dominant U.S. data center corridors — utility commission appointments, municipal elections, and county board races have become proxy battles over data center policy. In Loudoun County, Virginia, the tension between residential communities and data center operators over noise, visual impact, and infrastructure strain has generated real electoral consequences. In Georgia, debates over utility load forecasting — which directly determines whether large industrial customers get favorable treatment — have influenced Public Service Commission races.

The pattern matters for infrastructure investors because policy risk cuts in both directions. A utility board sympathetic to large industrial loads can accelerate development and deliver favorable rates. A board that prioritizes residential ratepayers may impose cost-allocation structures that make a project's economics unworkable.

The Economic Case — And Its Limits

Data centers are legitimate economic engines. A 100 MW hyperscale facility represents $800 million to $1.2 billion in construction investment. It generates significant property tax revenue even after abatements expire. It employs hundreds of construction workers during the build phase and, once operational, supports a smaller but well-compensated permanent workforce — typically 50–150 direct employees for a facility of that scale.

The tax revenue argument is real, but communities should read the fine print on abatement agreements before treating data centers as fiscal silver bullets.

Virginia's experience illustrates the tension. The state's data center tax exemptions — which waive sales tax on equipment purchases — have cost the commonwealth an estimated $1.1 billion annually in foregone revenue. In exchange, localities receive property tax payments and a degree of economic activity. Whether that tradeoff makes sense depends entirely on local fiscal conditions, infrastructure capacity, and opportunity cost. A county that abates property taxes for a data center that strains its electrical grid and consumes limited industrial-zoned land may find the economics less attractive than the initial announcement suggested.

That said, the indirect economic effects are substantial. Data centers anchor fiber infrastructure development, support local electrical contractors and engineering firms, and signal to other technology companies that a market has the connectivity and power infrastructure to support digital operations. Markets like Northern Virginia and Phoenix have built genuine technology industry ecosystems on that foundation.

The Road Ahead: Sustainability, AI, and the Grid Reckoning

Two forces are converging to reshape data center development over the next decade, and neither is fully priced into current market assumptions.

The first is sustainability pressure. Major hyperscale operators — Google, Microsoft, Amazon, Meta — have made aggressive public commitments to carbon-free operations and water neutrality. Those commitments are reshaping site selection toward markets with renewable energy resources, driving direct investment in solar and storage co-located with data centers, and creating demand for long-term power purchase agreements that are restructuring utility revenue models. Developers who can offer sites with a credible path to 24/7 clean energy will command premium positioning.

The second is the AI compute surge. The power density requirements of AI inference and training workloads are forcing fundamental redesigns of data center architecture — from air cooling to liquid cooling, from 10–20 kW per rack to 50–100 kW per rack or higher. That density shift has real implications for site selection: not every market with available land and grid access can support the thermal management infrastructure that next-generation AI facilities require. Markets that can will attract disproportionate investment.

The developers and landowners who understand that AI infrastructure isn't just more data center demand — it's a qualitatively different infrastructure challenge — will be positioned to capture the most value.

For anyone active in the infrastructure asset market, the takeaway is direct: data center development has outgrown its status as a niche technology real estate play. It now sits at the intersection of energy policy, local governance, environmental regulation, and geopolitics. The sites that win will be the ones where all of those variables align — and the investors who map those variables accurately, rather than chasing headline announcements, will be the ones who build lasting positions in this market.

Ready to dive deeper into the evolving landscape of data centers and infrastructure? Explore our marketplace for insights and opportunities: [InfraSale Marketplace](https://infrasale.com/marketplace).