Why This Site Fails as a Data Center Location

Every few months, a developer looks at a large parcel of land, runs the numbers on construction costs, and decides a data center belongs there. Sometimes they're right. Often, they're not — and the gap between those two outcomes is where hundreds of millions of dollars get quietly destroyed.

The case examined here is instructive precisely because it isn't unique. Talbott and Harris laid out a clear-eyed assessment of why a specific property fails as a viable data center location, citing a lack of power stations as a primary disqualifier. That single deficiency tells you almost everything you need to know about how these decisions should be made — and how frequently they aren't.

Site Selection Is the Whole Game

Data center developers spend enormous energy optimizing for server density, cooling efficiency, and PUE ratios. All of that matters. But none of it matters if you pick the wrong site.

The most technically sophisticated data center on earth is worthless if it can't be reliably powered. Site selection isn't a preliminary checkbox — it's the foundational decision that every subsequent dollar either validates or wastes.

The criteria that separate viable sites from expensive mistakes are well established in the industry, even if they're not always applied with the rigor they deserve. Power availability sits at the top of that list, followed closely by network infrastructure, physical accessibility, geological stability, and regulatory environment. Miss on any one of these in a significant way, and you're not just facing a cost overhaul — you're potentially looking at a project that simply cannot be built to spec.

Power Isn't Just a Utility — It's a Hard Constraint

Here's the number that reorients how you think about data center economics: a hyperscale facility can consume anywhere from 100 MW to over 500 MW of power. For context, 100 MW is enough electricity to power roughly 80,000 average American homes. That isn't something you solve with a call to the local utility.

Power station availability isn't a preference — it's a binary gate that determines whether a project is viable at all.

When Talbott and Harris identified the absence of nearby power stations as a core deficiency of this site, they weren't flagging a line item that could be renegotiated or phased in over time. They were describing a structural problem. Bringing adequate power to an underserved site requires transmission infrastructure — new substations, upgraded grid connections, and potentially new high-voltage lines running miles to the nearest adequate source. The cost of that buildout can easily run into the tens of millions of dollars before a single server rack is installed, and the timeline can stretch three to five years, depending on permitting and utility cooperation.

That's assuming the utility is willing to prioritize the project at all, which isn't guaranteed in regions where grid capacity is already strained. ERCOT, PJM, MISO — grid operators across the country are managing interconnection queues that stretch years deep. A site without existing power infrastructure isn't starting from zero; it's starting from behind.

Network Connectivity: The Other Half of the Equation

Power gets the headline, but network infrastructure is equally non-negotiable. Data centers live and die by latency and redundancy. A facility that can't connect to multiple fiber routes — preferably diverse, physically separated paths — is a liability for any enterprise client with serious uptime requirements.

Sites that lack proximity to existing fiber corridors face a similar infrastructure gap as power-deficient sites. Dark fiber can be leased or laid, but that comes with lead times and costs that erode the economic case for choosing a remote or underdeveloped location in the first place.

What "Myriad of Reasons" Actually Means Financially

Talbott and Harris described the site as unsuitable for "a myriad of reasons." That phrasing often gets glossed over, but it deserves unpacking in financial terms because the compounding effect of multiple deficiencies is rarely linear.

A site with one significant infrastructure gap is expensive. A site with several is often unfixable at any reasonable cost basis.

Consider the math: if closing the power gap costs $40 million in infrastructure upgrades, and addressing network connectivity adds another $15 million, and site accessibility improvements (road grading, utility easements, permitting) contribute another $10 million — you're now $65 million deeper into the project before construction begins. For a mid-sized colocation facility targeting $80–100 million in total development cost, that's a project that no longer pencils.

The financial risk isn't just capital expenditure — it's opportunity cost. Every month a development team spends trying to make a bad site work is time they aren't spending on a site that could actually deliver returns. For operators in a market where hyperscalers and colocation giants are signing 20-year leases and locking up quality sites at speed, timing is everything.

There's also the downstream risk that often goes underestimated: even if you build on a compromised site, you'll spend more to operate it. Higher power procurement costs, limited redundancy options, and reduced ability to attract anchor tenants all compress margins over the life of the facility.

What Serious Site Selection Actually Looks Like

The due diligence process for data center site selection that sophisticated developers use looks nothing like a standard real estate acquisition. It involves utility load studies, fiber path mapping, geotechnical surveys, flood plain analysis, and sometimes multi-year conversations with grid operators before a site ever goes under contract.

The checklist includes:

• Megawatt availability within 18–24 months — not theoretical future capacity, but committed, deliverable power
• Multiple utility feeds from independent substations, ideally from different parts of the grid
• Proximity to carrier-neutral fiber or the ability to connect to diverse long-haul networks
• Sufficient land for expansion, since a 10 MW build today frequently becomes a 100 MW campus within a decade
• Local and state regulatory environment, including permitting timelines, tax incentives, and zoning flexibility
• Water availability for cooling, increasingly critical as liquid cooling deployments scale

When a site fails on the first item in that list — power station availability — it rarely redeems itself on the others. Infrastructure deficiencies tend to cluster geographically. The same regions that lack transmission capacity often lack fiber density and face longer permitting timelines.

The Contrarian Reality Developers Should Sit With

Here's the observation that doesn't get made enough: the sites that look cheapest on paper are almost always the ones that fail these tests. Land in areas underserved by power infrastructure is inexpensive for a reason. The market has already priced in the friction.

Developers who convince themselves they can arbitrage that discount by building the infrastructure themselves are betting on execution timelines they rarely control and cost estimates that almost always expand. The utility doesn't answer to your project schedule. The permitting agency doesn't care about your IRR.

The developers who consistently build and operate successful data centers aren't the ones who found cheap land — they're the ones who paid appropriate prices for sites that were genuinely ready.

That distinction is worth internalizing at the board level, not just the site selection team. Because by the time a bad site has consumed 18 months of development activity and significant pre-development capital, the organizational pressure to continue — to not write off the sunk cost — becomes its own form of risk.

What Comes Next

If you're evaluating a site that has been flagged with power or infrastructure concerns, the honest question isn't whether those problems can be solved. Most problems can be solved with enough money. The honest question is whether solving them produces a project that still makes economic sense compared to the alternatives available to you right now.

The data center market in the United States is not short of demand. Northern Virginia, Dallas, Phoenix, Chicago, and a growing list of secondary markets are absorbing capacity faster than it can be built. Quality sites — sites with real power, real fiber, and real access — are what constrain the market, not capital or demand.

That scarcity makes disciplined site selection more valuable, not less. Walk away from the site that Talbott and Harris described. The next deal is out there, and it won't require you to build a power station first.

Explore InfraSale Marketplace for quality data center locations.

INTERNAL LINK SUGGESTIONS:

• [INTERNAL LINK: site selection criteria]
• [INTERNAL LINK: data center economics]
• [INTERNAL LINK: infrastructure challenges]