Why Data Center Development Is Changing

Data center development is transforming the landscape of infrastructure. A developer, speaking to the Southern California News Group, felt compelled to clarify something: "I'm not a data center developer." This small moment reveals something significant about where the industry stands right now. Data center development has become so dominant, so capital-intensive, and so structurally different from conventional real estate and infrastructure work that developers are actively defining themselves against it.

That distinction matters. The companies and investors who *are* stepping into data center development are discovering that the rules of the game are unlike anything else in infrastructure.

The Demand Is Real — and It's Relentless

Hyperscalers like Microsoft, Amazon, and Google have collectively committed hundreds of billions of dollars to data center expansion over the next several years. Microsoft alone announced $80 billion in data center investment for fiscal year 2025. That capital has to land somewhere — and it's creating a frenzied scramble for suitable sites, power access, and entitlement pathways that existing infrastructure development pipelines simply weren't built to handle.

The bottleneck isn't money. It's everything else.

Land with adequate power proximity, water access for cooling, and favorable zoning is genuinely scarce. A 100 MW hyperscale campus requires roughly the same electrical infrastructure as a small city. Securing that kind of power interconnection — particularly at the utility transmission level — can take three to five years in many markets, sometimes longer. Developers who entered this space expecting it to move like industrial real estate learned quickly that it doesn't.

What's driving the demand underneath all of this? Artificial intelligence, primarily. Training a single large language model can consume as much electricity as hundreds of homes use in a year. Inference workloads — actually running these models at scale — may ultimately be more power-hungry than training. The compute requirements aren't leveling off; they're accelerating.

The Challenges That Actually Trip Developers Up

Regulatory friction is the issue most developers mention first, but it's worth being specific about where the friction actually lives.

Utility interconnection queues are clogged across every major U.S. market. PJM's interconnection queue, which covers much of the Mid-Atlantic and Midwest, had over 2,600 projects waiting as of recent reports. MISO and CAISO aren't much better. A data center project that needs 50 to 200 MW of dedicated capacity can spend years navigating this queue — during which time site control costs money, and market conditions shift.

Permitting at the local level is its own obstacle course. Communities that welcomed the promise of tax revenue are increasingly pushing back on projects that strain water supplies, generate significant heat output, and — critically — create relatively few permanent jobs relative to their footprint. A 500,000-square-foot data center campus might employ 50 full-time workers once operational. That math doesn't always satisfy local elected officials who need to answer to constituents about infrastructure impacts.

Environmental considerations are becoming harder to sidestep. Traditional air-cooled data centers consume massive amounts of water for evaporative cooling. A single large facility can use millions of gallons per day — a serious concern in drought-prone regions like the Southwest. Liquid cooling and immersion cooling technologies are advancing fast, but retrofitting existing designs or specifying new ones adds cost and complexity to projects already under financial pressure.

The Financial Architecture Is Unusual

Data center development doesn't look like industrial or commercial real estate from a capital stack perspective, and treating it that way is a mistake.

Development costs for hyperscale facilities routinely run $10 to $15 million per megawatt — and that figure is climbing as construction costs, electrical equipment lead times, and land prices all move upward simultaneously. Transformer procurement, once a routine line item, has become a critical path issue. Lead times for large power transformers have stretched to 18 to 24 months in some cases, and the supply chain disruption shows no sign of rapid resolution.

The investors who are winning in this space understand that data center development is fundamentally an infrastructure play, not a real estate play.

That reframes the underwriting. Long-term lease commitments from creditworthy hyperscalers — 10 to 15-year terms with renewal options — function more like utility offtake agreements than commercial leases. The cap rate compression that's characterized data center REITs reflects this: investors are essentially pricing these assets like regulated infrastructure, not like office parks.

For developers without hyperscaler anchor commitments, the calculus gets harder. Speculative construction at 100+ MW scale is difficult to justify given land, power, and construction costs. The viable path for most non-hyperscale developers is colocation at smaller scale or purpose-built edge facilities serving specific enterprise or regional needs.

Where Technology Is Reshaping the Development Equation

AI integration is changing data center design from the inside out. The power density requirements for GPU clusters — the hardware doing heavy AI compute — are dramatically higher than traditional server infrastructure. Older facilities designed for 5 to 8 kilowatts per rack are inadequate for modern AI workloads that may demand 30 to 100+ kilowatts per rack.

This creates a bifurcation in the market. Legacy facilities that can't be upgraded to handle high-density compute are losing relevance. New development has to be designed with future power density increases in mind, which means over-provisioning electrical and cooling infrastructure upfront — adding cost before a single server is installed.

Energy efficiency is no longer just a sustainability talking point. It's a financial necessity. Power Usage Effectiveness (PUE) — the ratio of total facility power to IT equipment power — is a metric that hyperscalers track obsessively. Every fraction of improvement in PUE translates directly to operational cost savings at scale. Facilities running PUEs above 1.5 are increasingly uncompetitive. The best operators are pushing toward 1.2 and below.

Renewable energy procurement has become standard practice for hyperscalers, both for ESG commitments and for long-term cost hedging. This is reshaping site selection: proximity to renewable generation resources — wind in the Plains states, solar in the Southwest and Southeast — is now a real development criterion, not just a marketing consideration.

Building for What Comes Next

The developers who will define this sector over the next decade aren't simply building warehouses for servers. They're functioning as infrastructure developers in the fullest sense: navigating power markets, water rights, transmission policy, and long-duration energy storage in ways that no previous generation of real estate developers had to.

Site selection increasingly involves analysis that looks more like utility siting than commercial real estate brokerage. Proximity to high-voltage transmission, fiber backbone routes, and access to large renewable energy zones are the variables that drive competitive advantage — not proximity to labor markets or retail amenities.

The rise of behind-the-meter power generation — on-site natural gas, fuel cells, and increasingly, solar-plus-storage — is giving developers more optionality in markets where grid interconnection is constrained. It's not a complete substitute for utility power, but it allows projects to move faster and hedge against grid reliability risks.

What the industry needs — and is slowly developing — is a new class of developer with fluency across power, land, construction, and technology domains simultaneously. The quote from the beginning of this piece makes more sense now. Traditional developers aren't equipped for this. Saying "I'm not a data center developer" isn't modesty. It's an accurate reading of how much specialized expertise this work actually demands.

The infrastructure professionals who build that expertise — or partner strategically with those who already have it — are the ones who will capture the enormous capital flows chasing this asset class for the next decade and beyond.

Explore the InfraSale Marketplace for more insights and opportunities.

[INTERNAL LINK: data center trends]

[INTERNAL LINK: infrastructure challenges]

[INTERNAL LINK: AI in data centers]