Is the Data Center Boom Unsustainable?
Are we underestimating the challenges in the data center boom? Discover the critical issues impacting growth and sustainability.
The numbers are staggering. Global data center capacity is projected to more than double by 2030, driven by AI workloads, cloud migration, and a seemingly insatiable appetite for compute power. Hyperscalers are signing land deals in the hundreds of acres. Power purchase agreements are being executed at a pace the grid was never designed to handle. And somewhere between the groundbreaking ceremonies and the ribbon cuttings, a quiet crisis is forming — one that developers, utilities, and grid operators are only beginning to reckon with.
The boom is real. But so are the cracks.
A Build-Out Running Ahead of Its Infrastructure
Northern Virginia still commands roughly 70% of U.S. data center inventory, but that concentration is exactly what's pushing developers into secondary markets — Phoenix, Columbus, Hillsboro, Spartanburg. The logic is straightforward: land is cheaper, permitting can be faster, and communities are often hungry for tax revenue and jobs. What these markets don't always have is grid infrastructure ready to absorb 100 MW, 300 MW, or 500 MW campuses that didn't exist in any utility's long-range forecast three years ago.
The fundamental tension in data center development right now isn't capital or land — it's electrons and the pipes to move them.
When a hyperscaler announces a 1 GW campus, that figure tends to dominate the press release. What gets buried is the transmission infrastructure required to serve it, the substation upgrades, the new generation that needs to come online, and the queue position that determines whether any of it happens on time. In many regions, that queue is measured in years, not months.
Interconnection: The Bottleneck Nobody Talks About at the Press Conference
Interconnection — the process of connecting new power generation or large loads to the grid — has become the defining constraint of the clean energy and data center buildout simultaneously. FERC's interconnection queue reforms (Order 2023) acknowledged the problem openly: as of mid-2023, over 2,600 GW of generation capacity sat waiting in queues across the country, most of it renewables that data centers increasingly need to meet sustainability commitments.
The interconnection issues facing data center developers aren't just about waiting in line. They're about what happens when the studies come back. Interconnection studies — the technical analysis utilities and grid operators conduct to determine what upgrades are needed — frequently return with upgrade cost allocations that weren't in anybody's pro forma. A project that penciled out at one cost basis can look very different after a cluster study assigns it $40 million in network upgrades it didn't anticipate.
Developers who underestimate interconnection risk aren't just facing delays — they're facing fundamental project economics questions.
For data center operators accustomed to moving fast, this is a cultural collision as much as a technical one. Utilities operate on planning cycles measured in years. Interconnection queues are governed by rules that prioritize earlier applicants regardless of project readiness. And transmission buildout — the actual physical infrastructure that carries power from generation to load — often requires its own separate regulatory approval processes that run parallel to, but independent of, everything else.
Capacity: The Demand Signal Utilities Weren't Ready For
Grid operators and utilities traditionally build load forecasts from historical consumption trends. Data centers have broken that model. PJM, which manages the grid for 65 million people across 13 states, revised its 10-year load forecast upward by roughly 40% in 2024 — largely because of data center growth. That's not a rounding error. That's a planning crisis.
The challenge cuts both ways. Data centers represent firm, predictable, high-value load — exactly what utilities like. But serving a 500 MW campus requires infrastructure investments that take 5 to 10 years to permit, engineer, and build. When a hyperscaler wants to be operational in 24 months, the math doesn't work without either pre-existing excess capacity on the local system or creative solutions that bend conventional utility planning timelines.
Some regions are better positioned than others. Areas with recent industrial facility closures — retired coal plants, shuttered manufacturing sites — sometimes offer existing transmission infrastructure that can be repurposed. That's part of why certain Rust Belt markets have attracted serious data center interest despite lacking the brand recognition of Northern Virginia or Phoenix. The infrastructure is already there. The queue position is already established.
Data center capacity concerns extend beyond raw power availability. Cooling infrastructure, water rights, fiber density, and physical security requirements all constrain site selection in ways that don't show up in a simple megawatt analysis. A site that checks every power box can still fail on water availability for cooling systems — a growing concern as data centers get denser and as water stress becomes a real risk factor in markets like Phoenix and Las Vegas.
Utility Rate Cases: The Hidden Variable in Project Finance
When a utility needs to recover the cost of major infrastructure upgrades — new transmission lines, substation expansion, generation additions — it files a rate case with its state public utility commission. These proceedings can take 12 to 18 months, sometimes longer. During that period, the approved cost recovery structure remains uncertain, which ripples into how utilities price large industrial interconnection agreements and special contracts for major customers like data centers.
For a data center developer modeling a 20-year return, a utility rate case outcome can move the needle on operating economics more than almost any other variable.
Rate cases matter to the data center sector in at least two distinct ways. First, they determine how costs for grid upgrades are allocated — whether they're socialized across the ratepayer base or assigned directly to the new load requesting interconnection. That allocation decision can mean tens of millions of dollars in direct project costs. Second, approved rates affect long-term power pricing, which is a critical input for operators who've made energy cost assumptions in underwriting.
There's also a less obvious dynamic: rate cases create regulatory uncertainty that can slow utility decision-making. A utility in the middle of a contested rate proceeding may be less willing to commit to novel tariff structures or special contracts for large data center customers until it has clearer signals from its commission. Developers who need customized interconnection agreements or unique service arrangements often find the rate case calendar dictates their timeline more than any technical constraint.
Building Smarter, Not Just Faster
The developers who are winning in this environment aren't just the ones with the most capital. They're the ones who've built genuine expertise in grid planning, utility regulation, and transmission economics — or who've partnered with people who have.
A few strategies are separating sophisticated players from the rest:
Early utility engagement — before site control, before financing, before anything. Developers who show up to a utility's planning team after signing a purchase agreement are already behind. The ones getting infrastructure commitments are the ones who've been in conversation with utility resource planners 18 to 24 months before they need the power.
Queue position as a real asset. In markets where interconnection queues are measured in years, a project that already has a queue position — or a site with existing large-load service — carries value that doesn't show up on a traditional appraisal. Some of the most interesting acquisitions happening quietly in the data center infrastructure space are effectively queue-position plays.
Distributed and on-site generation. Fuel cells, on-site solar, battery storage, and even small modular reactors (still early-stage but actively being evaluated by hyperscalers) represent ways to reduce grid dependence for baseload power. These aren't complete solutions — a 100 MW campus can't run on solar and storage alone with today's technology — but they reduce peak grid demand and can accelerate project timelines by lowering the interconnection ask.
Flexibility commitments. Some developers are beginning to offer demand response flexibility to utilities as a way to accelerate interconnection. A data center that can curtail load during grid stress events is a fundamentally different grid citizen than one demanding firm, uninterruptible power at all times. That flexibility has real value to grid operators and can unlock better treatment in the interconnection process.
The data center growth challenges aren't going away. If anything, as AI model training and inference workloads continue to scale, the power demands will intensify before they moderate. The question isn't whether the grid can handle it in theory — given enough time and capital, it can. The question is whether the infrastructure can keep pace with the deployment timeline that capital markets and technology roadmaps are demanding.
Developers who treat grid infrastructure as someone else's problem will find it becomes their problem at the worst possible moment — after they've committed capital and signed customer agreements. The ones who internalize the interconnection queue, the rate case calendar, and the transmission planning horizon as core project variables will be better positioned to deliver what the market is asking for.
The boom isn't unsustainable. But the way some players are pursuing it is.
Call to Action
Ready to navigate the complexities of the data center boom? Explore our marketplace for innovative solutions: InfraSale Marketplace.