──── ENERGY INFRASTRUCTURE · PLANNING & DEVELOPMENT
A Distributed Model for Data Center Growth
By David Swank, CEO, i3 Power & Energy
The data center industry is experiencing a structural transformation that most market commentators have not yet fully reckoned with. The familiar pattern of the past twenty years, hyperscale campuses aggregated in a handful of major markets drawing power from high-capacity transmission corridors, is giving way to something more dispersed, more varied, and in many respects more technically demanding. The distributed model is not a trend. It is a response to physics and economics that the concentrated model can no longer evade.
I spent years on both sides of this dynamic. As a technology CEO, I watched our infrastructure requirements outpace the available power in markets we had relied on for years. As a utility CEO, I watched load growth projections change faster than interconnection and transmission infrastructure could respond. The collision of those two experiences is precisely what the industry is navigating now, and understanding it requires looking at both sides of the equation simultaneously.
Why Concentration Is Becoming a Liability
Imagine if you will a developer who locked in a 500-megawatt site in Northern Virginia five years ago, certain that power would follow. Today that developer is staring at a multi-year interconnection queue, land carrying costs that never stop, and customers who needed those megawatts yesterday. That is not a hypothetical. It is playing out across the industry's largest markets right now.
The hyperscale campus model made sense in an era when power was abundant, land was cheap near major network hubs, and load growth was relatively predictable. Data center developers could negotiate favorable utility rates, aggregate enormous capacity on contiguous sites, and benefit from economies of scale in construction and operations. That model delivered extraordinary value for two decades. It is now running into hard constraints that those favorable conditions masked for far too long.
“The next gigawatt of data center capacity will not look like the last one. It will be spread across more sites, in more markets, consuming power from a more varied mix of sources.”
The Distributed Alternative
A distributed data center strategy disaggregates the hyperscale campus into a network of smaller, strategically located facilities. Rather than pursuing a single 500-megawatt site with a single utility interconnection, a distributed developer builds a portfolio of 20 to 100-megawatt facilities positioned to access available power, leverage existing transmission capacity, and serve the latency requirements of specific customer segments.
This is not simply a workaround for constrained markets. In several important respects, the distributed model is technically superior. Network latency requirements for emerging AI inference workloads, edge computing applications, and real-time industrial AI demand that compute capacity be positioned closer to end users than hyperscale geography typically allows. The physics of data creates inherent advantages for distributed placement in a world where milliseconds matter.
The Power Advantage of Distribution
From a power perspective, distributed facilities can access grid capacity that a large campus simply cannot aggregate. Rural and secondary markets often have available transmission headroom that was built for industrial load that has since declined. Renewable energy resources, solar, wind, emerging geothermal, tend to be located where land and transmission capacity are most available, not where the largest data center markets are concentrated. A distributed strategy can co-locate compute and generation in ways that a concentrated model cannot.
This creates a meaningful opportunity for what I would describe as purpose-built energy infrastructure. At i3, we develop data center projects alongside the power generation and storage assets that will serve them, designing the compute and energy components together rather than treating power as an afterthought to be negotiated after the data center is sited. This integrated approach compresses development timelines, reduces exposure to utility interconnection queues, and produces bankable projects with operating economics that concentrated, grid-dependent campuses struggle to match.
Organizational and Operational Complexity
The distributed model comes with real challenges. Operating a portfolio of facilities across multiple markets, utility jurisdictions, and regulatory environments is substantially more complex than operating a concentrated campus. The organizational capabilities required, multi-market permitting expertise, distributed operations management, portfolio-level power procurement and dispatch, are different from what hyperscale developers have historically needed to build.
This is where experience at the intersection of technology and energy infrastructure becomes genuinely valuable. Managing distributed energy assets across multiple utility jurisdictions is a capability that the energy industry has spent decades developing. The data center industry is learning it now. Firms that can bridge those domains, applying energy sector operating discipline to data infrastructure, are positioned to move faster and operate more effectively than pure-play technology developers navigating unfamiliar terrain.
The fundamental thesis is straightforward. The growth of AI, the electrification of the economy, and the evolution of the grid are all pointing toward a world where compute and power infrastructure need to be developed together, distributed intelligently, and operated with the same rigor we apply to generation assets. The firms that understand that thesis and can execute against it will define the next chapter of data center development.
