On-site generation has emerged as the data center industry’s primary answer to a growing infrastructure problem. Interconnection queues now run three to seven years in many markets. Grid capacity is most constrained in exactly the regions where demand is surging fastest. And operators constructing facilities measured in hundreds of megawatts cannot afford to wait for traditional utility timelines to catch up.
So, they’re turning to an unusual strategy: generating their own power. More than 60% of data centers are actively exploring on-site generation. By 2030, over 35 GW of data center load is projected to come from self-generation. That number that would have been almost unthinkable just five years ago.
On-site generation addresses real constraints. However, the downside is significant. What inevitably results is a patchwork, fragmented and inefficient grid that impacts data center operators, utilities and customers alike.
There is a better way forward. It requires utilities and large customers to stop viewing each other as obstacles and start working together as partners in financing and developing the infrastructure the grid genuinely needs.
What On-Site Generation Gets Right
The data center industry is responding to a real and structural constraints. Time-to-power has become a significant bottleneck. While a data center can be designed and constructed in 18 to 24 months, securing power through conventional utility interconnection takes an average of three to seven years.
On-site generation directly addresses that gap. Rather than waiting years for grid interconnection, facilities can be powered from day one through dedicated generation assets designed and operated to serve customer load. For an industry where a single minute of downtime can represent millions of dollars in losses, reliable, dedicated power is a core operational requirement.
The move toward behind-the-meter generation reflects a fundamental mismatch between the pace of data center development and the pace of grid infrastructure. On-site generation will not resolve that mismatch. It relocates cost, risk and operational complexity away from systems purpose-built to manage them efficiently, and onto corporate balance sheets that were not designed to carry them. Compressing time-to-power is a meaningful achievement. Treating it as a complete solution is where the analysis deserves greater scrutiny.
Understanding the Full Picture of On-Site Generation
On-site generation offers real advantages but realizing them requires a clear-eyed view of the full capital and operational commitment involved. Operators building generation assets take on responsibilities that extend well beyond the initial equipment purchase.
Those commitments include:
Redundancy and reliability infrastructure: Unlike grid-connected facilities, which can draw on a utility’s broader system during maintenance or outages, on-site operators must ensure backup capacity is built into the portfolio itself. Every planned maintenance window requires a parallel asset on standby which is a structural cost that grid-connected operators share across a much larger system.
Power quality systems: Flywheels, power electronics and sophisticated control systems are delivered as shared services by the grid. Off-grid operators own, staff and maintain these individually, adding both capital and operational complexity.
Fuel supply and logistics: Continuously operating assets require long-term fuel contracts and supply chain management, with limited flexibility to adjust when market conditions shift.
Specialized operational capabilities: Managing on-site generation requires engineering expertise that most data center operators are building for the first time whether developed internally or acquired through the market.
Financing costs: Investor-owned utilities exist to raise capital cheaply for the kind of large-scale, long-duration power infrastructure data centers require. Technology companies are not historically infrastructure companies. The borrowing costs that come with funding generation assets on a corporate balance sheet reflect that.
From an operational standpoint, achieving grid-equivalent reliability off-grid requires meaningful overbuilds in capacity. Without neighboring assets to draw from, operating reserves must be physically owned rather than called from the broader system. The result is a generation portfolio that demands more capital and carries less flexibility than a grid-connected alternative.
What a Collaborative Model Looks Like
The path forward is about designing a structure where utilities and large customers collaborate to develop generation infrastructure that serves both parties without either bearing the full cost and risk alone. Fixed mindsets and siloed approaches will never serve a world in flux. The opportunity lies in asking better questions and building outcomes that benefit the broader system.
The core mechanism is straightforward. When a large customer requests service, the utility evaluates whether that load can be served more cost-effectively or more quickly, through a co-located distributed energy resource. The utility finances and constructs that asset using its lower cost of capital and infrastructure expertise. The asset enters the rate base. The customer pays for the asset’s costs over its useful life, rather than carrying them on their balance sheet upfront.
The customer may maintain a grid connection for redundancy and backup. The interconnection request is, however, sized according to the requirements of the customer and the judgement of the utility. The utility earns its regulated return. And the generation asset, rather than sitting on a corporate balance sheet, becomes part of the shared system, available to support grid stability when not fully utilized by the customer, and visible to planners managing the broader network.
This model succeeds because it plays to each party’s structural strengths. Until recently, technology companies were among the most asset-light businesses in the economy, owning servers and software rather than power plants. Moving into generation infrastructure represents a significant shift in how these companies operate. Utilities are built for exactly this: financing large capital assets at regulated cost, constructing and operating power systems and managing long-duration infrastructure risk. Data center operators are built to run computing infrastructure at scale. The collaborative model lets each do what they do best.
When the lifecycle cost of utility-financed generation is modeled against operator-financed generation for equivalent assets, the net present value advantage of the utility path is material. The difference in cost of capital alone, sustained over a multi-decade asset life, is large enough to change the decision entirely.
The Opportunities Ahead
The data center industry’s energy needs are significant and the opportunity is equally large. On-site generation can meet the need at an individual level. The stronger case for collaboration is one of efficiency and scale. A model that brings utilities and large customers together on financing, siting and system planning achieves outcomes that benefit both parties and the broader system in ways that individual projects, however well-executed, rarely capture on their own.
That model already has a clear shape. When utilities and large customers collaborate on the financing and development of distributed energy resources, they see faster time-to-power, lower lifecycle generation costs, infrastructure that remains in the shared system where it benefits all ratepayers and the planning visibility both parties need to invest with confidence over the long term.
The data center industry has already demonstrated its capacity to move quickly, invest boldly, and solve problems at scale. Bringing that same ambition to grid partnership, rather than grid separation, is how the industry transforms its most complex infrastructure decision into a lasting competitive and social advantage.
The decisions data center operators make over the next five years will shape the grid that everyone else depends on for decades to come. Getting the model right at the outset is significantly easier than correcting it later. That is the real stakes of this conversation: what kind of energy system gets built, and who benefits from it.
Next Steps: TRC Can Help
TRC helps utilities and large energy users navigate the complexities of a rapidly evolving energy landscape. By combining deep power system expertise with strategic planning, regulatory insight and technology integration, we help clients address growing demand from data centers, industrial facilities and electrification initiatives.
From interconnection studies and engineering design to stakeholder engagement and rate impact analysis, TRC delivers integrated solutions that accelerate time-to-power, strengthen grid reliability and create long-term value for utilities, customers and the communities they serve.