As data center demand continues to accelerate, access to firm power is becoming one of the industry’s most significant development constraints.
On August 12, 2026, Montera Infrastructure’s Chief Technology Officer, Craig Pennington, took the stage at Conn3ctData Chicago to share his perspective on delivering power in a constrained market, followed by a panel discussion exploring how data center developers can deliver capacity when utility power timelines no longer align with customer demand. Craig was joined by Mike Biltimier of Bay City Electric Works, Drew Terwilliger of Geronimo Power, Santiago Suinaga of STACK Infrastructure, and Brian Wolf, Ph.D., of NuScale Power.
Opening the discussion with brief remarks, Craig addressed a challenge facing data center developers across the U.S. – demand for capacity is growing faster than the grid’s ability to provide firm power. U.S. data centers drew approximately 65 gigawatts of grid power last year, with forecasts projecting that figure could reach roughly 130 gigawatts by 2030. Meanwhile, much of the new capacity being added to the grid is solar, wind and battery storage, while additions of firm, dispatchable generation remain limited. For developers, the result is a shift in how power availability is evaluated. In constrained markets, the critical question is increasingly not how many megawatts a utility can provide, but when those megawatts will be available.
Craig discussed onsite generation as one strategy for addressing that gap, allowing projects to move forward before utility capacity becomes available. He also emphasized the conditions that determine whether the approach is viable, including natural gas availability, pipeline capacity, and emissions permitting. In non-attainment areas, those constraints can require developers to carefully manage generation within permitted operating limits or evaluate alternative technologies such as fuel cells and emerging linear generators.
The discussion also addressed the importance of planning for the full lifecycle of onsite generation assets. Where onsite power serves as a bridge to a future grid connection, developers need to determine from the outset what happens to that infrastructure once utility power arrives. Local utility structures, ownership requirements, and regulatory frameworks can all influence whether a bridge strategy is viable. AI workloads introduce additional technical consideration. Large GPU clusters can create rapid fluctuations in facility demand that generation equipment cannot follow directly. Craig outlined how battery energy storage, medium-voltage UPS systems, reactive-power equipment and workload management can provide a buffer between the compute load and the generation source, creating a smoother and more predictable electrical profile.
Craig also shared how Montera is applying these principles within their portfolio at one of the sites, using gas turbines as bridging power until utility power is available. The turbines are paired with ON.energy’s medium-voltage UPS system to smooth and isolate the GPU loads from the turbine supply.
Once the utility connection becomes available, the onsite generation asset is designed to take on a second role, providing backup power for the data centers and serving as a peaking resource for the grid rather than becoming stranded infrastructure.
The conversation at Conn3ctData underscored a central point from Craig’s opening remarks: much of the technology required to deliver data centers in power-constrained markets already exists. The challenge is bringing those technologies together in the right sequence, within the right jurisdiction, and with a clear plan for the full life of the asset.
The future of digital infrastructure is being built now. Learn more about Montera’s approach and connect with our team at montera.com.
