New Power Plant Build Requirements for Large Load Data Centers

The Department of Energy is turning the inevitable into policy: the era of "just plug into the grid" for hyperscale compute is ending if not already. Power is no longer a cheap, abundant commodity you assume exists; it's the constraint. Local utilities and RTOs are staring at overloaded transmission, delayed interconnections, and a generation pipeline that can't keep pace with load growth. The result is predictable - costs rise, upgrades get socialized, and residential and small commercial customers end up subsidizing the infrastructure required to serve massive new loads. That's the political pressure point, and it's why the conversation is shifting toward requiring large data centers to bring their own capacity and stop treating the public grid like a blank check.

Read the full Department of Energy Fact Sheet for more detail.

Nistar's thesis has been built for this environment because our projects are power plants first and data center sites second—always designed to be compatible with both. We are not gambling on a single narrative. We develop dispatchable generation as the core asset, then pair it with compute as the offtake. At the same time, we maintain bankability through grid interconnectivity as a baseline requirement: the ability to import supplemental power, export generation, and remain fully financeable as an energy infrastructure platform. That matters because the fastest way to create a stranded billion-dollar asset is to build something that only works under one market assumption. We don't do that. Our sites are engineered so the energy stands on its own, and the data center enhances it—commercially, operationally, and strategically.

The grid-first story that hyperscalers and operators like to tell—"behind-the-meter is inferior" or "we prefer the grid"—has always been more about convenience and optics than fundamentals. When interconnection timelines stretch, capacity prices spike, and regulators start asking why ratepayers are carrying the burden for private AI buildouts, the math changes. Behind-the-meter isn't a compromise; it becomes the cleanest solution to two competing realities: society can't absorb unlimited upgrade costs, and the AI economy can't tolerate power uncertainty. If regulation formalizes what the market is already signaling, BTM generation becomes not just acceptable, but standard issue.

And we're not building in fringe markets. One of our projects sits inside PJM—13 states plus D.C.—where the system is facing a multi-gigawatt shortfall and capacity stress that's already rewriting forward power economics. Two more projects are in MISO, the center strip of the country where industrial-scale power development is most rational right now and where the next wave of load and infrastructure investment is colliding head-on with grid limitations. These are exactly the markets where owning the power stack is the competitive advantage, not a nice-to-have.

The risk allocation is the point. We take the power risk—the hard part, the scarce part, the bottleneck part—by controlling the land, the fuel and interconnection strategy, and the generation pathway. The hyperscalers and operators take the compute risk, which is what they're built to manage. We structure it so we control the deal because we control the asset that matters: capacity. They sign long-term PPAs for the power and pay lease fees for the site, creating a durable, creditworthy revenue profile anchored by counterparties that can survive a downturn. If anyone wants to argue AI is a bubble, fine—our structure is designed so the energy asset remains bankable and monetizable regardless, and the compute is contracted with tenants that don't disappear in a contraction.

Bottom line: This policy direction validates our game plan built over the past 18 months. The market is moving toward a world where power is strategic, local grids are constrained, and self-generation is not an exotic edge case but a requirement. We're not chasing the trend; we've been positioned ahead of it, solving both sides of the equation at once: grid shortages and data center demand, with a platform built to be financeable, expandable, and resilient under any macro scenario.