FERC Criticizes US Tech Giants for Weak Engagement on Grid Planning for Data Centers

The US federal energy regulator has publicly warned Big Tech: it is no longer possible to ask for ever more electricity for data centers while staying out of the conversation with the agency. For the market, this is a signal that AI infrastructure has finally become a question of energy and regulation.

What the complaint is about

The chair of the US Federal Energy Regulatory Commission, FERC, said that large technology companies do not engage actively enough with the agency, even though they need additional capacity for fast-growing data centers.

The complaint is not about formalities. It is a rebuke over the lack of systematic dialogue: the regulator expects companies to come not at the last minute, when the capacity is already needed, but early, with demand forecasts, an understanding of timelines and a readiness to discuss grid constraints.

In effect, FERC is saying the tech sector talks to the energy industry too late, when a project already has to be connected to the grid quickly. For the agency this is not a procedural trifle but a risk to the entire chain of decisions, from capacity planning to the date a facility comes online. The larger the data center, the costlier a mistake in assessing its impact on the system, and the less room for improvisation regulators and grid operators have.

"Large technology companies that need more power for data centers are

not engaging enough with the agency."

Strain on the grid

The problem is bigger than bureaucratic correspondence. New data centers consume huge amounts of power, and connecting large loads affects generation, transmission lines, the balance of supply and demand, and local grid reliability. The more projects launch in parallel, the higher the price of planning errors: if grid operators, regulators and customers work out of sync, commissioning dates slip, disputes over tariffs intensify, infrastructure decisions are made too late, and project costs rise.

For the US market this is especially sensitive, because a significant share of new energy demand is now concentrated around data centers. Even when a company has land, server hardware and money for construction, that is not enough. It needs clear connection rules, agreed technical parameters and, sometimes, new investment in grid infrastructure. So FERC's remark is addressed not only to tech corporations but to the whole ecosystem of contractors, utilities and regional operators.

What is expected of companies

The brief statement does not list FERC's specific requirements point by point. But from the logic of such projects, one can conclude that the regulator expects not public announcements of billions for AI, but early and substantive participation in energy planning. For Big Tech this means discussing needs before construction starts, building in realistic connection timelines, not treating the grid side as a secondary detail, and answering basic questions about load in advance.

In practice, such a dialogue usually covers several things:

• power consumption forecasts by site and launch phase
• discussion of connection timelines and bottlenecks in grid infrastructure
• coordination with local utilities, grid operators and regulators
• clarity on who funds the necessary grid upgrades, and when

If companies come to the government with an almost finished project and an expectation of fast connection, the room for maneuver narrows sharply. Any capacity shortfall then turns into a political and tariff problem. That is why the FERC chair's comment can be read as an invitation to earlier, more technical and less PR-driven work from the industry, before the conflict between construction speed and grid capability becomes systemic.

What this means

The era when a data center could be treated as an ordinary piece of real estate with servers inside is ending. The more aggressively Big Tech expands its computing capacity, the more closely it will have to work with energy regulators. This is where the speed of scaling, the cost of new sites and the limits of AI service growth will be decided.

For the market this is bad news in only one case: if companies keep thinking about the grid after they have already promised the world yet another hyperscale cluster.