When the House passed the reconciliation bill last week, it sent a chill through the clean energy industry. It was largely regarded as much worse than expected, with barely a whiff of compromise or moderation.
This version of the legislation — which has now been sent to the Senate — largely guts the Inflation Reduction Act and its support for the clean energy industry. Support has instead shifted to relying more on nuclear and fossil fuels to meet anticipated load growth in the U.S.
For those interested in urgently powering hyperscale data centers, this is a mixed signal at best. Renewables and storage have represented nearly all the new capacity coming on the grid in recent years, and will undoubtedly dominate in 2025. Saddling them with worse economics and inconsistent federal support only makes project development more costly and limits the options for developers eager to add capacity and flexibility to the grid.
But at the AI-energy nexus, there was more to the bill than just support for fossil energy and expedited permitting. Federalist Society be damned, there was a 10-year moratorium on all state and local regulation of AI.
(This may run afoul of the Byrd Rule, which prohibits the addition of extraneous provisions that have no budgetary effect. But running afoul of anything doesn’t seem to concern this administration or the Republican-controlled Congress, so the move should still be taken seriously.)
Proponents of this moratorium argue that a patchwork of state and local regulation will stifle innovation when it’s needed most, and slow the technology race against China. They argue that AI is “interstate infrastructure,” so only Congress should have regulatory authority over it. But 45 states have already introduced legislation to regulate AI, and 31 states already have laws or rules on the books. If the bill passes as written, all these would immediately be unenforceable.
It’s 2025, so fine, having a nuanced regulatory argument is likely a waste of breath. But it is worth pointing out a few things about what this moratorium implies that should concern those in the energy industry.
First, there isn’t a federal alternative proposed to replace those state laws on the books; nothing is even in the works. There is only talk, and most of it is decidedly libertarian. This means that without state or local regulation, consumers or businesses would remain unprotected from AI harms, which could lead to the unintended consequence of softening demand for AI tools as trust erodes — because of course bad actors will exploit this unregulated technology and people, or systems, will be harmed.
And there is also real potential for direct impact on the power sector.
To begin with, the language used in the bill is quite broad, and lumps “automated decision systems” in with “artificial intelligence models” and “artificial intelligence systems.” Where do automated decision systems reside in the power sector? Demand response, virtual power plants, managed EV charging, smart meter home energy management, power market bidding, system load balancing, emergency response, and many more technologies either in operation today or under development.
Choosing to wait for Congress to develop regulations for all of these feels a lot like waiting for them to regulate Facebook or X. It’s extremely fraught.
Allowing unregulated AI to be added to systems that impact the health and operation of the power system feels like lazy lawmaking at best. At its worst, the law is a deeply irresponsible rejection of the well-established principle of states as laboratories of democracy, where ideas are tested “without risk to the rest of the country.”
Particularly when it comes to AI and energy, regulatory innovation and pace at the state and local level feels exceedingly worth preserving.
A version of this story was published in the AI-Energy Nexus newsletter on May 28. Subscribe to get pieces like this — plus expert analysis, original reporting, and curated resources — in your inbox every Wednesday.
