Federal regulators just moved to clear the biggest bottleneck standing between AI companies and the power they need. On Thursday, the Federal Energy Regulatory Commission ordered six major grid operators to fast-track interconnection requests from data centers and other large electricity users, according to TechCrunch AI. The commissioners approved the orders unanimously, and the data centers themselves will foot the bill for connecting.
This is significant because grid access has quietly become the hardest constraint in the AI buildout. You can raise the capital, secure the chips, and design the campus, then wait years to actually plug in. FERC’s orders are aimed squarely at that wait.
What the orders actually require
The directives put grid operators on a tight clock and ask them to prove they can connect large loads “in a timely and orderly manner.” Per TechCrunch AI, operators now have to:
- Submit a report within 30 days detailing how much spare generating capacity they have, if any
- Defend or revise their regional electricity rates within 60 days
- Be more accommodating to behind-the-meter power, meaning on-site generation that bypasses the public grid
- Consider “alternative transmission technologies”
That last point is worth watching. FERC didn’t name names, but the opening could favor grid-tech startups working on solid-state transformers or superconducting transmission lines. For a sector that rarely gets a regulatory tailwind, that’s a real opening.
Why this matters now
The backdrop here is a grid under serious strain. Operators spent two decades planning around near-zero demand growth. Now electricity demand from data centers is expected to nearly triple through 2035, and the system wasn’t built for it.
The pressure is already showing. PJM, the country’s largest grid operator, has descended into what TechCrunch AI describes as something resembling chaos, with major utilities threatening to walk away. Wholesale electricity rates have climbed as much as 267% compared with five years ago, according to Bloomberg figures cited in the report. Developers that couldn’t get a timely connection have turned to behind-the-meter power out of desperation, even though it’s typically pricier and more complicated.
There’s a competitiveness angle too. Energy Secretary Chris Wright prodded FERC to act, warning back in October that connection delays threatened to undermine U.S. standing in AI. The orders are the response to that pressure.
The catch FERC didn’t fix
Here’s what stands out: a fast lane only helps if there’s something to connect to. FERC sped up the line, but it didn’t address the shortage of actual generating capacity.
The numbers tell the story. At the end of 2023, grid connection requests for new power plants exceeded the total capacity of the existing power plant fleet. The queue to get on the grid was longer than the grid could theoretically serve. Power plants are stuck in the same slow process as the data centers waiting on them, so moving data centers up the line doesn’t generate a single new megawatt.
Supply policy is also pulling in a different direction. TechCrunch AI reports the Trump administration agreed on Wednesday to pay $765 million to wind developer Invenergy to cancel offshore wind leases near California, Maine, and New York. One of those projects would have generated up to 2.4 gigawatts, enough at peak to power roughly 1.8 million homes. Invenergy says it’ll redirect the money toward natural gas plants in the Midwest and geothermal in the West. All told, the administration has now spent about $2.6 billion to scuttle offshore wind.
What to expect next
The 30- and 60-day clocks mean we’ll learn a lot fast. Those capacity reports will show, region by region, just how much headroom the grid really has, and the rate filings will signal who ends up paying as demand climbs. Public sentiment toward AI and data centers has soured as power prices rise, so expect the cost question to stay loud.
For anyone building or financing AI infrastructure, the message is clear: the regulatory door to the grid just opened wider, but the supply problem behind it is still very much unsolved. More detail is available in the original TechCrunch AI report.