Explainer

Power and the grid

A data center is a large, steady electrical load. Somebody has to build the capacity to serve it, and somebody has to pay for that.

What it is, in plain words

Utilities size the grid for peak demand, the worst hour of the worst day. A large data center adds load that runs flat out around the clock, which is different from a factory that runs a shift. Serving it can mean new generation, new transmission lines, or both. Those cost billions, and the bill gets spread across whoever the regulator says it gets spread across.

For scale: a 100 megawatt facility draws roughly the continuous average consumption of about 80,000 homes.^[1]

What the evidence actually shows

Kentucky has a state-government source for this, which is unusual and useful. EPIC, the Kentucky Energy Planning and Inventory Commission, was created by the General Assembly in 2024. It published a report for the legislature in June 2026.^[1] The numbers below are from it.

• Kentucky already hosts 37 data centers across nine markets, 24 of them in Jefferson County.^[1]
• In a March 2026 filing with the Public Service Commission, LG&E and KU reported 29 potential data center projects in the pipeline. The 11 the utility rated at 50 percent probability or better add up to about 3.5 gigawatts.^[1]
• PPL, the utilities’ parent company, told investors in May 2026 that total prospective data center demand in that territory could reach roughly 12 gigawatts.^[1]
• East Kentucky Power Cooperative has 11 active data center projects seeking over 10 gigawatts.^[1]
• For comparison, Kentucky utilities generated a maximum of 18.4 gigawatts in summer 2024.^[1]
• LG&E and KU’s 2025 filing projected a 29 percent increase in seasonal peak demand by 2032, almost entirely from large economic development projects.^[1]

The response so far is new generation. In October 2025 the Public Service Commission approved two new units, Mill Creek 6 and Brown 12, at a cost of $2.798 billion. The approval was conditioned on the utility’s commitment not to proceed if the data center load doesn’t materialize.^[1] That condition matters more than the price tag. Without it, ratepayers could end up paying for capacity built for a customer that never arrives.

EPIC also tracks prices in the PJM capacity market. They went from $28.92 per megawatt-day in the 2024/25 auction to $269.92, then $329.17, then $333.44 for 2027/28.^[1] That’s close to a twelvefold increase across three auctions. The 2027/28 auction was also the first where cleared capacity fell short of the reliability requirement market-wide.^[1]

What’s contested or unknown

How many of these projects actually get built is the open question, and the regulator has said so on the record. EPIC notes the Public Service Commission found it troubling that the utility had not provided evidence supporting the probability estimates it assigned to individual projects.^[1] So the pipeline figures are a range built on developer intentions, not a forecast. A project can request interconnection in several territories at once while committing to none.

Other states have answered the who-pays question with rate design rather than moratoriums. Virginia’s State Corporation Commission created a GS-5 rate class in November 2025. It requires data centers over 25 megawatts to sign 14-year contracts. Minimum billing demand is 85 percent for transmission and distribution, 60 percent for generation.^[1] AEP Ohio’s July 2025 large-load tariff set an 85 percent minimum billing floor and 12-year contracts with exit fees. It followed a two-year moratorium triggered by a queue over 30,000 megawatts. It produced roughly a 50 percent drop in speculative interconnection requests.^[1] Whether Kentucky does anything similar is open.

Questions worth asking at a hearing

• Is there a signed interconnection or service agreement, or only a request in the queue?
• What rate class does this customer fall into, and is there a minimum-take contract? For how many years?
• Who pays for the transmission upgrades this project needs, and over what period?
• Is the utility asking regulators to approve new generation on the strength of this load? Is that approval conditioned on the load actually arriving?
• What happens to the cost recovery if the project shrinks, delays, or leaves?
• What probability did the utility assign this project, and what evidence supports that number?
• Is any of this in an open Public Service Commission case the public can file comments in?