This is a quarterly market report for PJM, the largest power grid operator in the US. It’s for January to March of this year and was published yesterday. It discusses many things, including the rising electricity costs for customers. It’s 937 pages, so it’ll take a lot more time to unpack it all, but let’s just take a look at the Introduction/Quarter in Review section.
“Data center load growth is the primary reason for recent and expected capacity market conditions, including total forecast load growth, the tight supply and demand balance, and high prices. But for data center growth, both actual and forecast, the capacity market would not have seen the same tight supply demand conditions, the same high prices observed in the 2025/2026 BRA, the 2026/2027 BRA, and the 2027/2028 BRA, and the currently expected tight supply conditions and high prices for subsequent capacity auctions. Forecast data center load growth has been the primary cause to date and the accuracy of those forecasts is highly questionable.
…
The impact on the 2026/2027 and 2027/2028 BRA revenues would have been higher had PJM not used the Agreement VRR curve [2]. If the 2026/2027 BRA had been run with PJM’s proposed unrestricted VRR curve, total revenues would have been $19,294,286,100, an increase of $3,169,915,210, or 19.7 percent, compared to the actual auction results. If the 2027/2028 BRA had been run with PJM’s proposed unrestricted VRR curve, total revenues would have been $26,324,850,846, an increase of $9,913,272,621, or 60.4 percent, compared to the actual auction results. In other words, the inclusion of data center load in the last two auctions increased customers’ bills by $13,768,851,483, even with the maximum price from the Agreement in place. The market distorting effect of data center load would have been $26,852,039,314 but for the Agreement. The result of the Agreement was to reduce that impact by $13,083,187,831. Contrary to assertions that the Agreement was a typical regulatory response to high prices, the maximum price resulting from the Agreement is more consistent with a competitive outcome than would have occurred without the Agreement. The Agreement was a logical and reasonable response to the unmitigated market distortions created by unsubstantiated data center load forecasts that should not have been included in the market clearing in the first place.”
The Agreement VRR curve is, from what I can tell, a curve that predicts the demand and resources required by electrical infrastructure in the future. In other words, your electricity costs are not just affected by current use, but predicted future use and demand.
In summary, this market report claims that PJM has unnecessarily inflated electricity costs due to taking into account unreliable, overly high predictions of future datacenter use. In other words, certain models may have been exaggerating the future energy usage of data centers, and this exaggeration has led to electrical companies preemptively raising electricity costs when they shouldn’t have.
Now let’s look at the Bloomberg reporting on this exact market report:
Their reporting very feeds into and reinforces the very exaggeration of AI costs that the original report said was the issue in the first place!
Sidenote: Obviously, I’m simplifying from excerpts, so I would encourage people to read the intro of the market report itself. There is a lot more to analyze that I have also not read. Based on the introduction alone, it’s very clear that the overexaggeration of AI energy needs does directly impact people’s costs of living and people need to be very careful about how they’re reporting.
