Economics is full of counterintuitive insights. Rent control can make housing less affordable. Business failures can strengthen an economy. And, yes, higher demand can sometimes lower prices. That’s a possibility worth remembering in the ongoing debates over data centers and electricity bills.
It sure seems like we’re going to build a whole lot more data centers in this country—at least, that’s what Wall Street is expecting. JPMorgan just raised its forecast for data-center capacity growth to 138 gigawatts through 2030, up from 122 gigawatts in its forecast last November. The bank also said it expects $5.5 trillion in AI-related capex through 2030, up from its previous forecast of $5.1 trillion.
If nothing else, that expansion likely means a lot more controversy over how data centers are affecting electricity prices. Yet one shouldn’t automatically assume that massive electricity consumption by these warehouse-sized supercomputers means higher electricity prices for most folks. In the new pre-print paper “Have Data Centers Raised Your Electric Bill? Causal Evidence from the United States,” researchers Asa Watten, John Bistline, and Geoff Blanford explain that they find “patterns of economies of scale for transmission, distribution, and generation costs as well as within and across retail customer classes.”
Higher electricity demand can lead to lower prices because electricity systems have huge fixed costs. If new demand persists, those costs can then be spread across more kilowatt-hours. And if utilities build new capacity to serve that demand, the new, more efficient supply of energy may be cheaper to generate than older plants and grid assets.
Watten-Bistline-Blanford found the following:
We estimate that data centers caused average retail electricity rates to fall modestly in the United States from 2015 to 2024 using an instrumental variables approach. Despite prevailing sentiment, the finding is consistent with economic reasoning: existing large power system fixed costs, economies of scale in transmission and distribution, and declining unit costs for generation imply that durable demand growth lowers average prices. We find patterns of economies of scale for transmission, distribution, and generation costs as well as within and across retail customer classes.
No guarantees going forward, of course. Maybe expected demand doesn’t come through. Or maybe current supply constraints continue or even worsen:
Several supply constraints are currently impacting the U.S. electric sector: orders on new natural gas turbines are backlogged, which is bidding up their prices; persistent supply-chain issues limit the availability of transformers and other electrical equipment; high tariffs on photovoltaic arrays from China have increased the price of solar in the United States; and aggressive executive action has stalled federal approvals for wind and solar projects while they are litigated in courts. Each of these constraints make new generation capacity investments more expensive than they would be otherwise and possibly more expensive than incumbent supply in some instances.
Or perhaps the sheer size of the AI infrastructure buildout leaves “open the question of whether the system can continue to expand quickly enough to accommodate concurrent growth in data center demand and end-use electrification without triggering the supply constraints discussed above.” Maybe extrapolating from the recent past is a doomed endeavor given the tsunami of investment pouring into the sector. The paper notes that data centers used about 4.5 percent of U.S. electricity in 2024 and are projected to account for 9 percent to 17 percent by 2030.
Certainly a role for policy here to make sure high electricity prices don’t undermine the emerging AI revolution.