People who want to understand the implications of exploding electricity demand growth for the clean energy transition can be excused if they find the news confusing lately. Whereas the news was once mainly about what “is,” the competition for clicks and the spread of advocacy journalism has made much of the news about scary versions of what “might be” in an attempt to influence readers (read: voters). But of course, the future has yet to be written, so different futures are possible.
Demand growth and rates
Everyone knows that demand for electricity is growing rapidly (due in part to energy-hungry data centers) and that our electric grid is struggling to keep up. Consumer prices are increasing, because demand growth is outpacing supply growth. But will consumers will end up subsidizing investments made to serve data centers owned by cash-rich tech firms?
A recent quantitative study by the Lawrence Berkeley National Laboratory (LBNL) suggested that places experiencing the largest demand growth have the lowest electric rates. And utilities deny that they are subsidizing data center investments with higher rates to households. However, another study by Harvard Law researchers argues that investments made to serve new data center demand ought to be paid exclusively by those data centers, and that traditional bundling of capital costs into per-kwh rates means that ratepayers will be subsidizing data center rates. By contrast, some economists argue that the most efficient approach is to charge data centers less than other customers because unlike captive residential customers data centers have other power supply options. This is an approach known as Ramsey Pricing.[1]
But when it comes to the electric grid, all such generalizations should be viewed skeptically. Reasonable people can disagree about who should pay; and who will pay will depend upon location-specific facts on the ground, including state energy policy.
For example, the LBNL study may already be dated, and may simply be reflecting the fact that data centers want to locate where rates are low (though LBNL’s multivariate analysis supposedly controlled for some of that). On the other hand, LBNL’s conclusion may also reflect situations in which new data centers make use of unused capacity on the transmission or distribution system, thereby spreading existing fixed costs across more kwhs, reducing household rates in the process.
The Harvard study cites “secret contracts” and political pressure that may indeed shift costs caused by data centers to other ratepayers. In that sense the Harvard researchers identify a risk that public utility commissions ought to guard against. But cost causation is a nebulous thing. When a utility upgrades a transmission line or builds a power plant to serve new data center load, that doesn’t mean that all the benefits of that new investment flow to the data center. Depending upon the evolution of demand and supply on the system over the long life of these investments, they may yield significant reliability or other benefits to other system users as well.
And when economists argue that Ramsey Pricing is efficient, they mean that it maximizes the sum of producer and consumer surplus, which is not most voters’ definition of a just or fair price. Indeed, Ramsey Pricing can be thought of as a kind of price discrimination among customers, and politicians prohibited price discrimination when they created today’s public utility statutes.
It seems that state politicians are keenly aware of the risks identified by the Harvard study, and of the political risks of letting ratepayers foot the bill for new data center loads. Politicians have an electoral incentive to be vigilant on this issue; though contrary to what the empirical studies tell us, some assume that politicians can ignore popular opinion.[2] In any case, it seems that a lot of what we read today about these distributional questions is intended as warning — a kind of lobbying for the writer’s preferred policy.
What about the energy transition?
What does all this demand growth mean for the energy transition? Here again the news seems contradictory.
A recent email blast from the energy consultancy Wood Mackenzie announced that “shale gas is back!” and identified growth in demand for gas as an investment opportunity. Gas-fired power plant investment is ticking up after years of decline. Yet elsewhere we read that the clean energy transition refuses to be slowed by either these market forces or the hostility of the Trump Administration. The Economist reports that “the solar industry is buzzing,” and that inexpensive clean energy technologies from China are pushing the transition forward irrespective of U.S. policy. NPR reports surging growth in battery storage and renewable energy despite these headwinds.
When thinking about the relationship between data center electricity needs and the energy transition, keep in mind these three truths about the big tech companies that are building data centers:
- They want reliable (24/7) power.
- They have longstanding company goals to procure low-carbon emission energy
- They are swimming in cash
Each company will reconcile those goals differently, but together they explain why clean and dirty forms of energy are growing simultaneously. The market push for more natural gas-fired power, plus the Trump Administration’s orders requiring use of uneconomic coal-fired power plants, mean more carbon emissions in the near term.
But there is reason to suspect that big tech firms will push for cleaner electricity options even when they can, sometimes even if those options are more expensive, as long as those options offer reasonably firm supply. Several are apparently willing to pay above market rates for nuclear power. Some who are relying on natural gas in the near term pledge to switch to zero-emission firm options when they become available, such as geothermal energy or natural gas plus CCS.
So yes, the energy news is confusing. Unfortunately, in the attention economy stories that help readers understand an uncertain future don’t sell as well as stories that warn of a scary future. Bold conclusions get more clicks than nuance and circumspection. But bold conclusions are premature when it comes to the impacts of data center growth on electric rates and carbon emissions.
For now the rush to build data centers is pushing up demand (and in some places, prices) and giving a lifeline to all sorts of energy investments. And to the extent that that leads to more greenhouse gas emissions the harms associated with a warming planet will be exacerbated by those emissions. But beyond that, time will tell. – David Spence
——————
[1] Economist Lynne Kiesling made this argument in comments to the Federal Energy Regulatory Commission’s rulemaking proceeding on data centers last week.
[2] The Economist magazine warned that a popular “backlash is brewing” against data centers (article behind paywall). On the other hand, news articles have recounted cases of power suppliers favoring data centers over other customers (see e.g., here) or data centers complicating power planning by backing out of plans to locate (see e.g., here).
