Every so often a new set of lobbyists rises up to tell us that the electric industry has to do something because the expected demand from X will overwhelm the utilities. Back in the early days of the personal computer, the electric industry sponsored a study that showed that PCs would just about eat electricity. Everyone got excited until somebody realized that the researcher had incorrectly calculated consumption per computer. Recently we had the crypto crisis. Now it’s data centers that will consume all our electricity. These estimates, even when correct, suffer a flaw. They assume that there is no offsetting reduction in demand on the part of consumers affected adversely by the new user. (More use of crypto reduces electricity consumption at ATMs, for instance?) We should note that the recent concern that data centers will put strains on the US electricity grid comes accompanied by a PR push to boost nuclear power and a warning from the copper industry that demand for the red metal will outstrip mining capacity.
Not that we disagree about the prospects for a revival of demand for electricity. Several years ago we argued that percentage demand growth for electricity would triple and that the industry was unprepared. So with the recent fuss (and doubling of the growth rate in the past three years), we decided to do a simplified analysis (no econometric model here). Let’s make 2024 the base year. Next, use some rough estimates for demand growth over the coming decade. Data centers will consume 9% of electricity according to news stories. Electric vehicles would take 30% of annual electricity output once the automobile fleet were 100% electric. Well it won’t reach 100% for a long time at the current rate of sales. So let’s say 20% penetration by 2034, making a 6% addition to demand. Let’s also assume that the industry’s normal growth rate (assuming normal economic growth) is about 1% per year (or 10% for the total period). Finally, hotter weather and conversions from natural gas to electric heat pumps will increase the air conditioning load by about 5%. All this adds up to a 30% increase in sales over 10 years. Doesn’t sound spectacular? Well, this century annual growth in sales has been 0-1% per year. Relatively speaking, this represents a sea change and more important, implies significant capital spending to meet this demand.
Next question. Where will the electricity come from, given that the US government (EIA) forecasts 1% growth, not the 3% we believe. Plus, the industry will have to replace some of those ancient coal-fired power plants. It will, no doubt, plead that it cannot run a reliable network unless it keeps base load gas-fired stations on line, and even needs to add meaningfully to the gas fleet. We think the industry will get its say in terms of failing to decarbonize for no other reason than that the renewable producers won’t be able to get enough plant on line to help satisfy demand. After all, they have to wait for the transmission line. In some places, that is like waiting for Godot. Then there’s the hope to refurbish mothballed nukes like the plant formerly known as Three Mile Island. We assume that all mothballed plants come back on line (probably wishful thinking). Overall, at least on paper there is plenty of power generating plant in the domestic pipeline. Not an excess but enough. Figure 2 shows our estimates of the generation mix for 2024 and 2034. If we are right, natural gas stays in the picture, and plays a bigger role, while it replaces coal as a utility boiler fuel.
There is something bizarre about the notion that the US electric industry, the preeminent vehicle we have to decarbonize our economy, now produces over half of its output by burning fossil fuels, and in ten years will still produce half of its output by burning fossil fuels. That’s progress?
By Leonard S. Hyman and William I. Tilles for Oilprice.com