The AI Boom’s Hidden Cost: A Tale of Two Trends

A recent working paper from the Electric Power Research Institute has challenged a widely held notion about data centers and their impact on electricity costs. Contrary to popular opinion, data center operations actually caused retail electricity costs to decrease until at least 2024.

The research found that for every doubling of data center capacity, average retail electricity prices decreased by 3.5%. This phenomenon is rooted in the way electricity markets function. Unlike other commodities, electricity prices are based on cost recovery – how much is consumed – rather than production costs. As fixed costs are divided among more consumers and kilowatt hours, the greater division of these costs leads to lower prices.

The effect is amplified by load increases from increased data center usage, which causes generators to come online, many of which are becoming more energy efficient. This creates a virtuous cycle that benefits both consumers and the environment.

However, this trend may be short-lived. The rapid buildout of data centers – expected to reach $7 trillion by 2030 – is already correlated with increased power costs. PJM, the largest power grid operator in the country, projected a $6.3 billion increase in consumer electricity costs over the next three years, largely attributed to increased data center power demands.

The key determinant of future electricity prices will be whether the AI buildout lives up to its hype. If it doesn’t, we could see a reversal of this trend, leading to higher electricity costs for consumers. Asa Watten, coauthor of the study and EPRI researcher, notes that if customers for data centers aren’t there, fixed costs will be spread among fewer people.

The scenario raises questions about the AI bubble and when – or if – it will pop. Some investors are already growing more skeptical of the technology’s promise, with Mark Cuban warning that many data centers may become redundant as AI adoption increases but power efficiency improves.

However, energy efficiency is expected to continue improving, with electrification reducing household energy costs through increased use of electric vehicles and heat pumps. This could lead to positive spillovers for neighbors, keeping prices stable or even decreasing them.

As the AI buildout continues apace, it’s essential to consider both the short-term gains from economies of scale and the long-term implications of a potentially overbuilt infrastructure. The debate around the AI bubble is far from over, but one thing is clear: we need to be mindful of the hidden costs associated with this trend and ensure that they don’t come at the expense of consumers’ wallets.

The data center boom has been hailed as a driver of innovation, but its true impact must be carefully examined. As Watten noted, “This clearly efficiency-increasing thing or total budget-reducing thing could have positive spillovers to your neighbors.” The question is: will we be prepared for the consequences if this trend reverses?