The Washington Post claims, “Data centers have little impact on residential power rates.”

In a situation where residents' backlash against data centers has emerged as an issue during the recent U.S. midterm elections, the American daily Washington Post (WP) has drawn attention by publishing an article introducing the conclusion of a preprint (a draft of a paper released in an unreviewed state before formal publication) that "the impact of data centers on consumer electricity rate increases is minimal."

It is unusual for a research conclusion published in a preprint that has not been verified by the academic community, such as peer review, to be reported in an article and introduced to a credible media outlet as if its credibility has been proven.

The owner of WP, which published this article, is Amazon founder Jeff Bezos.

Amazon generated 63% of its operating profit (as of the third quarter of this year) from Amazon Web Services (AWS), a cloud business based on data centers.

On the 6th (local time), WP published an article titled "Everyone blames data centers for high electricity bills. But these are the real reasons," with the subtitle "Data centers have had only a small impact on residential electricity bills."

At the same time, he introduced the contents of a preprint (draft of a paper before submission to an academic journal) co-authored by three people, including Vishrut Rana and Sergei Neteshine of the Wharton School of the University of Pennsylvania, and Christian Kopps of Harvard Business School.

This preprint was posted on a public site for preprints in the social sciences called ‘Social Science Research Network’ (SSRN). The writing date is September 30th, the first posting date is the 9th of last month, and the most recent revision date is the 12th of last month.

The title of the preprint is “The Cost of Intelligence: The Impact of AI Data Centers on U.S. Electricity Bills.”

As a result of analyzing the impact of data centers on U.S. electricity rates from 2020 to 2025, the authors of the paper argued that the impact of data centers accounts for only 4% of the increase in residential electricity rates borne by the general public.

The paper authors assessed this as “significant but small.”

“Overall, our findings suggest that existing discourse about the impact of data centers on electricity rates is exaggerated and that the average effect on residential electricity rates is small and concentrated among co-location developers and specific utilities,” the study authors wrote.

However, 15% of the increase in industrial electricity rates was analyzed to be due to data centers, which was larger.

WP introduced the contents of this preprint and reported co-author Professor Netesine's comment, "It was somewhat surprising even to me."

WP said, "Multiple studies by several researchers and institutions have not found a strong correlation between data centers and the overall trend of U.S. electricity rates in recent years. According to another analysis by researchers at the Massachusetts Institute of Technology (MIT) earlier this year, the average increase in electricity rates for residential customers due to data centers between 2010 and 2024 was only 2.1%."

'Another analysis by MIT researchers' cited by WP also did not undergo academic verification as a source, and was a 'working paper' (research report) posted on the website of MIT's Energy and Environmental Policy Research Institute (CEEPR) in June of this year.

In the recent U.S. midterm elections, Democratic candidates who pledged to regulate or review data center construction won a large number of victories.

In a midterm election exit poll commissioned by a consortium of major U.S. media outlets, including the Associated Press, 76% of those surveyed opposed the construction of a new data center in their area, and 56% said their stance on data centers was a "key or major" factor when voting.

Negative opinions about data centers were more evident among those who said they voted for Democratic candidates.

Of the 41 data center-related referendums held across the U.S. simultaneously with the midterm elections, 40, excluding one in Harwood, North Dakota, concluded that local voters would limit or ban new construction.