Data centers built in the US between 2020 and 2025 explain about 4% of the rise in residential electricity prices over that period, according to a working paper by researchers at the University of Pennsylvania’s Wharton School and Harvard Business School (SSRN abstract).
The paper, “The Cost of Intelligence: Impacts of AI Data Centers on US Electricity Prices”, by Vishrut Rana, Christian Kaps and Serguei Netessine, is dated September 30, 2026. It estimates that the same data centers raised their region’s industrial electricity price by an average of 5.4 $/MWh, explaining 15% of the observed industrial increase, and residential prices by an average of 1.7 $/MWh.
Business Pill · Capex and Depreciation
An evergreen Business Engineer explainer on how fixed infrastructure costs are spread over time. It teaches the general idea only and does not discuss electricity prices or either paper.
The key insight: A Wharton and Harvard Business School working paper estimates that data centers built from 2020 to 2025 explain about 4% of the observed rise in US residential electricity prices and 15% of the industrial rise.
The authors linked quarterly data center capacity additions with local electricity price and utility sales data. They used what the abstract calls a “single-anchor synthetic-control design with cumulative, repeated treatment” to estimate the causal effect of data center growth on retail electricity prices.
The abstract describes the residential effect as “significant but small spillovers to residential consumers”. It says the 1.7 $/MWh average “accounts for 4% of the observed increase since 2020”.
Share of the 2020-2025 US residential retail electricity price rise that the Wharton and Harvard Business School working paper attributes to data centers built in that period: about 4%, or 1.7 $/MWh on average. The paper puts the industrial share at 15% (5.4 $/MWh). Not peer reviewed.
Which Data Centers Moved Prices
The paper finds that the effect depends on who builds. According to the abstract, “data center additions by Big Tech firms result in no discernible price increases, whereas co-location data centers do”.
It also finds that utilities with previously higher infrastructure utilization and a history of serving large industrial customers have lower residential spillovers. The authors conclude that “the prevailing discourse about data centers affecting electricity affordability is overstated”, and that the average residential effect is small and concentrated among co-location developers and specific utilities.
A Second Study, a Different Method
A separate paper by Asa Watten, John Bistline and Geoffrey Blanford, posted on arXiv on June 18, 2026, uses an instrumental variables approach over 2015 to 2024 (arXiv abstract). It estimates that data centers “caused average retail electricity rates to fall modestly in the United States” over that period, citing economies of scale in transmission and distribution.
Those authors add a caution: “We caution that future supply constraints could reverse the effect”. The two papers cover different years and use different methods, so their estimates are not directly comparable.
Why It Matters Now
The paper’s abstract says fears of rising prices “led to numerous bans on AI data center development in the US”. On November 3, limits or bans on data centers led in 40 of 41 local ballot measures on unofficial counts (our piece), and a Senate bill would make data centers pay for grid upgrades (our piece). Deutsche Bank has said data center limits could cap an AI bubble (our piece), and CoreWeave has written to Texas Governor-elect Gina Hinojosa about its data center projects in the state (our piece).
Our Numbers
By our arithmetic, the industrial effect of 5.4 $/MWh is about 3.2 times the residential effect of 1.7 $/MWh. A rise of 1.7 $/MWh equals 0.17 cents per kilowatt-hour.
Working back from the paper’s shares, also our arithmetic, the total observed rise implied is roughly 42 $/MWh for residential prices (1.7 divided by 4%) and 36 $/MWh for industrial prices (5.4 divided by 15%). The paper’s percentages are rounded, so these are approximate.
The Structural Read
As we read the abstract, the paper separates who builds: it finds no discernible price increase from Big Tech data center additions and a price effect from co-location data centers.
As we read it, the two papers differ in period and method: the working paper covers 2020 to 2025 with synthetic controls, while Watten, Bistline and Blanford cover 2015 to 2024 with instrumental variables and estimate that rates fell modestly.
By our arithmetic, the industrial effect of 5.4 $/MWh is about 3.2 times the residential effect of 1.7 $/MWh, which equals 0.17 cents per kilowatt-hour.
Rana, Kaps and Netessine, working paper abstract
“the prevailing discourse about data centers affecting electricity affordability is overstated”
Three Implications
FOR UTILITIES The paper finds that utilities with previously higher infrastructure utilization and a history of serving large industrial customers have lower residential spillovers.
FOR DEVELOPERS The paper finds that data center additions by Big Tech firms result in no discernible price increases, whereas co-location data centers do.
FOR THE DEBATE Watten, Bistline and Blanford caution that future supply constraints could reverse the effect they estimate for 2015 to 2024.
The analysis says: “But infrastructure reality is catching up with AI ambitions: a projected 19GW gap between new data center demand and available grid capacity threatens to deflate the AI bubble.”
As we read it, the working paper measures the price effect of capacity already built through 2025, while the framework concerns capacity still waiting for grid connections. The analysis does not discuss residential electricity prices.
What Is Not Established
This is a working paper and has not been peer reviewed. We read its abstract on SSRN; the full 42-page text was not available to us, so the regional and utility-level estimates are not reported here. The averages do not say what any single household paid. The estimates cover data centers built through 2025, not those under construction. This piece makes no forecast.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
A Wharton and Harvard Business School working paper estimates that data centers built from 2020 to 2025 added an average of 1.7 $/MWh to residential electricity prices, about 4% of the observed rise, and 5.4 $/MWh to industrial prices, about 15%. It finds no discernible increase from Big Tech builds and an effect from co-location data centers. A separate paper by Watten, Bistline and Blanford estimates that rates fell modestly over 2015 to 2024.
A note on sourcing. The figures and quotes are from the SSRN abstract of the Rana, Kaps and Netessine working paper (dated 30 September 2026) and the arXiv abstract of the Watten, Bistline and Blanford paper (18 June 2026). We did not read either full text. The ratios, the cents-per-kWh conversion and the implied totals are our arithmetic. This is not investment advice.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
