Is AI really driving up electricity bills? Here’s what the experts say

Lindsey Martin’s electric bill hit $314 in July, the highest bill she had received this year so far, she said on TikTok. In the video’s more than 4,000 comments, many users reported similar electricity bill spikes.

Martin’s bill rose again in August to $372, a notable jump from $150 two or three years ago.

“I thought it was so high that I couldn’t believe my bill was $150,” Martin, a registered nurse based in Kentucky, told CNN by phone. “And now I wish my bill was $150.”

Martin is not alone. Residential electricity costs are on the rise, according to data of the United States Energy Information Administration (EIA). The average price of electricity in the United States has increased 13% since 2022, with the cost of retail electricity expected to grow faster than the rate of inflation, the report said. Some regions, such as the Pacific, Middle Atlantic and New England, could see even higher increases than the national average.

These increases are largely driven by the costs of updating and maintaining the power grid and other necessary infrastructure, especially in the face of increasingly frequent extreme weather events, according to energy and IT experts who spoke to CNN.

But a new technology wave is also driving up electricity bills: The AI ​​boom is boosting electricity demand and energy resources as tech giants invest billions in what many consider the biggest IT shift in decades. OpenAI and Broadcom announced a partnership just this week to design and develop 10 gigawatts of custom AI chips and systems, more than enough to power a major city.

This trend is only expected to continue. Data centers are expected to consume about 6.7% to 12% of U.S. electricity in 2028, up from 4.4% in 2023, according to a December 2024 report from the Ministry of Energy. A Analysis by Bloomberg News found that areas near data centers experienced an increase in electricity costs of up to 267% compared to five years ago.

The energy sector simply isn’t equipped to keep up, said Bob Johnson, an analyst at market research firm Gartner who tracks the semiconductor industry.

“Explosion of demand” for AI

Investments in data centers have grown rapidly as tech giants stake the future of their businesses on AI. Meta said it spent $17 billion on capital expenditures, which generally refers to money spent on data centers and infrastructure, for the quarter ended in June, while Microsoft said it spent $24.2 billion. Data center construction spending hit a maximum of 40 billion dollars in June, according to a report from the Bank of America Institute.

Existing data centers will likely need to be upgraded to ensure they have sufficient capacity to handle new power-hungry AI services and products.

That growth is contributing to higher demand for electricity than the United States has seen in two decades, requiring more investment in electricity generation and transmission, said Rich Powell, CEO of the Clean Energy Buyers Association, a trade group that represents electricity buyers.

Other factors driving the increase in demand include the shift to electric heating systems in homes and at new manufacturing plants, according to the Bank of America report.

There hasn’t been enough data center construction and upgrades in the past to affect consumer prices, said Ram Rajagopal, a senior fellow at Stanford University’s Precourt Institute for Energy. But today, the AI ​​industry is experiencing an “explosion in demand.”

AI tools are also becoming more sophisticated, evolving beyond text queries to handle complex tasks such as generating realistic video clips in seconds and coding websites from scratch. This means they also require more resources than ever.

“AI requires a lot of calculations,” said Shaolei Ren, associate professor of electrical and computer engineering at the University of California, Riverside.

Rates vary by location, but retail electricity prices generally include the “cost of production, transportation and delivery” of electricity and can sometimes take into account investments in infrastructure upgrades, according to the EIA.

Large electricity buyers generally pay lower rates because the distribution infrastructure is less complex; electricity needs to go to one location rather than to hundreds or thousands of homes, Johnson said. Pricing models have not been updated to account for strong data center growth, he said, although that can vary by location.

Oregon, for example, adopted a bill requiring data centers to “pay for the actual strain they put on Oregon’s power grid” so the costs aren’t passed on to the consumer.

“In other words, owners should not have to pay for data centers, but that is not built into the pricing structure,” he said.