US, China can work together on AI energy demand, expert says

As artificial intelligence drives a surge in electricity demand from data centers, the United States and China face a shared challenge: meeting energy needs without undermining sustainability goals.

The different infrastructure and market dynamics could make it valuable for the two countries to study each other's approaches, a Stanford expert suggests.

Richard Dasher, director of the US-Asia Technology Management Center at Stanford University, says there are ways to work together even as the two countries compete in AI development.

"I certainly think that there are opportunities for cooperation," Dasher told China Daily. "In some ways, it's worth noting that even though it's a little bit like a tennis game where the two tennis players are playing against each other, they both get better, the better they play."

Some of China's latest AI models are already having an impact on the industry in the US, which Dasher described as a positive development.

From the US perspective, he said, understanding different energy infrastructure and market structures is also increasingly important as AI puts greater pressure on electricity systems.

"China has very different dynamics. It's very good for us to study each other," said Dasher, suggesting this could create openings for exchanges.

The International Energy Agency projects that global electricity consumption from data centers will double from 485 terawatt-hours (TWh) in 2025 to 950 TWh in 2030, with power use by AI-focused facilities set to triple over the same period. A terawatt-hour is a unit of energy equal to 1 trillion watt-hours.

In the US alone, data center electricity use is expected to roughly double to about 426 TWh by 2030, according to the American Public Power Association.

For Dasher, the part he calls "disturbing" is the amount of additional electricity expected to come from sources that are not sustainable, including coal, natural gas and oil.

"Less than half of the predicted additional capacity in the United States is expected to come from sustainable energy sources," he said.

The US has overtaken China as the leading builder of natural gas-fired power plants. It is currently constructing twice as much capacity as China and has nearly three times the amount of gas power capacity in development, according to a recent analysis by Global Energy Monitor.

During the first half of the year, the amount of gas-fired generating capacity under development in the US to directly supply data centers nearly doubled, the report said.

New proposals

US states also account for some of the world's fastest-growing portfolios of planned gas-fired generation. New proposals in Texas have increased faster than anywhere else in the world, according to the analysis.

Turning to gas, a fossil fuel, rather than renewables to run new data centers has raised concerns among climate researchers, because it would increase carbon emissions and could make it more difficult to achieve climate goals.

Public sentiment has been turning against data centers. A March Gallup poll found that 70 percent of US respondents oppose the construction of AI data centers in their local area, driven largely by environmental and quality-of-life concerns. Resistance to new projects has spread across the country.

"In some ways, the reaction against building data centers is because people are worried about sustainability, and so you can't ignore the people forever," Dasher said.

He said concerns about environmental sustainability remain strong despite changes in US national policy.

liazhu@chinadailyusa.com