#

Joe Biden signs executive order to speed AI data center construction

An aerial view of a large data center under construction next to a roadway.
A CloudHQ data center under construction on Waxpool Road in Ashburn, Virginia, US, on Wednesday, March 27, 2024. | Photo: Getty Images

President Biden issued an executive order today aimed at speeding the development of AI data centers in the US.

It directs the Department of Defense (DOD) and Department of Energy (DOE) to lease federal sites to private companies building gigawatt-scale AI data centers and clean power facilities. It also tells federal agencies to “prioritize” and speed up permitting of AI infrastructure. The measure could create “categorial exclusions” to speed environmental review under the National Environmental Protection Act.

Developing new AI tools is an increasingly energy-hungry endeavor. Nevertheless, the Biden administration seems to think it’s worth the risk of further derailing US climate goals and putting additional pressure on already stressed power grids.

“We will not let America be out-built when it comes to the technology that will define the future,” Joe Biden said in a statement today.

Prior to the announcement today — in response to reports that the White House was considering measures to fast track data center development — environmental and consumer advocacy groups as well as Democratic lawmakers had urged the White House to avoid exempting AI from typical permitting procedures and environmental standards.

“We urge you to reconsider any potential executive action that could lead to increased pollution and costs for consumers,” says a letter sent by senators Sheldon Whitehouse (D-RI), Elizabeth Warren (D-MA), Edward Markey (D-MA), Brian Schatz (D-HI), and Peter Welch (D-VT) to the Biden administration on December 17th. “We are the United States of America; there is no doubt that we can win the AI race while accelerating our decarbonization efforts,” it reads.

Electricity demand from data centers has tripled over the past decade, according to estimates published by the Lawrence Berkeley National Laboratory (LBNL) on December 20th. It’s likely to double or triple again by 2028, according to the report. Data centers used about 4.4 percent of US electricity in 2023, which could rise to as much as 12 percent by 2028.

That rise in demand is the result of the tremendous compute power needed to train AI models. Utilities are already extending the lives of polluting coal and gas infrastructure to try to meet skyrocketing electricity demand. Customers also face rising electricity bills as a result.

Developers building new AI data centers on federal land will be required “to pay all costs of building and operating AI infrastructure so that this development does not raise electricity prices for consumers,” according to the White House.

That includes constructing the data center itself, as well as power facilities and transmission lines. Companies will be responsible for sourcing electricity that data centers use from new sources of “clean” electricity. They’ll also have to assess the security implications of AI models developed at federal sites and purchase “an appropriate share” of American-made semiconductors.

“In the race to dominate AI, we can’t lose sight of the very real race to stop the pollution that’s warming our planet and harming our health,” Johanna Neumann, a senior director at the Environment America Research & Policy Center, said in a December 19th statement.

Neumann argued that the focus should be on making sure that new computing facilities are more efficient, and that they run on renewable electricity. “Without those guardrails, AI’s insatiable thirst for energy risks crashing America’s efforts to get off dirty and dangerous fossil fuels,” Neumann added.

The government already leases federal lands for energy production, including fossil fuel exploration and renewable energy projects. Under the executive order, by February 28th, the Secretary of Defense and Secretary of Energy are supposed to find at least three sites each to host new AI data centers on land that their departments manage.