AI Could Consume 25% Of US Power By 2030: 'The More Information They Gather... The More Power It Takes,' Warns Arm CEO

Without improved energy efficiency, artificial intelligence (AI) data centers could potentially reach a quarter of the U.S. power requirements by the end of the decade.

The stark warning came from Arm Holdings plc ARM CEO Rene Haas on Tuesday as the chip-design giant and its parent company, SoftBank Group, pledge $25 million to help fund a $110 million initiative aimed at supporting AI research at universities across the U.S. and Japan.

The Rising Cost of AI’s Electricity Demand

As reported by the Wall Street Journal, Arm is now setting its sights on reducing the power consumption of artificial intelligence technologies.

According to Haas, AI models, including OpenAI’s ChatGPT, require substantial electricity, which could lead AI data centers to consume 20% to 25% of U.S. power needs by 2030, a significant increase from today’s 4% or less.

These models require an “insatiable” amount of electricity to operate effectively, said Haas. “The more information they gather, the smarter they are, but the more information they gather to get smarter, the more power it takes,” he said, according to WSJ.

The International Energy Agency underscored the issue in a recent report, noting that a single interaction with ChatGPT uses nearly 10 times the electricity of an average Google search.

Seeking Sustainable Solutions

With the alarming increase in power requirements, Haas emphasized the challenge of advancing AI research without more efficient energy solutions.

The U.S.-Japan research collaboration, receiving funds from Arm, will involve Carnegie Mellon University and Keio University, focusing on innovative ways to curb electricity use.

This AI initiative is part of a larger strategy by U.S. ambassador to Japan, Rahm Emanuel, who previously orchestrated a $150 million U.S.-Japan quantum computing research program.

Alongside Arm, major tech players such as Amazon.com Inc. AMZN and Nvidia Corp. NVDA are also investing in the initiative, with commitments to support additional research efforts in Washington and Tsukuba, Japan.

Emanuel pointed out the strategic importance of U.S.-Japan cooperation in AI and quantum computing as a counterbalance to China’s aggressive tech policies, which he argues stifle entrepreneurship.

Read now: US Stock Futures Flatline On Inflation Data Worries; Bond Yields Slip While Gold And Oil Surge: Analyst Says ‘Momentum Is Still There’ For Bull Run

Image generated using artificial intelligence via Midjourney.