AI has an "insatiable" hungry for more power — and if AI firms don't figure out how to be more energy efficient, it'll slow down research, the CEO of a top chip design firm said.
Arm CEO Rene Haas told The Wall Street Journal that AI can't keep devouring so much energy at huge data centers if the industry wants to keep advancing.
"It's going to be difficult to accelerate the breakthroughs that we need if the power requirements for these large data centers for people to do research on keeps going up and up and up," Haas said.
If progress isn't made, "by the end of the decade, AI data centers could consume as much as 20% to 25% of US power requirements," Haas told the Journal. "Today, that's probably 4% or less."
Haas did not immediately respond to a request for comment from Business Insider.
It's a warning sign for the buzzy AI field that's courted major investment from Wall Street's top firms.
In March, The New Yorker reported that ChatGPT uses more than half a million kilowatt-hours of electricity daily. The average US household uses just 29 kilowatt-hours each day.
By 2027, the energy required by the sector could spike to match countries the size of Sweden, according to another recent study.
Meanwhile, data centers are sprouting up and changing the rural American landscape, as environmentalists warn of potential repercussions.
UK-based Arm, which SoftBank Group owns, is contributing to a $110 million program for AI research between the US and Japan at top universities in each country.
Amazon and Nvidia are also participating in the program, according to the Journal, which was spearheaded by Rahm Emanuel — the US ambassador to Japan — to help both countries better compete against China.
