Photo from the BrockNews.
Arm, a renowned chip-design firm, initially gained prominence for its innovations aimed at reducing power usage in smartphones, thereby prolonging battery life. According to the company's leader, the imperative for energy efficiency in artificial intelligence tasks mirrors this same commitment.
Rene Haas, CEO of Arm, addressed the public before an upcoming announcement on Tuesday regarding a joint initiative between the U.S. and Japan. This $110 million program aims to support AI research in universities across both nations. Arm, headquartered in the U.K., along with its parent company SoftBank Group based in Tokyo, has pledged $25 million in funding for this initiative.
Haas highlighted in an interview that AI models like OpenAI's ChatGPT have an unquenchable need for electricity. He emphasized that as these models gather more information to enhance their intelligence, they consume increasing amounts of power.
Haas warned that if efficiency isn't significantly improved, AI data centers might consume around 20% to 25% of the United States' power needs by the end of the decade. Currently, this figure stands at approximately 4% or less, which he finds unsustainable.
In recent months, technology executives have increasingly focused on the power challenge, leading to a surge in the stock prices of companies involved in owning and running electric-power plants.
In a January report, the International Energy Agency revealed that the energy consumption of a ChatGPT request averages 2.9 watt-hours, roughly equivalent to powering a 60-watt lightbulb for almost three minutes. This is nearly ten times more than the average energy usage of a Google search. The agency also projected that the power demand of the AI industry is anticipated to increase by at least tenfold between 2023 and 2026.
Haas expressed concerns about the challenge of advancing necessary breakthroughs if the power demands of large data centers for research purposes continue to rise persistently.
He voiced optimism that the U.S.-Japan research collaboration would focus on addressing the power challenge. Arm's financial support will contribute to a joint effort between Carnegie Mellon University in Pittsburgh and Keio University in Japan.
The collaboration is overseen by the U.S. ambassador to Japan, Rahm Emanuel, who previously initiated a $150 million U.S.-Japan research initiative in quantum computing, with support from IBM and Google. In addition, Amazon.com and Nvidia will each contribute $25 million to fund AI research at the University of Washington and Japan's University of Tsukuba as part of the broader AI partnership.
Emanuel stressed the significance of collaboration between the two allies amid China's advancements in AI and quantum computing research. He emphasized the importance of the U.S. and Japan combining the resources of both industry and academia, contrasting this approach with Chinese leader Xi Jinping's efforts to regulate the tech giants within the country in recent years.
Emanuel pointed out that one drawback of Xi's actions is the suppression of entrepreneurship. He suggested that the U.S. and Japan's approach, which fosters entrepreneurship, is likely to achieve greater success.
Liu Pengyu, spokesperson for the Chinese Embassy in Washington, criticized the comments as irresponsible. He stated that the Chinese government values entrepreneurship and innovation and has implemented measures to support the private sector. Liu emphasized that technology collaboration between nations should promote stability, open trade, and not undermine the interests of any other party.