Andrew Stokols
The training and deployment of artificial intelligence (AI) models is predicated on an enormous energy supply. This has ushered in a new rush for resource security—both for power generation and water use, as the data centers on which these technologies rely consume large amounts of water to keep them from overheating (Semianalysis, February 13). Currently, the United States leads the People’s Republic of China (PRC) in number of data centers by a wide margin (Statista, October 11, 2024). However, given the energy needs of cloud computing and AI, which continue to expand, there is a growing imperative for coordinating energy and cloud investment.
In the PRC, the “Eastern Data Western Computing” (东数西算) plan is a multiagency national initiative to do just that. While the United States lacks a national strategy to coordinate cloud computing and energy use, instead relying on disconnected regional initiatives and private-sector led investment, the PRC’s plan serves as an illustration of the potential benefits such coordination offers at national and local scales (NDRC, February 22, 2022; NCSTI, accessed February 26).
No comments:
Post a Comment