Microsoft’s Cloud Chief: Energy Problem Is People, Not Power
Microsoft’s cloud chief says AI’s power demands are manageable, but public trust and people pose a bigger challenge.
Microsoft's Cloud Chief Noelle Walsh recently spoke about the challenges facing the company's cloud operations amidst the rapid expansion of artificial intelligence. Despite the fact that U.S. electricity demand has remained relatively stable over recent decades, Walsh predicts that the data center industry's share of the national grid could reach 12% by 2030.
While acknowledging that power consumption is not her primary concern, Walsh emphasizes that the main issue is people - specifically the availability of skilled workers needed to build and maintain the necessary infrastructure.
Walsh highlights the impressive efficiency improvements in Microsoft's data centers, noting that each new generation of chips is significantly more efficient than its predecessor. However, she admits that efficiency is not keeping pace with demand, and the cooling of AI chips often relies on closed-loop systems that consume a substantial amount of electricity.
In response to this challenge, Microsoft proactively seeks out regions with abundant power supplies to locate its new data centers. For example, in Sweden, the company moved to a region with ample power resources and connected it to the national grid through network fiber.
To address the labor shortage, Microsoft is taking a proactive approach by investing in training programs and partnering with local educational institutions. The company operates data center academies in collaboration with nearby colleges, aiming to train and hire local residents for the numerous positions required to operate large-scale facilities. By focusing on community development and sustainability, Microsoft seeks to balance its rapid expansion with the needs of the people living in proximity to its data centers.
Written by urgent.news from Newsweek's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.