Global race for AI infrastructure amid soaring energy costs

Japan plans to invest ¥1 trillion ($6.34 billion) over five years in a domestic artificial intelligence company. This initiative aims to build infrastructure for AI, despite rising electricity costs raising concerns about the industry’s sustainability.The government will collaborate with SoftBank Group and Preferred Networks to develop the largest foundation model in Japan, employing around 100 engineers to compete globally. This effort reflects worries about the risks of relying on foreign AI technology.

China is also increasing its focus on technological independence. Domestic chipmakers are preparing for public offerings to enhance their capabilities, following successful launches by Moore Threads and MetaX in Shanghai. Companies such as Biren Technology and Baidu’s Kunlunxin are also planning to go public.

SoftBank is pursuing a $22.5 billion funding commitment to OpenAI by year-end, by selling assets and securing loans. CEO Masayoshi Son’s significant investment signifies a strong commitment to AI infrastructure.

AI Infrastructure

The rapid expansion of data centers is putting pressure on energy resources. Projections suggest data centers will consume 945 terawatt-hours by 2030, nearly tripling from 415 TWh in 2024. In the U.S., energy use by data centers could reach 9% to 12% of total supply by 2028.

Concerns have arisen over whether tech companies’ activities are raising residential electricity costs, with investigations launched by three Democratic senators targeting major companies like Alphabet, Microsoft, and Amazon. Utilities face significant expenses to upgrade grids, which may be passed on to consumers.

The energy demand is benefiting companies that supply power infrastructure. Jefferies recently upgraded GE Vernova to Buy, citing expected growth in gas pricing and long-term service demand amid rising energy needs.