AI demand leads to power shortages, 3-4year implied
With the surge in AI demand, power and transmission infrastructure, rather than semiconductors, have become the bottlenecks. Semiconductor capacity can be expanded in 3-4years, but building power facilities and transmission networks takes much longer.
As demand for AI services surges, power and transmission infrastructure are emerging as new bottlenecks. Semiconductor manufacturing capacity may take 3-4years to expand. By contrast, power generation capacity and transmission infrastructure require more time because of permitting and construction.
Large data centers and AI accelerators require a stable power supply. If power supply is unstable, data center utilization and expansion plans are affected. Some companies are reportedly considering their own generation facilities or long-term power purchase agreements to secure electricity.
Expanding power capacity and transmission networks involves complex approval processes that vary by region. If this process takes longer, AI infrastructure expansion is increasingly likely to be constrained by the pace of securing power. The article explains that this is creating investment opportunities in power and data center infrastructure.
Ultimately, the pace of AI expansion is said to depend not only on chip supplies but also on how reliably power can be secured. It points out that companies with power and transmission infrastructure may gain an advantage. The article states that companies securing access to power are attracting attention from an investment perspective.
Sources: The Motley Fool
Bullstory synthesized this report from the linked sources. AI may have assisted the writing process. Check the original reports for factual confirmation.