AUTO-UPDATED

Growing AI energy demands, the search for subatomic particles and the hottest amoeba

Artificial intelligence data centers are projected to consume 3% of global electricity by 2030, raising significant concerns regarding environmental sustainability, water usage, and regional power grid strain.

Key Points

  • AI data centers could reach 945 terawatt-hours of annual electricity consumption by 2030, tripling current usage levels.
  • Six major tech companies—Microsoft, Google, Meta, Oracle, Amazon, and Apple—will account for approximately two-thirds of this total energy demand.
  • Large data centers can consume up to 5 million gallons of water daily for cooling, impacting local resources in states like Virginia and Arizona.
  • Infrastructure operations contribute to noise pollution and air quality issues through the use of backup diesel generators and industrial cooling fans.

Why it Matters

The rapid expansion of generative AI infrastructure creates a massive, localized environmental burden that challenges current energy and water management policies. As these facilities cluster in specific regions, they risk overwhelming local utility grids and natural resources, necessitating more sustainable infrastructure planning.
Scientific American Published by Rachel Feltman, Andrea Gawrylewski, Alli Rodgers, Alex Sugiura
Read original