California and the EU Move to Rein In the Environmental Cost of AI

California and the European Commission this week introduced new transparency and permitting rules aimed at curbing the growing water, energy and land footprint of AI infrastructure. The moves follow a United Nations University study warning that AI could by the end of the decade consume water equivalent to 1.3 billion people in sub‑Saharan Africa and greatly expand global energy and land use.

By AI Newsroom· Reviewed by Pranav, Founder & Editor-in-ChiefPublished 1 minute agoUpdated 1 minute ago0 views

Why It Matters

Policymakers are responding to evidence that hyperscale data centers and AI compute are placing large, uneven burdens on water, power grids and land, and early disclosure and permitting requirements are being positioned as the first steps toward binding limits. Those rules could shape how and where high‑intensity AI infrastructure is built as governments seek to balance digital competitiveness with resource and community impacts.

Key Facts

  • Study: United Nations University report on AI resource use
  • Projected water use (by end of decade): Equivalent to 1.3 billion people in sub‑Saharan Africa
  • Projected energy comparison: Could triple the combined energy consumption of Pakistan, Bangladesh, and Nigeria (population ~650 million)
  • Projected land footprint: Potentially more than 14,500 square kilometres
  • European action: European Commission introduced rules requiring data centers to disclose energy and water use and report links to local water stress

Governments in California and the European Union moved this week to increase oversight of the environmental impacts of large-scale artificial intelligence infrastructure. The European Commission unveiled new disclosure rules that obligate data centers to report their energy and water consumption and to disclose how their water use relates to local water stress. The measure is a transparency-first approach, without immediate caps on usage, intended to inform future policymaking as the EU plans to expand its data center capacity over coming years.

In California, Governor Gavin Newsom signed seven bills that tighten permitting and approval conditions for data centers in the state. The package requires developers to estimate water demand, provide information on energy efficiency and drought contingencies, meet specified requirements for energy, water and fuel consumption to proceed in the approval process, and contribute funding for local grid upgrades where facilities will be sited.

These policy steps respond to findings from a United Nations University study highlighting rapidly rising resource demands from AI. The report estimates that, by the end of the decade, AI could consume water on a scale comparable to 1.3 billion people in sub‑Saharan Africa, increase energy use to levels exceeding the combined consumption of Pakistan, Bangladesh and Nigeria, and require a land area possibly surpassing 14,500 square kilometres. UN researchers framed the findings as a call to manage AI’s growth so it fits within planetary limits and to address impacts on communities that supply critical minerals or host infrastructure and electronic waste.

Both jurisdictions framed the new measures as necessary to protect local communities and infrastructure while the sector scales. The EU’s disclosure rules aim to reduce opacity around data center resource use, a key challenge for crafting effective regulation, while California’s statutory package ties approvals to concrete estimates and mitigation measures, including funding for grid improvements. Policymakers and advocates say transparency and labeling can lay the groundwork for eventual limits or other binding controls if needed.

Keep Reading