Artificial intelligence is driving the largest increase in US electricity demand in decades, yet environmental impact remains a secondary consideration in most companies’ responsible AI strategies, according to a new report from The Conference Board.
The study found that only 13% of surveyed sustainability leaders consider environmental sustainability a core or major element of their responsible AI approach, while 31% said it ranks behind concerns such as ethics, bias and security. A further 42% said environmental sustainability is a minor consideration or not a priority at all.
The findings come as data centres supporting AI workloads account for a growing share of US electricity and water use. Energy demand, consumption and emissions were identified as the top environmental concerns linked to AI, with 63% of respondents citing data centre energy demand, 58% pointing to overall energy consumption and 56% to greenhouse gas emissions from electricity use. Water use was flagged by 37% of respondents, while only 5% said they were not concerned about AI’s environmental footprint.
“At a time when AI investment continues at record pace, its environmental footprint is becoming impossible to ignore,” said Andrew Jones, author of the report and Principal Researcher at The Conference Board. “Data centres already account for a growing share of US electricity demand, and water use is rising as AI workloads scale. Yet the same technology is also unlocking new tools for decarbonisation, grid optimisation, and operational efficiency.”
Jones added that leading companies will need to take a “dual lens” approach. “In 2026, the leading companies will be those that manage AI’s resource demands while harnessing AI to accelerate sustainability outcomes,” he said.
The report draws on responses from more than 60 corporate sustainability leaders at large US and multinational companies. It also examines how firms are using AI to advance sustainability objectives. While more than 60% of respondents said they are using AI for environmental purposes, applications remain concentrated in lower-impact areas.
Around 34% of surveyed leaders said they are using AI for sustainability-related disclosure and reporting, while 22% are applying it to carbon accounting and emissions tracking. Fewer companies are using AI for operational improvements, with 15% using it for climate risk modelling and scenario analysis, and 12% for circularity, waste and water management.
“AI’s environmental story is not only about its footprint—it is also a promising toolkit for sustainability performance,” said Brian Campbell, Leader of The Conference Board Governance & Sustainability Center. “While early applications centre on reporting and disclosure, the highest-value opportunities lie in emerging operational uses that can drive far greater environmental impact.”
The report highlights the scale of AI’s environmental footprint, particularly through the rapid expansion of data centres. Hyperscale data centres have doubled globally over the past five years, with more than half located in the US. The country now hosts more than 4,250 data centres, with around a third concentrated in California, Texas and Virginia.
Energy consumption among major cloud providers has risen sharply. Three of the largest US cloud companies doubled their electricity use between 2021 and 2024, with their combined consumption in 2024 equivalent to around 2% of total US electricity generation.
Beyond energy and water use, the report points to wider environmental pressures linked to AI, including the carbon- and water-intensive production of hardware such as graphics processing units, increased demand for emissions-heavy construction materials, rising electronic waste from faster hardware refresh cycles, and growing reliance on critical minerals including cobalt, nickel, copper and rare earth elements.
Despite these challenges, the report notes that AI could play a larger role in improving sustainability performance if applied more widely to operations. Potential use cases include optimising energy use, improving logistics and fleet routing to reduce fuel consumption, enhancing water and climate resilience through predictive tools, and supporting circularity through AI-enabled waste sorting and recycling systems.
However, the study concludes that most companies have yet to fully exploit these higher-impact applications, leaving a gap between AI’s growing environmental footprint and its potential to support sustainability outcomes.
