Nearly half of executives report that their organisation’s use of this technology is directly contributing to a rise in greenhouse gas (GHG) emissions, new research has suggested. According to a new study from the Capgemini Research Institute which polled 2,000 executives from large companies across 15 countries and 12 sectors, firms measuring the environmental impact of their generative AI (Gen AI) deployments found that their predicted total share of emissions will double from 2.6% to 4.8% over the next two years. Meanwhile, only 12% of organisations currently track this footprint, revealing an additional gap in sustainability practices as Gen AI adoption accelerates.
“If we want Gen AI to be a force for sustainable business value, there needs to be a market discussion around data collaboration, drawing up industry-wide standards around how we account for the environmental footprint of AI, so business leaders are equipped to make more informed, responsible business decisions, and mitigate these impacts,” said Capgemini’s global sustainability services and corporate responsibility head Cyril Garcia.
The report points to the energy-intensive nature of Gen AI, which requires significant computational resources and data processing power, as a key contributor to its growing carbon footprint.
Despite the increasing reliance on Gen AI, only 38% of executives are aware of its environmental impact, and just 20% rank sustainability as a top-five consideration when selecting or deploying AI models. Scalability, performance, and cost remain dominant factors in decision-making processes.
However, more than half of the survey respondents acknowledged that prioritising sustainability in vendor selection could help reduce emissions, signalling an opportunity to align business objectives with environmental goals.
Gen AI adoption outpaces mitigation efforts
Gen AI adoption has grown significantly, with 24% of organisations reporting its integration across business functions in 2024, compared to just 6% in 2023. Despite this growth, only 31% of organisations have introduced sustainability measures into their Gen AI operations. Most remain reliant on pre-trained models provided by vendors, with only 4% developing their own models.
Efforts to mitigate emissions are underway, with over half of the organisations either adopting or planning to adopt smaller models and renewable energy-powered infrastructures within the next year.
The report highlights the need for collaboration across the Gen AI ecosystem to establish clear methodologies for assessing and managing the environmental impact of the technology. Nearly three-quarters of executives cited the lack of standardised measurement frameworks as a major obstacle.
Governance frameworks and stricter guardrails were identified as critical tools, with 62% of respondents expressing confidence that these measures could effectively mitigate the environmental footprint of Gen AI.
The Capgemini study calls for a strategic approach to deploying Gen AI, including comprehensive evaluations of financial returns and environmental impact prior to implementation. It also recommends adopting sustainable practices, such as optimising model architecture, leveraging renewable energy, and establishing usage policies that minimise unnecessary resource consumption.
Currently, one-third of organisations are using Gen AI for sustainability initiatives, including ESG reporting, circular product design, and material optimisation. Two-thirds anticipate that these efforts will result in GHG emission reductions of over 10% within three to five years. However, the study cautions that such outcomes require robust measurement and monitoring to validate their impact.
Read more: Data centres could drive 2.5bn tons of CO2 emissions by 2030, warns Morgan Stanley
