AI Fuels CEOs’ Net-Zero Confidence, Raises Environmental Concerns

KPMG reported that more than 50% of CEOs are confident in reaching their net-zero emissions targets by 2030. The consultancy noted that AI has become a decisive tool for advancing these commitments. However, the environmental implications of this technology remain significant.

The 11th edition of the KPMG Global CEO Outlook reveals that although confidence in the global economy has dropped to levels similar to those seen during the pandemic, 79% of CEOs remain optimistic about their own organizations’ prospects. In the realm of corporate sustainability, the report shows that most business leaders continue to demonstrate strong commitment to their environmental goals and feel increasingly confident in their ability to meet them.

According to the report, 61% of CEOs say they are on track to achieve their net-zero targets by 2030—an increase of 10 percentage points from 51% a year earlier. This progress may be the result of companies reviewing and adjusting their interim climate goals to better align with their broader business strategies, the document explains.

Another key area of focus is ESG reporting. About 51% of CEOs said they are prioritizing regulatory compliance and disclosure standards to respond to the changing demands of investors and regulators. In addition, 65% stated that sustainability has been fully integrated into their business models and that they view it as critical to long-term success.

However, there are still gaps in the incorporation of sustainability criteria into capital investment decisions, with only 29% reporting that they have fully embedded ESG considerations into their investment frameworks. Among the main challenges to achieving these objectives is the complexity of decarbonizing supply chains, cited by 25% of CEOs. Meanwhile, 21% pointed to a lack of skills and expertise to implement effective solutions, and 11% mentioned cost constraints, reflecting a trend similar to that seen in 2024.

At the same time, more CEOs are recognizing AI’s potential to support decarbonization and sustainability efforts. Over 71% identified AI as a key investment priority, and 69% said they are allocating between 10% and 20% of their budgets to AI-related initiatives. The main applications of AI in the corporate sector include improving the quality of sustainability data and reporting, as noted by 79% of CEO’s, identifying opportunities to optimize resource use with 79%, and reducing emissions while enhancing energy efficiency with 78%.

However, AI itself comes with its own environmental challenges. Generative AI requires enormous power density. Noman Bashir, Computing and Climate Impact Fellow, MIT Climate and Sustainability Consortium (MCSC), noted that generative AI training clusters may consume seven or eight times more energy than a typical computing workload.

In North America, the power demand of data centers rose from 2,688MW at the end of 2022 to 5,341MW by the end of 2023, partly driven by generative AI’s growing requirements. Globally, data center electricity consumption reached 460TWh in 2022, enough to make data centers the world’s 11th-largest electricity consumer, between Saudi Arabia and France, according to the Organization for Economic Co-operation and Development. By 2026, data centers’ electricity use is projected to reach 1,050TWh, which would make them the fifth-largest consumer of electricity globally, between Japan and Russia. 

Although not all data center operations involve generative AI, the technology has been a major driver of rising energy demand. “The demand for new data centers cannot be met sustainably. The pace at which companies are building them means most of the electricity to power them must come from fossil fuel-based plants,” Bashir warned.

Beyond energy, data centers also consume vast amounts of water. Chilled water systems are used to cool computing equipment by absorbing heat. Bashir estimated that for every kilowatt-hour of energy consumed, a data center requires about 2L of water for cooling.

Additionally, the computing hardware used in data centers contributes to environmental impacts of its own. GPUs, powerful processors designed for handling intensive generative AI workloads, carry significant carbon footprints due to emissions from material sourcing and product transportation. Extracting the raw materials needed to manufacture these components often involves environmentally harmful mining practices and the use of toxic chemicals in processing.

“We need a more contextual way to systematically and comprehensively understand the implications of new developments in this space. Because of the speed of progress, we have not had the chance to catch up with our ability to measure and understand the trade-offs,” said Elsa Olivetti, Professor in the Department of Materials Science and Engineering, MIT.