With the global need to transition toward net zero, focus has been on carbon intensive industries with the greatest potential to accelerate action - images of smoke-stack industries, exhaust fumes, mineral extraction, fossil-fuel energy generation and livestock production historically dominate the conversation around climate change.
However, significant advancements in the innovation and adoption of AI have changed that and the role of technology in contributing to climate risks is becoming increasingly apparent. In 2020, Microsoft committed itself to be carbon negative, water positive, and zero waste by 2030 to meet its sustainability promises.[1] However in its latest Environmental Sustainability Report, Microsoft revealed that its total carbon emissions had risen 30% over the past four years, primarily driven by innovation in artificial intelligence (AI) and increased construction and usage of data centers. Goldman Sachs[2] now projects that AI is poised to drive a 160% increase in data center power, while Arm CEO Rene Haas warns that AI may consume up to 25% of the entire power grid usage in the U.S by 2030.[3] Presently, 200 million ChatGPT requests are made daily, translating to over half a million kilowatt-hours of electricity each day. In comparison, a typical US household uses about 29 kilowatt-hours per day. The International Energy Agency projects that by 2026 AI globally will consume as much electricity as Japan[4].
Unsustainable practices often result in higher operational costs due to inefficiencies. They can also result in very real costs for organizations that don’t comply with growing regulation requirements and their own targets. Volkswagen and the financial impact of having to pay more than $20 billion in fines, penalties and settlements over vehicle emissions[5] is a worst case scenario of mismanagement, but a reminder of the values at risk in an era of growing litigation.
This puts the industry somewhat in a place of contradiction. Technology in the broadest sense is sensibly and justifiably seen as offering solutions to many of the issues associated with the greenhouse gas emissions that are causing climate change. Yet, until those solutions are more numerous and widespread, and during the transition to net zero, AI is itself increasingly seen as a contributor and susceptible to increasing climate risks without meaningful action.
“While fossil fuel consumption remains a large obstacle for organizations to neutralize their carbon footprint, the energy needed to power AI as outlined here will assert itself quickly as an emerging risk for ESG.”
Jennifer Caldarella | Managing Director, WTW Large Accounts Strategy
In our latest WTW Risk & Resilience Review we shine a spotlight on the sustainability of AI. Supporting the WTW smarter way to risk, this report introduces research and opinions that provide new perspectives to support risk management and resilience. Our previous report honed in on emerging risks from geopolitical shifts, with a focus on supply chains and national competition. For this edition, we incorporate research work from our academic partners alongside expert opinion and views from within WTW to focus on the AI sustainability dilemma.
Written using research work undertaken as part of the Wharton Mack Institute Collaborative Innovation Program, our opening piece in Section 2.1 drills into one of the key areas in which the growth in AI is challenging sustainability aims – energy consumption. The piece identifies three key areas of energy use in the architectural framework supporting AI, and present practical best practices to consider for technology firms.
In Section 2.5, we concentrate on the energy sector with our Natural Resources team. Energy use could double by the end of the century, with geopolitical tensions, new technologies, a changing climate and variable economic outlooks all key drivers influencing the direction and pace of change. The momentum behind electrification is growing, and the power industry is building resilience to meet the growing energy needs, with AI at the forefront of both increased demands, but also helping to manage energy needs.
When fully realized, AI could revolutionize how organizations capture and utilize data to enhance risk management and overall sustainability in the supply chain, alongside improvements in sales and operations planning. With advancements in technology, organizations can model the impact of decisions on risk, sustainability, and efficiency, particularly in supplier procurement and network management. In Section 2.2, we explore how AI can transform supply chain sustainability and risk management through embedded AI.
Supply chain issues have specific considerations when looking at the semi-conductor industry – a crucial component of AI. The sustainability and politics of natural resources is creating competition and friction across the global supply chain. In Section 2.3 we explore these dynamics with Cullen Hendrix from the Petersen Institute
In November 2023, then-U.K. Prime Minister Rishi Sunak invited world’s leaders, tech executives and academics to a large AI Safety Summit at Bletchley Park, a symbolic choice of location as the home of the codebreakers that cracked the enigma code, with the aim to align the world view on the development of AI and to agree on a way forward to help manage the integration of AI into society. The results included notable announcements on a multilateral agreement for governments to test AI models, for developers to submit their models to testing, and an international advisory panel to advise on AI risk based on the Intergovernmental Panel on Climate Change. Section 2.4 looks at the UK regulatory framework as a case study to foster innovation and boost public trust in AI applications.
PWC estimates that the application of AI levers could reduce worldwide greenhouse gas emissions by 4% in 2030, an amount equivalent to 2.4 gigatons CO2e – equivalent to the projected 2030 annual emissions of Australia, Canada and Japan combined.[6] In our final piece, Section 2.6, we explore creating a sustainable future with AI, focusing on how AI could be key to tackling environmental challenges across industries.
