Investment dilemma: The hidden carbon cost of AI

From optimising energy consumption to waste management, artificial intelligence (AI) has the potential to be a powerful ally in the fight against climate change. Several companies already employ AI to reduce their environmental footprint. Take Amazon, for example, which uses AI tools to cut emissions across its facilities; their ‘The Package Decision Engine’ has avoided over two million tons of packaging waste.1 Similarly, Google’s DeepMind has reduced energy consumption in its data centres by 40%. This has been achieved by using predictive models for efficient system adjustments.2 However, the deployment of AI comes with a trade-off: increased energy consumption.

As AI’s capabilities grow, so does its energy consumption. Take, for instance, a single interaction with ChatGPT, which is estimated to consume ten times the electricity of a standard Google search.3 This increased demand for power is largely driven by data centres, the backbone of AI services. By the end of this decade, these data centres are projected to account for over 20% of the growth in global electricity demand. In 2030, the United States is projected to use more electricity for data processing than to produce all energy-intensive goods combined - such as aluminium, steel and cement.4 Tech giants are already responding to this challenge. For instance, Microsoft has secured nuclear power to meet AI related energy demands5 and Google is investing in clean energy.6 Furthermore, Trane Technologies designs energy-efficient cooling and heating systems that are critical to reducing the carbon and water footprint of hyperscale data centres.

While AI drives up emissions, it simultaneously holds significant potential to combat climate change. The emissions depend on the energy sources used in the local power grid. If the grid uses more renewable energy, the emissions from AI will be lower; if it relies on fossil fuels, the emissions will be higher. ESG investors can direct capital toward companies that power data centres with renewable energy, as well as toward solution providers that work to reduce AI’s environmental burden.

_{1 How AI Tools are Powering Amazon’s Sustainability Goals | Technology Magazine
2 DeepMind: Google’s AI saves the amount of electricity used in data centres | WIRED
3 International Energy Agency. “Electricity Market Report - Analysis and Forecast to 2026.” IEA, July 2023.
4 AI is set to drive surging electricity demand from data centres while offering the potential to transform how the energy sector works - News - IEA
5 Microsoft in deal for Three Mile Island nuclear power to meet AI demand
6 Google kicks off $20B renewable energy building spree to power AI | TechCrunch}