AI is being used to address the major issues that are hindering our planet . As semiconductor technology continues to squeeze more ‘power per watt’, is AI’s environmental cost likely to be outweighed by the ‘green dividend’ it pays back ?
From predicting extreme weather to optimising energy use, AI is tackling various environmental challenges. But there is a concern: AI craves a lot of energy. So, is AI an environmental boon or a burden? The answer is a bit of both – but with reason for optimism.
AI grappling climate change issues
AI’s ability to process vast amounts of data makes it a powerful tool to help solve complex environmental problems, for instance forecasting weather faster and potentially more accurately by learning patterns from decades of climate data. AI’s pattern-spotting skills are also saving energy: Google has been able to achieve more than six times more computing power per unit of energy versus five years ago through various efficiency measures. (Source: https://sustainabilitymag.com/news/how-is-google-balancing-ai-demand-with-sustainability)
In wildlife conservation, AI is an impressive forest guardian. Autonomous drones and acoustic sensors can monitor protected rainforests, instantly alerting rangers to illegal logging. In Africa, cameras with AI software distinguish endangered animals from poachers, enabling quicker interventions. In agriculture, farmers now deploy AI-powered apps and sensors to monitor soil moisture and pest outbreaks in real time – using less water, fewer chemicals, and boosting crop yields.
AI can also help with renewable energy, where unpredictable sun and wind pose a challenge. AI systems can now balance energy supply and demand by predicting output from solar panels and wind turbines, helping utilities store energy when it is plentiful and distribute it efficiently when it is needed.
The invisible footprint of AI
AI’s environmental impact is a real consideration. Those algorithms run on physical hardware – thousands of servers humming away in data centres, consuming electricity and generating heat. Training a single advanced AI model involves immense computing power.
Data centres rely on energy-intensive air conditioning and vast amounts of water for cooling. Collectively, all data centres, according to the International Energy Agency (IEA) – including those for AI – consume over 1% of global electricity and emit roughly 0.5% of worldwide greenhouse gases. (Source: https://www.carbonbrief.org/ai-five charts-that-put-data-centre-energy-use-and-emissions-into-context/).
Greener computing
Tech companies are acutely aware of this challenge – and they’re innovating at speed to improve AI hardware energy efficiency. One big trend is specialised semiconductor chips designed just for AI tasks.
Simultaneously, engineers are reimagining how to power and cool big data centres. Smarter cooling systems dynamically adjust cooling loads and even use outside air or liquid cooling to curb power use. Companies are also investing in renewable energy, placing new data centres near wind and solar farms to ensure clean energy sources.
Can AI solve its own carbon footprint?
Making AI environmentally friendly is about balance and ingenuity. Done well, AI could save far more emissions than it produces. According to the IEA, by 2035, AI-driven improvements in energy, transport, and agriculture could help slash global carbon pollution by billions of tonnes (Source: https://www.iea.org/reports/energy-and ai/ai-and-climate-change). If AI’s superpower is solving complex problems, one of the biggest problems it may solve is its own carbon footprint.
How can investors benefit?
We have always focused on identifying compelling long-term investment themes and seeking effective ways to benefit from them. On this basis, we have identified opportunities in semiconductors – in our view, the backbone of all digitalisation – and in the enterprise software space, which is characterised by high recurring revenues. We have also invested in companies exposed to the data centre build out, where the lines between our ‘atoms’ and ‘bytes’ approach start to blur.
We believe that AI is undergoing a multi-decade investment boom, and as with all new trends or technological developments, there will be winners and losers, making a targeted approach essential.
Past performance is not a guide to future performance. The value of investments and the income from them may go down as well as up and is not guaranteed. Capital at risk. Investors may not get back the amount invested. The views expressed are those of the speaker at the date of publication unless otherwise indicated, which are subject to change, and is not investment advice.
Contact details: For more information on Royal London Asset Management’s product please contact bdsupport@rlam.com
