AI and environmental impact: from a vicious circle to a virtuous one

The rapid growth of artificial intelligence (AI) has cast a shadow over the climate ambitions of leading tech companies. Emissions from their data centres are rising strongly, as a result of the power-hungry processing needed for AI computations.

This recent trend has been seen as a negative for efforts to slow and reverse global warming. But there is another side to the story: the AI revolution will spur tech firms to accelerate efforts to address their rising energy use, boosting investments in climate technologies, driving the next stage of growth within environmental markets and, ultimately, accelerating wider efficiencies in how the global economy uses energy.

The International Energy Agency has warned that energy consumption by data centres could double between 2022 and 2026. Its latest electricity market forecast estimates that power consumption by data centres (and cryptocurrencies) could reach 1,000 TWh by 2026, roughly equivalent to that of Japan. That’s up from 460 TWh in 2022. 

The AI industry realises it has a problem. “An energy breakthrough is necessary for future artificial intelligence, which will consume vastly more power than people have expected,” Sam Altman, CEO of OpenAI, told the World Economic Forum meeting in January 2024.

Tech companies, many of which have set ambitious climate targets, are seeing those targets threatened by skyrocketing growth in AI.

But these companies are not passive actors in the face of rising energy use and emissions. They are already major investors in renewables, with Amazon the world’s largest purchaser of renewable energy for the fourth year in a row in 2023, followed by Meta in second place and Google in fourth. In May, Microsoft was reported to have signed the world’s largest single contract for renewables, investing an estimated $10 billion with Brookfield to develop more than 10.5 GW of wind and solar.

They have also helped drive the outperformance of companies working to improve the energy efficiency of the data centres they use. Big Tech has been quietly but effectively boosting data centre efficiency behind the scenes for many years, whether through chip design or cooling, building hyperscale data centres or the use of virtualisation technology.

Our FTSE Environmental Opportunities All Share (EOAS) index tracks the performance of companies active in the green economy. They been growing strongly, delivering a 13.8% compound average growth rate (CAGR) over the previous 10 years. If the green economy were considered as a standalone sector, it would be the second-best performing industry over the period, outpaced only by the Technology sector. From its inception in 2008 to the end of March 2024, the EOAS outperformed its benchmark by 82%.

Energy Management and Efficiency has long been one of the largest sectors within the index: it currently accounts for 46% of the green economy by market capitalisation. It is also the second-best performing, with a five-year CAGR of 17% (after Transport Equipment, with a five-year CAGR of 24%, driven by electric vehicle, battery and railway equipment names).

Demand from tech firms has helped drive the growth in Energy Management and Efficiency. These firms have also contributed to a re-shaping of the theme, shifting its orientation from green buildings and industrial applications towards cloud computing (the largest sub-sector) and efficient IT equipment and electronics. 

However, the rapid progress the sector has made in improving the energy efficiency of IT has stalled.

The dramatic step-up in energy demand from AI will require a similar step-up in efficiency, particularly in the energy use of chips, if Big Tech is to meet its climate goals and rein in costs.

If the past is any guide, we can be reasonably confident that chip designers and manufacturers, data centre operators, and providers of related IT services will deliver such improved efficiency.

First, there is an obvious incentive to do so to cut costs associated with energy use. This incentive is reinforced by issues around energy security and geopolitics. There is a clear technological imperative: the more efficient that chips become (through reducing the heating effect from wasted energy) the more can be done with them in terms of capacity and power.

There are also regulatory tailwinds: in the EU, for example, revisions to the Energy Efficiency Directive will, from September 2024, require data centres to report key performance indicators. It introduces a sustainability rating scheme that will encourage improved performance.

The EU AI Act has also introduced energy-efficiency as a necessary priority for AI developers. As part of the EU AI Act implementation, the newly established EU AI Office and the Member States will facilitate the drawing up of codes of conduct. These will concern among others the voluntary application of specific standards assessing and minimising the impact of AI systems on environmental sustainability, including as regards energy-efficient programming and techniques for the efficient design, training and use of AI. Moreover, providers of general-purpose AI models will need to keep technical documentation which now includes information about known or estimated energy consumption of the model.

There is also a virtuous circle at work. Chip designers and manufacturers are applying AI to the challenges of chip production. AI is being increasingly applied to power management, improving the efficiency of manufacturing plants, data centres and power grids alike. AI can be applied to improve sustainability, which in turn can make AI more sustainable.

The growth in AI is also good news for another sub-sector within environmental markets. Renewable energy names have struggled since 2022, in the face of overcapacity, rising interest rates and collapsing equipment prices. However, they rallied in the second quarter of 2024, with the sub-sector climbing 20% during April and May, in anticipation of higher demand from AI.

Without question, the growth of AI represents a significant new source of power demand, adding to that created by the electrification of large swathes of the global economy. But, as we have seen many times over the relatively short history of the green economy, that demand will drive innovation, investment and, we believe, outperformance for long-term investors in environmental markets.