Watts Next for AI?
Artificial intelligence (AI) is currently the hottest topic being discussed by companies and investors alike. Although much has been spoken about how it will revolutionise almost every industry, far less focus has been placed on its potential environmental impact. AI is incredibly energy-intensive, primarily due to the vast amount of data that the models are trained on, as well as the volume of requests made to the AI by users. These factors are further amplified by the complexity of the models, with more advanced programmes requiring a greater amount of power. According to Forbes, it is estimated that the daily requests sent to ChatGPT amount to half a million kilowatts of electricity, equivalent to the daily power use of nearly 180,000 US households.
Moreover, the rapidly growing AI industry is also heavily reliant on water, Earth’s most precious resource. The massive energy consumption of AI servers creates a significant amount of heat, which needs to be quickly dissipated through cooling systems that are fed by fresh water. This need for water further exacerbates the industry’s growing environmental footprint. Microsoft consumed around 22bn litres of water in 2022, an increase of 34% over the year before, while Google’s usage grew by 22%. A single conversation with ChatGPT uses about a 500ml bottle’s worth of water. With 200 million requests being sent through daily, an extremely substantial amount of water is required.
Positively, the companies involved in AI have recognised their role in pursuing these new technologies responsibly. Microsoft has committed to being “water positive” by 2030, which means they will replenish more water than they use. Both Microsoft and Amazon have been exploring the use of next-generation nuclear reactors to power their data centres, with Amazon purchasing a nuclear-powered data centre in Pennsylvania earlier this year. Nuclear is considered a clean non-renewable source of energy, aiding Big Tech in achieving their sustainability goals.
As AI technology is still in its infancy, we can expect a significant amount of innovation to occur within the energy sector, as industry players grapple with the potential of heightened demand in the near future. In March’s newsletter, we wrote that we expect copper to have a very strong growth profile because of rising demand from the energy transition. We believe that this, alongside the energy consumption associated with AI, will continue to make copper an attractive commodity to hold. We own mining companies with significant copper assets in our various funds, namely BHP Group, Glencore, and Teck Resources.