Artificial Intelligence and the increasing demand for electricity
In this week’s video insight, I discuss the significant impact of artificial intelligence (AI) on companies, particularly the “Magnificent Seven.” AI is driving up the power needs of data centres, far beyond traditional levels. I explain this with the concept of rack power density and highlight predictions about U.S. data centre electricity use. I also mention Amazon Web Services’ recent energy deals and how these changes might influence Australia’s energy policies, including the debate on nuclear power.
Transcript:
Hi, I’m David Buckland and welcome back to this week’s video insight.
Artificial intelligence (AI) is a powerful theme driving the share price and outlook for several technology businesses, particularly many of the “Magnificent Seven”, which on average jumped by 111 per cent in calendar 2023, and by a further 37 per cent in the June 2024 half-year.
A big shift coming from AI is associated with incredible demand from data centre power compared to traditional data centres. Firstly, it is essential to define the term rack power density. This is the power draw of a single, fully populated server rack, measured in kilowatts (KW).
In a recent presentation from the U.S.-based Dominion Energy Inc (NYSE:D), a typical data centre without AI demand has a rack power density of 6-12 kilowatts (KW) from central processing unit (CPU) based servers.
However, a data centre supporting AI in the Training phase – the process that enables AI models to make accurate inferences – has a rack power density of 26-80KW, up 4 to 7.5 times, from high-powered graphics processing unit (GPU) based servers.
A data centre supporting AI in the Inference phase – the process that enables AI models to produce predictions or conclusions – has a rack power density of 12-40KW, up 2 to 3.5 times, from a combination of CPU and GPU based servers.
In terms of U.S. data centre electricity demand from 2023 to 2030, Goldman Sachs is forecasting compound annual average growth of 15 per cent, driving data centres to account for eight percent of total U.S. electricity demand by 2030, up from the current three per cent.
Interestingly Amazon Web Services recently announced the purchase of Talen Energy’s “1,200-acre data centre campus”, which is next to its 2.5 gigawatt (GW) nuclear power plant, in Pennsylvania. Amazon Web Services is also looking at securing energy for its data centres from Constellation Energy, the largest owner of U.S. nuclear power plants.
Down under, businesses like Next DC (ASX:NXT) and Infratil (ASX:IFL) have data centre aspirational targets by 2033 approximating six times their current capacity, to 950 megawatts (MW) and 1,870 MW, respectively.
Given big tech is heading down the nuclear path, and France is deriving around 70 per cent of its electricity from nuclear energy, it seems likely the increase in electricity demand in Australia from AI over the medium-term will only boost the debate on nuclear energy.
The likes of Ziggy Switkowski and Dick Smith are calling for a coherent energy policy, lobbying the Australian government to lift the nuclear moratorium to promote clean energy whilst avoiding potential blackouts.
The Montgomery Fund owns shares in Infratil. This article was prepared 4 July 2024 with the information we have today, and our view may change. It does not constitute formal advice or professional investment advice. If you wish to trade Infratril, you should seek financial advice.