Double click here to edit new New Header component
Investment Bank
How can we help?
X
  • Contact Us
  • Client Login
    • Search
    Parsys 1
    Double click here to edit Hero Media component
    RESEARCH  |  IMPACT SERIES  |  INNOVATION EDGE

    AI revolution: Meeting massive AI infrastructure demands

    authorImage authorImage authorImage 
    Contributors: Christian Keller, William Thompson & Akash Utsav 
    16 Jan 2024

    The race to implement AI is well under way, as companies around the world have begun integrating AI tools. Big tech firms are responding with huge investment in data centres that can train and run models. This rapid growth is placing strains on resources and raises new, urgent questions for the public and private sectors on how to deliver AI responsibly and sustainably.

    Download the report
    Series Image
    An AI chip data centre using electricity
    Parsys 2
    Double click here to edit Go To Section component

    CHAPTERS

    Parsys 3

    AI’s thirst for power has always been strong – it is often cited that AI queries require about ten times the electricity1 of a traditional search over the web. And aggregate demand seems bound to rise further, as more and more companies look to exploit advantages that AI can bring, while tech firms train ever-larger models in an effort to achieve human-like intelligence.

    Beyond its energy demands, scaling AI also requires land, water and materials to make semiconductors and data centres. The companies and countries that rise to meet these demands stand to benefit. But scaling AI is a bit like walking a tightrope: balancing the desire to reap the economic opportunities and geostrategic benefits expected from AI adoption, while respecting existing frameworks on issues from net zero to intellectual property.

    Related content

    AI revolution: productivity boom and beyond

    Our Research analysts explore how recent breakthroughs in Artificial Intelligence could provide a boost to productivity, similar to past periods of revolutionary technology change.

    Can energy demand keep pace?

    Data centres are the “factories” for AI but also serve as warehouses and distribution centres for the storage and dissemination of information for many other applications including websites, streaming services and cloud computing.

    Most of the world’s more than 11,000 registered data centres are not yet involved in any kind of AI-related activity. Combined, their consumption of electricity (excluding cryptocurrencies) is about 1.0%-1.5% of the world’s total, according to a mid-2024 report from the IEA2.

    However, data centres' energy demands could change dramatically in coming years, thanks to the dissemination of AI. According to a June 2024 analysis from Barclays Research – which uses a bottom-up approach based on utilities’ forward-looking supply contracts – annual demand to power data centres in the US could grow by a range of 14%-21% every year through 2030. That would imply US data-centre demand roughly tripling by 2030, from 150-175 terawatt hours (TWh) in 2023 to as much as 560 TWh – equivalent to 13% of current US electricity demand.

    Granted, there is a lot of uncertainty around these bullish projections. But the broader point is that AI and data centres could provide a much bigger spur to global electricity demand than the sums initially factored in to net-zero targets.

    Related content

    AI is hungry for power

    Data centres could make up 9% of US electricity consumption by 2030. Our Research analysts examine the implications for investment and sustainability mandates.

    Shifting the geography of the global economy

    Industry growth is most obvious in the US, where spending on data-centre construction has more than doubled since the release of OpenAI’s ChatGPT in November 2022, according to US Census Bureau data.

    But the US will surely not be expanding alone. Concerns over the sovereignty of AI technology are likely to make all nations reluctant to outsource activities, to any great extent. In China, for example, the so-called "Eastern Data, Western Computing"3 project involves plans to construct eight major data-centre hubs. The Gulf region could become a particularly large player in the industry, as it combines abundant energy and vast amounts of capital. Europe, too, will have to respond, as it grapples with a number of disadvantages, including the loss of cheap Russian gas, tight fiscal constraints and ambitious emissions targets. It also has some of the strictest data-protection standards in the world.

    Ambitions at odds?

    Clearly, the hyperscalers themselves have a key role to play in the decarbonisation push. Amazon, Meta, Alphabet and Microsoft are among the biggest buyers of power under long-term, clean-energy contracts, according to Bloomberg New Energy Finance.

    But there is a tension here. What these companies are seeking is sources of power that are “firm” – available at all times, even under adverse conditions – and also “dispatchable” - programmed on demand at the request of grid operators, according to market needs. On both counts, renewables do not score highly.

    One source of power that does score highly is nuclear, as it provides stable, carbon-free electricity that also reduces overall decarbonisation costs: by reducing the over-procurement of renewables; by limiting the need for batteries and storage associated with solar and wind; and by creating extra benefits, such as the production of hydrogen.

    Small modular reactors (SMRs) – facilities that have about one-third the generating capacity of a traditional reactor – are in sharp focus, as they are relatively quick and cheap to build and enable on-site generation that can bypass the regulated utility process. Amazon said in October 2024 that it had bought a stake in nuclear developer X-energy, as part of a collaboration4 with the company aimed at deploying SMRs to provide low-carbon electricity to power its data centres.

    Geothermal energy, too, is gaining traction. Microsoft, flanked by Abu Dhabi-based AI company G42, is investing in a data centre powered by geothermal energy in Olkaria, Kenya5, and has also signed a geothermal PPA in Aotearoa, New Zealand6.

    Ultimately, the balancing act between achieving emissions targets, while allowing responsible AI technology to advance, is a global challenge. Policymakers, tech companies and the energy industry must forge new partnerships to deliver AI that is socially beneficial and environmentally sustainable.

    1 EPRI Study: Data Centers Could Consume up to 9% of U.S. Electricity Generation by 2030 
    2 International Energy Agency Electricity Mid-Year Update July 2024 
    3 Huawei
    4 Amazon
    5 Microsoft
    6 Microsoft

    Parsys 4
    Double click here to edit Banner component
    Double click here to edit Experts Panel Card component

    Barclays Research contributors

    Christian Keller

    Head of Economics Research

    Christian Keller is Head of Economics Research at Barclays, leading a global team covering both developed and emerging markets. He is based in London and joined Barclays in 2007 from the International Monetary Fund (IMF), where he had worked on programs with emerging market economies in Europe, Latin America and Asia. He graduated with a PhD in Economics from University of Köln, Germany, and holds a joined-MA in Economics and Finance from University of Köln and HEC, Paris.

    Will Thompson

    Director on the Thematic Investing Research team

    Will Thompson is a Director on the Thematic Investing Research team and focuses on the energy and industrial sectors. Will joined Barclays Equity Research in 2014 and was most recently a senior analyst covering Small and Medium Cap Oil & Gas Exploration & Production (E&P) companies. Prior to that, he was part of the Oilfield Services & Equipment research team, where he ranked as an Institutional Investor Rising Star in 2017. Will previously worked in equity research in various other investment banks, covering Oilfield Services & Equipment and Airfreight, Surface, and Marine Transportation. He also spent time as a Senior Research Associate at an investment firm. Will has a Bachelor of Engineering from Dartmouth College and a Bachelor of Arts in Chemistry from Colby College.

    Akash Utsav

    Economist in the Global Economics Research team

    Akash Utsav is an economist in the Global Economics Research team at Barclays. He joined the bank in 2017 and contributes to global macroeconomic thematic notes, as well as to quarterly publications. He holds an MSc Economics degree from University College London (UCL), London, UK, and completed his undergraduate from Loyola College, Chennai, India.

    Important content disclosuresImportant content disclosures
    Parsys 5
    iParsys for Double Pixel component