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  |  MACRO SHIFTS

    Talent and trade: The new battlegrounds in AI

    authorImage authorImage authorImage 
    Contributors: Zornitsa Todorova, Vincenzo Pota & Maggie O'Neal
    01 May 2025

    Artificial Intelligence advancement is not dependent solely on energy – it hinges on a global pool of critical resources, including minerals and human talent. Our Research analysts consider how these scarce, highly concentrated inputs are extremely vulnerable to disruption, making them the new battlegrounds for AI supremacy between developed and emerging markets.

    Client log-in for full report
    Series Image
    Close-up of rare, broken mineral fragments with sharp, angular edges and rough textures
    Parsys 2
    Double click here to edit Go To Section component

    GO TO SECTION

    Parsys 3

    It is well known that AI requires vast amounts of electricity to power huge data centres and a lot of water to cool countless racks of servers. Yet less is said about critical inputs, including the raw minerals used to make components such as semiconductors or the human capital required to design and develop new tools and applications.

    In the case of AI-related products, trading channels are highly concentrated and therefore vulnerable to disruption. When it comes to human capital, supplies are similarly constrained.

    Related content

    AI revolution: Meeting massive AI infrastructure demands

    With AI adoption picking up pace, Barclays Research examines new, urgent questions about the technology’s massive infrastructure demands, and how public & private sectors might collaborate to deliver AI responsibly & sustainably.

    Chains of command

    Concentration in supply chains tends to occur for two reasons: deliberate, strategic choices on the part of buyers or limited suppliers. The latter type of concentration is often more critical, because it means a small group dominates the production technology or supply of minerals, leaving buyers with few options.

    Our Research analysts assessed degrees of concentration in AI supply chains by examining more than 30 million trade corridors between countries in the UN’s Commodity Trade Statistics Database. They found that the range of countries supplying AI-related products is narrow, on the whole. As of 2023, 37% of the value of AI-related imports – from minerals to semiconductors to computers – came from just three or fewer suppliers per product. That concentration was five percentage points higher than the equivalent figure for non-AI-related trade, which suggests a more acute level of concentration.

    More significantly, of that 37% share, just 7% was concentrated by strategic choice, meaning 30% was because of a limited menu of supplier countries. Again, that's a significantly higher skew than for non-AI trade, where concentration was caused more by design (18%) than a lack of options (14%).

    Is that risky? Not necessarily, according to our Research analysts. In normal circumstances, high levels of concentration do not pose serious threats to operations. But sudden geopolitical tensions, conflicts or shocks such as those currently present in the global economy can transform dependencies into structural vulnerabilities.

    The supply of finished products such as chips, computers and electrical boards is dominated by Asian economies, especially China, Taiwan and Korea. But our Research team’s analysis shows that emerging and developing economies are the primary sources of the minerals essential for these products.

    AI-related trade is concentrated due to a limited number of supplier countries

    Mining the elements driving AI

    The minerals used in AI components span the periodic table, including readily available elements such as silicon and phosphorus, as well as rare-earth elements such as germanium, gallium and thallium, which are prized for their magnetism, conductivity and resistance to heat. Other minerals such as cobalt, lithium, aluminium, nickel, and copper (collectively, ‘the CLANCs’) are crucial for energy storage, battery materials and electrical wiring.

    And while China is the dominant buyer, purifier and processor of many of these minerals, almost all of the actual mining happens in countries such as Chile (supplier of 20% of the world’s raw copper), the Democratic Republic of Congo (73% of raw cobalt) and Guinea (70% of aluminium ores). As a group, emerging and developing economies supply close to two-thirds (62%) of the world's raw rare earth elements and minerals. And demand is likely to rise: Congo’s cobalt ore exports have grown five-fold since 2010, while Guinea’s aluminium ore exports have grown twelve-fold.

    Map shows how 10 countries dominate the supply of finished AI products with heavy concentration in China, Taiwan and S Korea

    The implications are significant. Nations rich in critical minerals have the potential to establish new mining operations and value-added processing industries, helping to create jobs, spur economic growth and attract new investment from abroad. Moreover, by developing deeper and more reliable supply chains, these countries could gain geopolitical leverage, forming alliances with tech-intensive economies such as the US, the EU and Japan.

    Meanwhile, importing nations can do their best to reduce vulnerabilities by diversifying suppliers or recycling used materials more efficiently, although extraction tends to be difficult and recovery rates low – 1-5%, for rare earths. If supply chains start to buckle, for whatever reason, importers could find themselves in an uncomfortable position: facing real, physical constraints on their AI ambitions.

    Map shows how 10 countries dominate the supply of finished AI products with heavy concentration in China, Taiwan and S Korea

    Attracting and sustaining critical AI talent

    Access to human capital should not be taken for granted, either. The US is currently the top destination for the world’s AI talent. People from all over the world, but particularly from India and China, have significantly contributed to the US AI labour force across various cohorts, from researchers to developers and entrepreneurs. The US is also a prime destination for international students pursuing AI majors, many of whom stay post-graduation.

    That has given a huge boost to America’s AI industry, where immigrants are twice as likely as US-born individuals to fill roles in AI companies across research, engineering and data science, according to our team’s analysis of US Census Bureau data. Many of these newcomers bring ideas and energy, as well as technical prowess. According to the National Foundation for American Policy, a non-profit, non-partisan group, 65% of AI start-ups in the US have at least one immigrant founder.1

    What’s more, demand for talent is only increasing. Most large companies in the US have hired for general AI roles in data science or machine learning over the past few years, and now many are seeking to fill more specialised positions. The team’s study of job postings from Lightcast, a labour-market analytics platform, suggests that demand for specialized sets of AI-linked skills – such as familiarity with CUDA, Nvidia’s programming interface – increased nearly five-fold between 2016 and 2024.

    However, continued US dominance is not assured. Plenty of other nations appear to be offering easier immigration pathways than the US, which maintains hard caps on specialty-occupation visas through its H-1B programme. Canada, for example, has introduced an open work programme specifically for US H-1B visa holders, aiming to attract skilled tech professionals. The United Arab Emirates offers a ten-year residency without a sponsor to people with an ‘outstanding specialised talent’ across numerous fields, including AI. Both countries ranked among the top for net AI talent migration in 2024, according to The AI Index.2

    Meanwhile, due to the rapid expansion of domestic AI sectors and increasing job opportunities, other countries – particularly China - are narrowing gaps, according to the team’s analysis of research papers accepted by the Conference on Neural Information Processing Systems (NeurIPS), the most prestigious global forum for AI and data science. In 2021, more than half (56%) of the top researchers who were invited to make an oral presentation at NeurIPS were based in the US, while just 5% were in China. By 2024, the US was still out front, with a 42% share. But China’s share had surged to 26%. Researchers from the UAE and Hong Kong, too, were beginning to make an impression.

    Unearthing potential

    For now, much is in flux. But for governments, what seems certain is that a resilient AI strategy will involve measures to ensure a stable supply of critical minerals and the steady development of talent, both imported and homegrown.

    Chart shows how the US retains highest number of AI researchers globally, however China’s share has surged between 2021-2024

    1 National Foundation for American Policy (NFAP)*Source: National Foundation for American Policy (NFAP)

    2 The AI Index 2025 Annual Report

     

    Important content disclosures

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

    About the analysts

    Zornitsa Todorova

    Zornitsa Todorova

    Head of Thematic FICC Research

    Zornitsa Todorova leads the Thematic FICC Research team at Barclays, based in London. She is a Director, and covers topics such as market structure, financial market innovation, and emerging macroeconomic trends. Zornitsa and her team employ quantitative, academic-style methods to analyse the global impact of these themes across credit, rates, and macro products. Before joining Barclays in 2019, she taught risk management and investments at Bocconi University and the Berlin School of Business and Innovation. Her research has been featured in peer-reviewed journals such as The Journal of Financial Markets, Institutions and Instruments, The Journal of Corporate Ownership and Control, and The Review of Finance and Banking. Zornitsa holds a PhD with honors in Economics and Finance from Bocconi University in Italy.

    Vincenzo Pota

    Vincenzo Pota

    Head of US Rates Research

    Vincenzo Pota is a Director on the Barclays Research Data Science team, based in London. He specialises in information extraction from big data and alternative data using cutting-edge data science techniques. Prior to entering finance, Vincenzo worked as a data scientist for IT consulting firms where he modelled customer behaviour using machine learning. He received a PhD in Astrophysics from the Swinburne University of Melbourne, and was a researcher at the University of California Santa Cruz.

    Maggie O'Neal

    Maggie O'Neal

    Global Head of Sustainable Investing Research

    Maggie O'Neal is a Managing Director and Global Head of Sustainable Investing Research. Prior to moving to ESG Research, Maggie covered Basic Industries/Industrials and was named in the top three of Institutional Investor European Fixed Income Research team for her coverage of Basic Materials and Industrials from 2016-2023. She co-developed Barclays' approach to ESG Research and assumed her current role in 2022. Maggie joined Barclays in 2015; she has 15 years of experience in research, having previously worked at Deutsche Bank. Maggie studied at UCLA, Sciences Po Paris and King's College London for undergraduate and holds an MSc from London School of Economics.

    Parsys 5
    iParsys for Double Pixel component