AI investors are overlooking the power, minerals and water behind the $1 trillion boom, EXANTE report warns

London, United Kingdom, 15 October 2026 — EXANTE, the global prime broker, today published Beneath the gold rush: The uneven economy of AI, an analysis of how AI investment is reshaping the global economy across three frontiers: labour, sovereignty and natural resources.

Authored by Renée Friedman, PhD, Global Head of Research at EXANTE, the report brings together research from global institutions, banks, consultancies and universities. From this evidence it builds a single view of where the costs and benefits of AI are falling, and which risks are not yet visible in company disclosures or market prices.

The report’s central argument is that AI is too often assessed as a technology allocation alone. Goldman Sachs forecasts $1 trillion in AI-related investment globally in 2026, including $581 billion in the US. The returns on that spending depend on factors that sit largely outside the technology sector: the energy grids, critical minerals and water supplies that data centres rely on, the regulation that governs them, and the workers and institutions that must adapt to them.

The report’s key conclusions include:

• The climate costs of AI are arriving before the benefits. Allianz estimates that, without grid decarbonisation, data-centre emissions could more than double to 643 MtCO₂ by 2030. The IEA estimates that AI could cut 1.4 Gt of emissions a year by 2035, but most of the applications behind that figure do not yet exist at scale. By 2030, roughly half of the footprint could sit in emissions embedded in servers and semiconductors. These are the emissions least visible to current ESG reporting, so the risk is likely to reach portfolios before it shows up in the data.
• Power over AI is concentrating at physical chokepoints. Peterson Institute for International Economics (PIIE) estimates show raw US AI compute capacity grew by more than 200% in both 2024 and 2025. No state can control every layer of the AI stack. Influence accrues instead to those who are indispensable at specific points: semiconductors, energy grids and the roughly 40 critical minerals beneath every model. The report argues that export controls, mineral supply and grid capacity are now policy variables as much as market ones. For companies operating across rival blocs, the cost of serving markets under stricter regimes is rarely disclosed and rarely priced.
• Markets are reacting to fear of AI job losses that the data does not yet support. In a King’s College London study, 57% of the public expect widespread unemployment from AI. Yet analysis by Stanford University’s Institute for Economic Policy Research (SIEPR) found that unemployment among the most AI-exposed fifth of workers rose by 0.77 percentage points since 2022, compared with 0.85 points among the least exposed. Pressure is concentrated at entry level. PwC found that entry-level roles requiring senior, human-intensive skills have grown by 35% since 2019, while other entry-level roles have shrunk by 10%. The report argues that the gap between perceived and measured disruption should be treated as a variable in its own right, because consumer confidence responds to the first and corporate earnings to the second.
• The productivity dividend is a claim to be tested, not a base case. OECD modelling suggests AI could add 0.4 to 1.3 percentage points to annual labour productivity growth. That range assumes reskilling, organisational change and adaptive institutions, and ageing societies tend to be weaker on all three. The report’s test is whether gains appear in national accounts and in wage growth outside the technology sector. Until they do, AI spending on its own is a cost line.
• In healthcare, the constraint has shifted from capability to evidence. By 2050, nearly one in three EU citizens will be aged 65 or older. EU countries already faced an estimated shortage of 1.2 million doctors, nurses and midwives in 2022. The US Food and Drug Administration (FDA) discussion paper of August 2026 signals a longer route into clinical use for generative AI tools than earlier medical software faced. That lengthens timelines for revenue, though not for research.

“The AI boom is being priced as a software story, but it runs on copper, power and water. Those bills are arriving now. Most of the promised benefits, from climate savings to productivity gains, have yet to show up at scale. That gap is the story markets are missing”, commented Renée Friedman, PhD, Global Head of Research, EXANTE.

The full report is available on request. 

About Dr Renée Friedman, PhD

 Dr Renée Friedman is Global Head of Research at EXANTE, specialising in macroeconomic trends, geopolitical risk and the intersection of technology and financial markets. She is available for interviews across all formats.

About EXANTE

Media contact: Julia Chapman, jch@exante.eu 

Disclaimer and risk warning: The information contained herein is provided for informational purposes only and should not be regarded as an offer or solicitation of an offer to buy or sell any investments or related services that may be referenced here. Investing involves a high level of risk. Past performance is not a reliable indicator of future results.