EXANTE, the global prime broker, has released a new research report arguing that investors are treating artificial intelligence purely as a technology play, while overlooking the energy, minerals and water on which the boom depends.
The analysis, titled ‘Beneath the gold rush: The uneven economy of AI‘, looks at how AI spending is changing the global economy across three areas: labour, sovereignty and natural resources.
It was written by Renée Friedman, PhD, EXANTE’s global head of research, and draws on studies from international institutions, banks, consultancies and universities. Its aim is to map where AI’s costs and gains are landing, and to identify risks not yet reflected in corporate disclosures or asset prices.
According to Goldman Sachs projections cited in the report, worldwide AI-related investment is expected to reach $1tn in 2026, with $581bn of that in the US. The report argues that whether this spending pays off depends largely on factors beyond the technology industry, including electricity grids, critical minerals, water supplies, regulation, and how well workers and institutions adapt.
On climate, the report finds that AI’s environmental costs are emerging ahead of its benefits. Allianz estimates that data-centre emissions could more than double to 643 MtCO₂ by 2030 if grids are not decarbonised. The IEA suggests AI could reduce emissions by 1.4 Gt annually by 2035, but most of the relevant applications have not yet been deployed at scale. Around half of the footprint by 2030 may come from emissions built into servers and semiconductors, which current ESG reporting struggles to capture, meaning portfolios may feel the risk before the data shows it.
The report also says influence over AI is gathering around physical bottlenecks. Figures from the Peterson Institute for International Economics show raw US AI compute capacity expanded by more than 200% in each of 2024 and 2025. Since no country can dominate every layer of the AI stack, power flows to those controlling key points such as semiconductors, energy grids and around 40 critical minerals. Export controls, mineral supply and grid capacity are therefore described as policy levers as well as market forces, and the cost to companies of operating under stricter regimes across competing blocs is seldom disclosed or priced.
On employment, the report suggests markets are responding to fears that the evidence does not yet back up. A King’s College London study found 57% of people expect AI to cause widespread unemployment. However, Stanford’s Institute for Economic Policy Research found unemployment among the fifth of workers most exposed to AI rose 0.77 percentage points since 2022, against 0.85 points for the least exposed. Strain is concentrated at entry level, where PwC found roles needing senior, human-intensive skills have grown 35% since 2019, while other entry-level positions have fallen 10%. The report recommends treating the gap between perceived and actual disruption as a separate variable, since consumer sentiment follows perception while corporate earnings follow reality.
The report also questions assumed productivity gains. OECD modelling indicates AI could lift annual labour productivity growth by 0.4 to 1.3 percentage points, but this relies on reskilling, organisational change and adaptable institutions, areas where ageing societies often fall short. The report’s benchmark is whether improvements appear in national accounts and in wage growth beyond the technology sector; until then, it views AI spending as simply a cost.
In healthcare, the report argues the bottleneck is now evidence rather than capability. Close to one in three EU citizens will be 65 or older by 2050, and EU countries were already short an estimated 1.2 million doctors, nurses and midwives in 2022. A US Food and Drug Administration discussion paper from August 2026 points to a longer path to clinical use for generative AI tools than earlier medical software faced, extending revenue timelines though not research ones.
EXANTE global head of research Renée Friedman said, “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.”
The full report is available on request.
Copyright © 2026 FinTech Global
