Who really wins the AI race? The answer lies in electricity and metals, not algorithms

As the AI boom accelerates, investment is shifting from software companies toward power grids, data centers, mines, and chip manufacturing. A University of Tartu economics student argues that the true AI race is a geopolitical competition for physical resources, raising urgent questions about whether surging energy demand can even be met.

The popular image of the artificial intelligence race — brilliant programmers competing to write the cleverest code — may be fundamentally misleading. According to Heili-Mae Möller, an economics student at the University of Tartu, the real competition is happening underground, in server halls, and along high-voltage transmission lines, not in software labs.

Capital Flows Are Shifting

Just a few years ago, the AI investment frenzy meant pouring money into technology firms and their valuations. That pattern is changing. An increasing share of capital is now flowing toward physical infrastructure: electricity grids, data centers, mining operations, and semiconductor fabrication facilities. The shift reflects a hard reality — no matter how sophisticated an algorithm is, it cannot run without vast quantities of power, rare materials, and custom-built chips.

What is emerging, Möller argues, is a geopolitical race for physical resources. Countries and corporations are racing to secure the raw materials and energy capacity that will determine who can actually scale AI systems. Copper for wiring, lithium and cobalt for batteries, silicon for chips — these unglamorous commodities are quietly becoming strategic assets of the first order.

Can Energy Demand Be Met?

The central question hanging over the entire AI industry is whether the exponentially growing appetite for electricity can realistically be satisfied. Data centers already consume enormous amounts of power, and the rollout of large language models and AI inference systems is accelerating that demand dramatically. Electricity grid operators in Europe and North America are warning that existing infrastructure was simply not built for this scale of consumption.

Möller's analysis suggests that investors and policymakers who focus solely on the algorithmic dimension of AI are missing the deeper story. The winners of the AI era may not be the companies that write the best models, but those that control the mines, the megawatts, and the manufacturing capacity needed to bring those models to life. For a small, digitally advanced country like Estonia, understanding this shift in the global resource competition carries direct strategic and economic implications.