The AI Race Isn’t About Models—It’s About Power

The contest over AI is no longer about algorithms—it is about the industrial scaffolding that makes them possible. Chips, capital, compliance, and kilowatts now define national power.

Image: “Tesla Optimus Gen-2 humanoid robot” — image by Tesla, licensed under CC BY 3.0

Railways and telegraphs once redrew the world map, giving their owners both economic advantage and political leverage. Generative AI is doing the same today. The decisive factor is not who builds the most dazzling chatbot, but who commands the infrastructure behind intelligence itself: semiconductors, finance, regulation, and electricity. These have become the thin red lines of twenty-first century power.

Washington: From Chips to Capital

The United States has tightened export curbs on advanced semiconductors four times since 2022. As of January 2025, it also bars American firms from investing in Chinese ventures tied to AI, quantum, or chipmaking. Finance now joins technology on the national security list.

Allies are pressed into the strategy. The Netherlands and Japan restrict sales of lithography tools. Nvidia, symbol of the AI boom, ships only neutered processors to China under strict licenses. Access to technology is no longer a market transaction—it is a political decision.

The risk is overreach. Squeeze too hard, and Washington alienates allies or drives Beijing to double down on independence. Ease up, and U.S. leverage over the global AI supply chain weakens.

Europe: Regulation as Power

Europe lacks a chip champion but possesses regulatory muscle. The EU AI Act, phased in from 2025, requires transparency for general-purpose models and stricter obligations for high-risk applications by 2026.

Any firm seeking Europe’s 450 million consumers must comply. As with GDPR, Brussels exports its rules globally, shaping the field to suit compliance-ready European firms while imposing costs on foreign rivals.

Europe’s danger lies in credibility. Overreach could push companies to friendlier jurisdictions; delay would erode its claim to leadership. If Brussels balances ambition and restraint, it becomes the world’s arbiter of AI governance.

Beijing: Substitution and Circumvention

China is sprinting toward self-reliance. Huawei has released AI accelerators to challenge Nvidia. Domestic fabs inch forward in lithography despite U.S. sanctions. Beijing courts the Global South with “good enough” AI—cloud services, pre-trained models, turnkey systems at bargain prices.

This approach appeals to Africa, Southeast Asia, and Latin America, where governments want AI for growth but cannot afford Western tools. China offers an affordable alternative, creating dependencies abroad while reducing its own.

Progress is uneven. Chip design advances, but scaling at cutting-edge nodes remains elusive. China’s red line is forced decoupling. If U.S. restrictions tighten further, Beijing could strike back with rare-earth export bans or disruptions across supply chains.

The Gulf’s Neutral Ambition

The Middle East has become an unexpected theater. The UAE and Saudi Arabia are investing billions in hyperscale data centers, importing advanced U.S. chips, and striking deals with Nvidia, Oracle, and OpenAI. Their ambition: to act as neutral compute hubs linking Europe, Asia, and beyond.

It mirrors their oil diplomacy—capital, geography, and studied neutrality as leverage. Yet neutrality is precarious. Washington demands compliance; Beijing searches for alternatives. If AI infrastructure is openly weaponized, Gulf balancing acts may collapse.

Japan: Betting on Rapidus

Japan is chasing autonomy through Rapidus, a state-backed venture aiming to mass-produce 2-nanometer chips with IBM and U.S. partners by the late 2020s. Tokyo remembers its 1980s trade clash with Washington and its present reliance on foreign chips.

But advanced fabs demand more than subsidies. They require decades of expertise that even generous funding cannot conjure overnight. Japan’s gamble is clear: if Rapidus fails, dependency deepens; if it succeeds, Tokyo reclaims a central role in semiconductor power.

The Constraint of Electrons

The most decisive factor may not be chips but kilowatts. AI data centers consume staggering amounts of electricity, with demand set to more than double by 2030. In Virginia, utilities delay coal retirements to keep up. In Ireland, grid operators warn of overload. Nuclear energy and storage, once niche debates, are now central to AI strategy.

If compute is the new oil, electrons are the pipelines. Energy geopolitics—long defined by hydrocarbons—is shifting to grids. Climate goals clash with AI demand, forcing governments into hard choices between emissions targets and technological edge.

The grid is the ultimate bottleneck. Without reliable power, AI expansion stalls no matter how advanced the chips. Nations that solve the energy puzzle will hold the real leverage.

⸻

Our Take: The AI revolution underscores a simple fact: sovereignty is industrial. States no longer fight over borders but over supply chains. And unlike steel or oil, AI magnifies every domain it touches—military, economic, scientific, social. Control the scaffolding of intelligence, and you shape the future of human progress.

Four interlocking arenas—compute, capital, compliance, and kilowatts—now define global competition. None can be mastered in isolation. Restrictions on chips drive capital controls. Regulations steer investment flows. Energy constraints limit computational growth. No single nation can dominate all four, creating both opportunities for coalitions and risks of fragmentation.

History offers two models. The nuclear age produced division but stability. The internet age (at first) produced integration. Which path AI takes will depend on whether nations can compete without collapsing the system they all need.

For now, the thin red lines hold. But they fray with every sanction, every export curb, every tilt toward coercion. The winners will be those who secure sovereignty without manufacturing scarcity. The losers may discover that in trying to dominate intelligence, they have engineered something more dangerous: artificial scarcity of the one resource no country can afford to lose.