Jensen Huang has mentioned one term in multiple public speeches: Sovereign AI.

When most people hear the word “sovereignty,” what comes to mind is territory, military, diplomacy. But in today’s world where AI is deeply embedded in every nation’s economy, defense, and public services, the definition of sovereignty is being rewritten.

If a nation cannot autonomously develop, deploy, and control AI systems, its sovereignty is incomplete—regardless of how strong its military or how large its territory.

Four Layers of Autonomy

Sovereign AI is not a single concept—it has four layers, each indispensable.

Technological sovereignty. Can you create core technologies yourself—chips, high-performance computing, AI frameworks? If your AI systems run on another country’s chips, you’re paralyzed the moment they cut off supply. Taiwan’s semiconductor industry has unique advantages at this layer, but this advantage isn’t eternal—it requires continuous investment to maintain.

Data sovereignty. Where is your data stored? Who can access it? If a nation’s medical data, financial data, demographic data are all stored in foreign clouds, that’s not digitization—that’s digital colonization. Data sovereignty means controlling data storage and processing rights, protecting sensitive information from external access.

Algorithmic sovereignty. AI models are not neutral. They reflect the biases and values embedded in their training data. If a nation’s public AI systems (education, healthcare, justice) all use models trained by foreign entities, the values embedded in those models may be completely misaligned with domestic societal needs. Algorithmic sovereignty means ensuring AI models are transparent, controllable, and aligned with national interests and social values.

Application and service sovereignty. Controlling the AI infrastructure of critical public services. If your traffic management systems, power grid control systems, defense simulation systems depend on foreign platforms, that’s not cooperation—that’s dependency.

These four layers together constitute complete Sovereign AI. Missing any layer leaves sovereignty vulnerable.

From Oil to Data

In the last century, the foundation of national power was oil and minerals. Whoever controlled energy controlled the world order.

In this century, data and computing power are replacing oil’s position.

In “Negentropy Strategy: The Survival Logic of Taiwanese Enterprises,” I discussed how Taiwanese enterprises need to establish order within chaos to survive. The same applies at the national level—establishing one’s own technological autonomous order within the chaos of AI geopolitics is a prerequisite for survival.

The difference is: oil’s geographic location is fixed—you either have it or you don’t. Data and computing power can be constructed—as long as you have the will and capability to invest. For countries like Taiwan that lack oil but possess technological capabilities, this is actually an opportunity.

But opportunity doesn’t equal guarantee. If Taiwan doesn’t proactively build its own AI infrastructure, this window of opportunity will be seized by others.

Technological Autonomy vs. Techno-Authoritarianism

Sovereign AI development has a dangerous dark side.

When nations pursue AI development under the banner of “technological autonomy,” how do we ensure these AI systems aren’t used to surveil citizens, suppress dissent, or manipulate public opinion? China’s social credit system is one example—it’s technically an implementation of “Sovereign AI,” but ethically it’s a demonstration of techno-authoritarianism.

This is the dilemma every nation pursuing Sovereign AI must face: you need sufficient technological control to protect national security, but without democratic institutional checks and balances, this control can easily slide toward abuse.

In “Safer-4 and the Future of Technology Governance,” I discussed how the core issue of AI governance isn’t technology, but the distribution and balance of power. The same applies to Sovereign AI—technological autonomy is necessary, but it must be embedded within democratic, transparent, and accountable institutional frameworks to avoid becoming another form of oppression.

Taiwan’s Position

Taiwan occupies a unique position on the Sovereign AI chessboard.

Our semiconductor manufacturing capabilities give us inherent advantages in the first layer of technological sovereignty. But in the other three layers—data sovereignty, algorithmic sovereignty, application and service sovereignty—our investment is far from adequate.

Our public AI applications heavily depend on foreign platforms. Our data governance framework is still immature. Our AI talent continues to flow overseas.

Taiwan doesn’t lack technological capability—what we lack is strategic-level integrative thinking to extend our semiconductor advantages across the complete AI stack, from chips to models to applications to data governance, constructing a complete Sovereign AI system.

This isn’t something one company can do, nor something one ministry can accomplish. It requires national-level strategic will.

A Civilizational Choice

Sovereign AI is not just technological competition—it’s a civilizational choice.

Do you choose to depend on great powers’ AI ecosystems, or build your own? Do you choose technological control to protect security, or democratic checks and balances to prevent abuse? Do you choose to entrust data to the most efficient platforms, or insist on storing it where you can maintain control?

Every choice has costs. But not choosing—letting things develop naturally—carries the highest cost. Because the direction of “natural development” always tilts toward the concentration of power and capital.

Establishing true autonomy in the digital age requires not just technological investment, but a nation’s clear answer to “what do we want to become.”