Estonia Redefines AI Competitiveness: Shifting Focus from Tech Infrastructure to Governance

When we talk about national competitiveness in the age of AI, the first things that usually come to mind are GPUs, data centers and foundation models. How much computing power a country has secured, how large its AI data centers are, and whether it possesses competitive domestic AI models have become important criteria for judging national AI competitiveness. Without the physical infrastructure that powers AI, no country can become an AI leader. Yet after spending two days in Tallinn, Estonia, observing the front lines of digital government and AI policy, I found myself returning to one question: can national competitiveness in the AI era really be built on technological infrastructure alone?

What I encountered in meetings with officials from Estonia’s Information System Authority, or RIA, was another kind of infrastructure, one that is far less visible. Data governance, digital identity, inter-agency collaboration, cybersecurity, trust in government, and the digital and AI literacy of citizens and public officials. I would call this the social infrastructure of the AI era.

Digital government is not built overnight. When people discuss Estonia’s digital government, X-Road and electronic identity, or e-ID, are almost always mentioned. But when one listens to people on the ground, what matters is not only the technologies themselves, but the order and principles through which they were built. Estonia first established citizens’ digital identities and built a foundation through which data held by different public and private institutions could be exchanged securely. On top of that, it gradually added numerous digital public services.

What is important here is that Estonia did not gather government data into one giant centralized database. Each institution still manages its own data. Each institution remains responsible for that data. But when necessary, it can connect with data held by other institutions. In other words, the core of Estonia’s digital government is not centralization, but interoperability.

This difference matters more than it may first appear. When governments try to solve the problem of bureaucratic silos, they often imagine gathering data in one place or creating a powerful control tower. Estonia shows a different path. It preserves each institution’s responsibility and expertise while connecting them. And precisely because they are connected, each institution must take greater responsibility for the quality of its own data and the operation of its systems. If one institution’s data is wrong, services that rely on that data elsewhere can also malfunction.

Governance is harder than technology. Each ministry has its own legal authority and function. Each has its own information systems, data, budget and organization. Whenever data sharing or workflow integration is attempted, the same question appears: “If something goes wrong, who is responsible?”

Korea has also made many policy attempts to strengthen government data sharing and inter-ministerial cooperation. But being technically able to connect systems and actually having organizations share data and work together are entirely different things. That is why one of the most memorable remarks from this visit was also one of the simplest: “It is mainly about governance.”

The problem is not technology. The core issue is governance.

The technology already exists. The question is under what rules it should be connected, who should take responsibility, and how institutions should be made to cooperate. X-Road alone did not solve every problem for Estonia. The responsibilities of each ministry and agency, the role of RIA in managing shared national digital infrastructure, political coordination, cooperation with private companies, and a relatively horizontal administrative culture all work together.

RIA, in particular, is not simply an IT support agency. It is responsible for the interoperability of the state information system and cybersecurity, creates shared foundations that government institutions can use, reduces duplicated investment and helps set the direction for new digital technologies. Estonia’s RIA describes its role as coordinating the development and administration of information systems to ensure the interoperability of the state information system, organizing information-security activities, and handling security incidents in Estonian computer networks.

Another striking theme during this visit was education. In other words, even in the AI era, education is again at the center. When people speak of Estonia’s digital success, they often think of X-Road and e-ID. But locally, the significance of Tiger Leap, which began in the 1990s, is still regarded as crucial. Its purpose was not merely to distribute computers and internet access to schools. It also built the capacity of teachers and students to use digital technology. More than a decade later, Estonia’s startup ecosystem began to grow in earnest.

Education Estonia explains that Tiger Leap was built around computers and the internet, basic teacher training and native-language electronic courseware for general education institutions. In other words, it was not only a hardware project, but also a capability-building project.

Now Estonia is trying a similar approach again with AI.

The program is AI Leap. In the 2025–2026 pilot, around 20,000 students and 4,800 teachers took part. What is interesting is that Estonia did not simply give AI to students first. It began with teacher training. The program addressed not only how to use AI, but also educational psychology, assessment methods, self-directed learning and critical thinking. The goal was not simply to bring AI into schools. It was to ask how education itself should change in an era where AI exists. AI Leap is designed to expand in 2026–2027 to vocational education and new student cohorts, and then to integrate validated approaches into the education system and curriculum.

This approach delivers an important message.

AI infrastructure does not mean only GPUs. People are infrastructure too.

Through this visit to Estonia, I came to believe that national AI competitiveness should be divided into two kinds of infrastructure. One is visible AI infrastructure. This includes GPUs, data centers, cloud infrastructure, electricity, data and foundation models. The other is invisible AI infrastructure. This includes data governance, digital identity, inter-agency collaboration, cybersecurity, institutional trust, education and AI literacy, and accountability structures.

The former makes AI possible.

The latter allows AI to actually function within society and the state system.

Estonia is not the country with the world’s largest number of GPUs. Nor is it a country developing the world’s most powerful foundation models on its own. Yet it can rapidly experiment with the next stage of AI entering government and social systems because it has accumulated this invisible infrastructure over more than two decades.

And this is where the question for Korea begins.

Are we focusing too much on visible infrastructure when we talk about AI competitiveness? Securing GPUs and data centers and developing strong models are, of course, important. But once AI begins to enter healthcare, education, public administration, welfare and industrial sites, entirely different questions appear.

Who will access the data?

How will data be connected across institutions?

How much authority should be given to AI?

If a wrong judgment occurs, who is responsible?

Will citizens trust the system?

And are citizens prepared to use AI properly?

The answers to these questions cannot be found in the number of GPUs alone.

As the AI transformation accelerates, the focus of national competition is likely to shift from securing technology to the governance capacity needed to use that technology. Estonia’s digital government in Tallinn showed precisely that. And Estonia is already asking the next question: if AI moves beyond answering citizens and begins acting on their behalf, how must the state change?

At the center of that question is Estonia’s emerging vision of the agentic state.

Next article: “From Chatbot to Agentic State: Estonia’s Bürokratt Experiment”