Summary in Plain Language
The “Sovereign AI Index 2.0” released by the renowned American think tank New America Security Center provides a comprehensive overview of the new AI competition underway around the world. In the past three years, the number of “sovereign AI” projects – those that are completely independent and not susceptible to foreign supply disruptions – has skyrocketed from one to 184, spanning 67 countries and the European Union. This is no longer a game exclusive to wealthy countries in Europe and America. The once-unshakable dominance of American large-scale AI models in global AI technology has begun to crumble, as Chinese technology companies have entered the official sovereign AI projects of six countries. This trend reflects a global concern that in the AI era, countries must not become dependent on others. The country whose AI infrastructure is adopted by more nations will gain more influence over future global technological norms. China now holds a unique position: it is both building its own AI ecosystem from scratch and becoming one of the few providers of such infrastructure to other countries, presenting both significant opportunities and challenges.
---
Detailed Analysis
1. Growth from 1 project to 184 in 3 years: Sovereign AI is no longer just a luxury for wealthy nations
Many believed that sovereign AI was a high-end concept reserved for tech giants like China, the US, and Europe. However, data shows the opposite: in 2023, there was only one sovereign AI project in action globally, but by 2026, this number had risen to 184, including countries with low and middle incomes such as Cambodia, Nepal, and Ghana. The logic behind this is similar to the drive for national power grids and communication networks decades ago. If a country relies entirely on foreign AI for government documents, hospital records, and traffic management, it could be severely impacted if the service is interrupted. Now, one-third of the world’s countries are pursuing sovereign AI, accounting for 90% of global GDP. This has shifted from a discretionary choice to a necessity.
2. The US’s dominance in AI infrastructure is weakening
Previously, most countries used American models as the foundation for their sovereign AI, with little alternative. However, this situation is no longer sustainable. Even American think tanks acknowledge that it’s no longer taken for granted that all countries should rely on US AI. Chinese models, which were virtually unknown by 2025, have quickly made it into the official projects of six countries. This is due to their cost-effectiveness and reliability, as they avoid expensive patent and service fees and the risk of being cut off for geopolitical reasons.
3. Countries prefer simpler models for security reasons
Forty percent of sovereign AI projects now adopt a radical approach, developing their own models from scratch without using any foreign models. For example, Switzerland’s domestically developed AI, although less powerful than Meta’s open-source models, performs better for translating confidential documents because all data remains on Swiss servers, eliminating security concerns. Countries prioritize security over performance, even if it means using less advanced models.
4. Sovereign AI is no longer just a government initiative; it’s a critical infrastructure
Sovereign AI has spread beyond government initiatives to various industries. For instance, South Korea uses it for ship design and production, Taiwan for finance, and Ukraine for military operations. In these critical sectors, countries cannot afford to risk their data on foreign servers. This shift means that the competition is about control over core data, which will determine future industry standards and practices.
5. China’s dual role as both player and provider
China is both a developer and supplier of sovereign AI infrastructure, with its technologies being adopted by six countries. This position is advantageous as it allows China to set global norms. However, it’s more than just selling models; China must adapt them to local languages and regulations and build long-term trust with clients. Even the US is adjusting its strategy, seeking to be a reliable partner rather than forcing all allies to use its AI. China is in a prime position to shape the future of global AI.
In summary, the global trend towards sovereign AI is a shift from reliance on foreign technologies to self-sufficiency and control over critical data, with significant implications for economic and strategic landscapes.