How can a national AI strategy be turned into a visible change in people's lives? A lecture by experts from government, business, and academia suggests that the key lies in concrete projects, realistic infrastructure, education, and continuity. Slovakia has what it takes—if it stops waiting for the "perfect" solution and acts pragmatically.
From ambitions to pilots: infrastructure and sovereignty
Strategic documents alone are not enough—the first concrete projects and confident decisions will make the difference. According to estimates, the Slovak market today needs roughly 10 MW of compute power, which is why a private AI "factory" of this scale is being built with the ambition to launch operations at the beginning of the year. A smaller but quickly implemented solution could even be among the first in Europe, while the big gigafactories grapple with bureaucracy and financing.
The question of sovereignty is not only about where the data reside, but also about the law that applies to them. Experts pointed out that with American hyperscalers the Cloud Act applies, so it makes sense to rely on European or local infrastructure and a hybrid cloud. At the same time, the state does not want to start with fines under the AI Act, but to systematically educate the market and help with implementation over the coming years, and only then introduce strict oversight.
Education and language: AI literacy from primary school
Since last September, the Ministry of Education has been revising key documents so that AI literacy will be taught in primary schools from this school year and in secondary schools from the next. In April, a competency framework for teachers was created, and they are undergoing intensive training in cooperation with the nonprofit and commercial sectors. From September, two AI competence centers are to operate in Bratislava and Košice, each with support of 6 million euros.
Universities (STU and TUKE) are launching an international study program with support from NVIDIA, which helped with the curriculum and trained 20 instructors. In the debate on Slovak in AI, they agreed on a pragmatic path: not to develop an "own" model for enormous sums, but to use high-quality open-source models and provide them with top-tier Slovak training data. The state can also collaborate on such a data package—and there is room as well for minority languages, where the problem is even more pressing.
Technology in practice: from hybrid cloud to quantum support
Rising hardware prices and rapid technological development force us to think about return on investment: AI pays off where it delivers a measurable impact. Experts recommend choosing 20–30 key public services and deploying AI where it shortens waiting times and removes routine work (for example, in application review). Healthcare is promising but difficult—experience shows that without alignment with local legislation and data, solutions are hard to scale.
In infrastructure, a hybrid approach scores best: European/local data centers, high containerization, and the ability to scale capacity across ministries. Quantum technologies do not appear to be a "competitor" but a partner to AI—they can help with programming and training some models with lower data requirements. Finally, the most important rule applies: continuity and collaboration. Even with the best strategy, what matters is whether the first pilots are launched quickly, evaluated, and improved—without an endless return to square one.
