Slovenia unveils first academic AI-focused supercomputer in region

The University of Ljubljana has unveiled Frida, a new high-performance computing infrastructure designed to support artificial intelligence research, scientific computing, and industry collaboration.

Located on the rooftop of the university's Faculty of Computer and Information Science, Frida is the first academic AI-specialised supercomputing facility of its kind in the region.

It will be accessible to researchers, students and companies, CE Report quotes The Slovenia Times.

According to the faculty, Frida delivers twice the AI training performance and energy efficiency of Slovenia's existing supercomputer, Vega, and represents a significant technological leap forward in the training and application of AI models.

Companion to Vega

Rather than replacing Vega, the new system is intended to complement it. "Frida specialises in calculations that are crucial for artificial intelligence, while Vega remains highly effective for high-precision scientific calculations," Iztok Lebar Bajec, associate professor at the faculty and the lead architect of the Frida project, explained at the inauguration ceremony on 16 June.

With a computing capacity of 1.42 exaflops and 16.8 terabytes of GPU memory, Frida combines seven generations of graphics processors, providing flexibility for a wide range of research applications.

University officials described the project as a major step forward for Slovenia's research ecosystem. Faculty Dean Mojca Ciglarič said Frida expanded opportunities for international collaboration and positioned the faculty among leading global academic and technology institutions.

Data sovereignty

A key feature of the facility is its focus on data sovereignty. By enabling advanced AI models to run locally, organisations can process sensitive data without relying on foreign or commercial cloud providers, improving both security and control over information, researcher Rebeka Kropivšek Leskovar said.

The system also incorporates advanced direct-source cooling technology, reducing energy consumption by cooling hardware components directly rather than entire server rooms.

Frida is expected to support projects ranging from Slovenian-language AI models and medical image analysis to climate modelling, industrial optimisation, speech recognition, and real-time translation technologies.

Professor Marko Šikonja pointed out that the infrastructure will allow the continued development of Slovenia's domestic language model, Gams, within the country rather than relying on foreign supercomputing resources. It will also "enable Slovenian companies and public institutions to develop and test AI solutions in a local environment".

Slovenian government and university representatives described Frida as a strategic addition to the country's research infrastructure, strengthening its capabilities in high-performance computing and AI while creating new opportunities for collaboration between academia and industry.

Photo: Pexels (Free Stock Photos)