AI landmark in the Baltics: Lithuania launched one of CEE’s most powerful AI platforms

May 21, 2025
Lithuania has launched one of the most powerful AI platforms in Central and Eastern Europe, built on the advanced NVIDIA DGX B200 server. Valued at €1 million, the system has been deployed at the Digital Defence Centre of Excellence at VILNIUS TECH University.
 
The new platform accelerates research and innovation in artificial intelligence, digital defence, and data science. According to university researchers, the infrastructure will help bridge the gap with leading global research centres and enable Lithuanian teams to surpass them in select areas.
 
In addition to academic use, the platform will be accessible to businesses, providing a high-performance environment for testing, developing, and deploying advanced AI solutions.
 
The new artificial intelligence (AI) computing platform enables researchers to work tens or even hundreds of times faster than before and to move their ideas from theory to real-world solutions more quickly. Shorter experimentation, simulation and analysis times on large datasets will allow faster prototyping and publication of results.
 
The platform will be available to the University's scientists, researchers, PhD students, staff, and external partners such as public institutions and businesses. The computing and AI platform is envisaged as a sandbox for collaboration between the university and the private sector, especially those involved in cybersecurity and defence projects.
 
The University, therefore, invites businesses to develop solutions jointly, use the testbed, and conduct joint research. Companies with valuable data and domain expertise without the infrastructure for AI experiments are welcome.
 
The latest DGX server model delivered
 
The platform was designed, delivered and installed for the university by 3RTechnology, the only technology company in Lithuania with NVIDIA Priority Partner status. According to the company's director, Rėdas Šimelis, the project included not only the latest air-cooled DGX server model available today, but also other components such as CPU and GPU servers, data storage, network switches, cooling and power supply systems. The total cost of all the equipment is more than €1 million.
 
"We designed and prepared the infrastructure to work according to the needs of the university, so that the platform is versatile and can provide different types of resources according to the researcher's tasks during the use of the platform. The NVIDIA DGX B200 server itself weighs more than 130 kg, and more than 180 kg with packing materials, so one of the challenges was transporting it and delivering it safely to the data centre premises", says Šimelis. The implementation of this project shows that Lithuania is keeping pace with global technology trends. 
 
Ambitious plans include an unparalleled solution
 
Professor Dr Artūras Serackis, Head of the Department of Electronic Systems at VILNIUS TECH, says the newly launched platform delivers capabilities previously beyond reach for university researchers. Tasks that previously required an entire month to train a single AI model can now be completed in under half a day.  Moreover, the platform unlocks the ability to develop and train AI models that were previously too large or complex to process.
 
“This platform enables us to study large language models at an entirely new scale,” says Prof. Serackis. “We can run models with significantly more parameters and specialise them for specific domains, resulting in highly accurate language models capable of delivering comprehensive responses with near-infallible precision.”
 
Serackis says this opens new business opportunities: “They can now evaluate whether a large language model can be tailored to meet their unique needs, estimate the infrastructure required, and make informed decisions—potentially avoiding costly and unnecessary upfront investments.”
 
Now the platform is ready for use, and according to project lead Prof. Dalius Matuzevičius, it is not easy to decide where to start with the research, as the range of ideas and projects is vast. "For the first time, we can launch and effectively train specialised AI models for generating 3D objects from a single photo. This is particularly important for our ongoing project to develop a unique solution for 3D modelling spectacle frames. We hope the new infrastructure will soon enable a significant technological breakthrough," says the VILNIUS TECH professor.

Galerija

Related news

New doctoral dissertation
New doctoral dissertation
VILNIUS TECH Library invites you to follow the published new dissertations. The dissertation „Analysis and modeling of deformations induced by the extrusion of fiber-reinforced polymers“ prepared at VILNIUS TECH by Mahmoud Samy Mahmoud Mohammed Farh. The dissertation was prepared in 2021–2026. Scientific consultant – Prof. Dr Viktor Gribniak. The dissertation was defended at the public meeting of the Dissertation Defense Council of the Scientific Field of Materials Engineering in the Aula Doctoralis Meeting Hall of Vilnius Gediminas Technical University at 2 p.m. on 10 June 2026. Additive manufacturing via fused filament fabrication (FFF) enables the creation of geometrically complex components. Yet, its use in structural and semi‑structural applications remains limited by anisotropic mechanical response, defect sensitivity, and fabrication‑induced residual stresses that cause warpage and geometric inaccuracy. This dissertation investigates polylactic acid (PLA)- based materials manufactured by FFF, including neat and partially recycled PLA, continuously reinforced PLA, and short-fiber-reinforced composites, to develop an integrated experimental-computational methodology for evaluating mechanical efficiency and predicting process-induced distortion. The research object comprises the mechanical, thermal, viscoelastic, microstructural, and thermo‑mechanical characteristics of these materials. The dissertation develops a unified approach linking reinforcement strategy, material structure, thermal history, mechanical performance, and warpage behavior. The adopted methodology combines quasi‑static tensile and flexural testing, thermomechanical characterization, scanning electron microscopy, and finite-element simulations. Continuous aramid reinforcement developed in this study for FFF increases the load-bearing capacity of the tension specimens by 67%. Still, reinforcement efficiency was limited by toolpath continuity, interfacial defects, and the absence of in‑process fiber tensioning. Short‑fiber-reinforced composites exhibit distinct fiber‑type‑dependent behavior: carbon‑filled PLA increases stiffness, while wood‑filled PLA enhances crystallinity, stiffness retention near the glass‑transition temperature, toughness, and dimensional fidelity. Wood‑fiber reinforcement reduces edge warpage by 43% and carbon fiber by 14.3% under identical conditions. A staged thermo‑mechanical simulation framework is developed to model printing, cooling, and detachment, transferring residual stress and distortion fields into subsequent mechanical simulations. The ABAQUS model for neat PLA predicts warpage with an average error of 8.2–10.6%, whereas a Digimat workflow captures the deformation in short‑fiber-reinforced PLA with an error of 14.3–17.9%. The latter predictions were obtained for the first time. The dissertation consists of an introduction, three main chapters, general conclusions, and references. The First Chapter provides a literature review of FFF of reinforced polymers, including material combination and modeling strategies. The Second Chapter specifies the chosen materials, test program, and thermo mechanical modeling concept. The Third Chapter evaluates experimental and numerical results, integrating mechanical, thermal, microstructural, and simulation based findings. The General Conclusions summarize the dissertation work, which is supported by four publications, including three articles in Web of Science indexed journals with impact factors, and four conference presentations. Doctoral dissertation readers can search via VILNIUS TECH Virtual Library.    
More
New doctoral dissertation
New doctoral dissertation
VILNIUS TECH Library invites you to follow the published new dissertations. The dissertation „Interaction between currency market evolution with monetary policy instruments in the age of digitisation“ („Valiutų rinkos evoliucijos sąveika su monetarinės politikos instrumentais skaitmenizacijos amžiuje“) prepared at VILNIUS TECH by Tomas Pečiuli. The dissertation was prepared in 2020–2026. Scientific consultant – Assoc. Prof. Dr Asta Vasiliauskaitė. The dissertation was defended at the public meeting of the Dissertation Defence Council of the Scientific Field of Economics in the Aula Doctoralis Meeting Hall of Vilnius Gediminas Technical University at 10 a.m. on 10 June 2026. The emergence of decentralised cryptocurrencies has created fundamental challenges for traditional monetary policy systems. Although these digital assets have the potential to increase financial inclusion and efficiency, their volatility and the lack of centralised oversight create systemic risks that cannot be properly managed using classical models. This dissertation presents an integrated hybrid analytical framework designed to quantitatively assess the impact of cryptocurrencies on monetary policy transmission mechanisms, providing policymakers with empirically grounded tools to analyse this evolving financial domain more effectively. The dissertation is divided into three main parts. The First Chapter summarises the theoretical role of cryptocurrencies in modern monetary theory. The Second Chapter presents and substantiates a new methodology that combines machine-learning techniques with advanced econometric modelling, specifically using an Elastic Net machine learning model with ARIMA residuals and MSGARCH specifications to capture regime-dependent behaviour. The Third Chapter empirically validates the framework using data from cryptocurrency markets and central bank policy operations. The empirical results show a significant asymmetric policy transmission effect, with the price of Bitcoin reacting by USD -15,348 to a 1% change in the Federal Reserve interest rate. The analysis also identifies critical volatility thresholds (σ>80%) at which cryptocurrency fluctuations increase inflation risk. These results indicate the growing systemic importance of cryptocurrencies in monetary policy dynamics. The study contributes to the emerging field of digital asset economics. The integrated modelling approach helps overcome the long-standing limitations of analysing nonlinear financial phenomena. Practical applications include real-time financial stability risk monitoring systems and evidence-based guidelines for regulatory interventions. The modular structure of the framework allows for future expansion by incorporating evolving market structures and new digital assets. The dissertation’s results have been presented to the scientific community in eight peer-reviewed publications in scientific journals and conference proceedings. This work provides central banks with essential analytical tools to maintain monetary stability and to promote responsible financial innovation in the digital era. Doctoral dissertation readers can search via VILNIUS TECH Virtual Library.
More