České Radiokomunikace (CRA) successfully launched a new element of its artificial intelligence infrastructure at its data center in Lužice during the weekend of April 25-26. This is a high-performance AI server HGX B300 equipped with eight NVIDIA B300 graphics processors, which is intended for direct production deployment at the end customer.
The entire procurement and installation process took an exceptionally short time. The hardware was delivered from the USA in just ten days, allowing the operator to immediately respond to specific client requirements without unnecessary time delays. According to the company, the goal was to quickly follow up on previous project phases and ensure technical readiness for computationally intensive operations.
From a technical perspective, the installation represented a demanding process, primarily due to the implementation of virtualization in the Proxmox environment. This configuration required a specific approach from the technical team, as in many aspects it involved deploying technologies in a less explored operating mode. The server is now fully configured and ready for live operation within the portfolio of services provided by CRA.
“We view the deployment of this server as progress in building an AI infrastructure that allows companies to work with computationally intensive models in a local environment, and as a significant step towards the development of CRA's AI services,” representatives of the company stated in an announcement on LinkedIn.
The technical and commercial teams of the company participated in the project. The technical part, including installation and configuration, was handled by Marek Antolik, Róbert Drotár, Jaroslav Pavlovec, and David Prokop. The commercial aspect of the project and customer communication were overseen by Ondřej Krumphansl, Petr Možiš, and Jakub Smrček.
The data center in Lužice is thus becoming a key hub for providing performance for the needs of generative AI and machine learning in the Czech Republic. Expanding local infrastructure is intended to facilitate companies' access to modern technologies without having to rely exclusively on foreign cloud providers, which brings advantages especially in terms of latency and data sovereignty.