AI Factory in CEE: Beyond.pl Infrastructure | M Computers
We delivered a key piece of infrastructure based on NVIDIA technologies for the Polish data center provider Beyond.pl, which created the first commercial sovereign AI Factory in the CEE region. This is one of the most significant AI projects in CEE to date, providing a powerful, secure and European regulation-compliant environment for the development and operation of next-generation AI.
For us, this meant working with the latest NVIDIA technologies, extreme time pressure and having to deal with the entire chain of continuity from component availability to technical readiness of the data centre.
Beyond.pl: a visionary pioneer
Beyond.pl is a long-standing leader in the field of secure and highly available infrastructure in Poland. The company focuses on the operation of critical IT systems and provides solutions that meet the most stringent requirements for availability, security and data sovereignty. With the launch of AI Factory, Beyond.pl is responding to the growing demand for a powerful AI computing environment that is available directly within the European Union.
Infrastructure for critical data and AI of the future
The foundation of the AI Factory is a cutting-edge infrastructure built on NVIDIA technologies, designed to train and run large-scale AI models. The delivered solution provides high computational power, scalability and efficient data handling while maintaining full control over data placement and security.
The solution is based on a reference architecture NVIDIA DGX SuperPOD™ which includes NVIDIA infrastructure and software. Computing runs on NVIDIA B200 GPUs with Blackwell architecture, interconnected via a high-speed network NVIDIA Quantum-2 InfiniBand, and connected via NVIDIA Spectrum Ethernet to Pure Storage FlashBlade scalable object storage. It was this technological complexity that presented one of the biggest challenges of the entire project.
This powerful configuration supports a wide range of tasks, from training large language models (LLM) and generative AI to scientific research, enterprise and industrial use cases, or AI startups. A software stack is installed to accelerate deployment NVIDIA AI Enterprise that optimises and accelerates the development and deployment of pre-trained AI models, including a suite of easy-to-use microservices, tools, libraries and frameworks.
Where the real work began: the role of M Computers
M Computers participated in the supply of key technological components.
“The Beyond.pl project was unique for us mainly because of its scope and the fact that it worked with the latest NVIDIA technologies that were just coming to the market. At the same time, everything had to be delivered in record time. We were dealing with the complex availability of individual components that are technologically dependent on each other – without one part, the customer could not install another,” explains Kamila Jeřábková, Product Manager at M Computers.
“We were working with very demanding scaling, timing and compatibility requirements for the entire NVIDIA platform. It was an important experience for us that has guided how we approach large AI projects today, not only technically but also in terms of partnerships.”
What we took away from the project
The Beyond.pl project was a valuable experience for us in implementing a large-scale AI infrastructure in real operation. It showed us how crucial it is to work not only with the technologies themselves, but also with the overall context of their deployment.
By communicating closely with the customer, we gained a deeper understanding of how performance and scalability requirements translate into the preparation of the environment, the planning of individual steps and the coordination of all parties involved.
Today, we use this experience to design other AI solutions where we emphasize long-term uptime, infrastructure readiness and realistic planning of the entire project from the first steps.
AI Factory and its impact on the CEE region
The establishment of AI Factory Beyond.pl represents an important milestone for the entire CEE region. It brings organizations the opportunity to use cutting-edge AI infrastructure without having to move sensitive data outside the EU or rely solely on global hyperscalers. The project thus supports the development of innovation, research and industrial applications of AI while maintaining high standards of security and regulation.
Technical Specifications
| PARAMETER | NVIDIA DGX B200 |
|---|---|
| GPU | 8× NVIDIA B200 |
| GPU memory | 8x 180GB HBM3e (1440 GB total) |
| CPU | up to 2x Intel Xeon Platinum 8570 2.1 GHz (56 cores) |
| FP8/INT8 Tensor Performance | 72 PetaFLOPS |
| # CUDA cores | 163,840 |
| # Tensor cores | 5,120 |
| RAM | up to 4 TB DDR5 |
| Storage | OS: 2× 1.92 TB NVMe M.2 data: 8x 3.84 TB NVMe U.2 |
| Network | 4x OSFP for 8x single-port NVIDIA ConnectX-7 VPI (400Gb/s InfiniBand/Ethernet) 2x dual-port QSFP112 NVIDIA BlueField-3 DPU (400Gb/s InfiniBand/Ethernet) |
| Max. power consumption | ~ 14.3 kW |
| Form factor | 10U rack |
NVIDIA AI Enterprise
Included with virtually every NVIDIA device, whether it’s a compact AI workstation or a supercomputer at the scale of an entire data center, is the extensive but intuitive NVIDIA AI Enterprise software suite. With the various features and settings that this software brings, you’ll have a perfectly optimized environment on your device for machine learning, data analytics, or working with AI models as they are trained, inferenced, and ultimately tested and deployed. Libraries and tools such as Cafe 2, Theano, TensorFlow, PyTorch or MXNet are also ready for these purposes, and more are no problem to get from the portal NVIDIA GPU Cloud (NGC). Thanks to careful tuning and optimization, such an environment provides up to 30% higher performance for Machine Learning applications compared to those deployed on NVIDIA hardware alone, according to NVIDIA.
