AI/GPU servers

We deliver an ecosystem of services and solutions supporting the development of AI projects.

Our multi-level offering consists of: GPUaaS, AIaaS based on NVIDIA AI Enterprise, GPU colocation services, and managed services.

Accelerate AI development with a package of secure and sovereign services from Beyond.pl – from access to GPU infrastructure to the AI platform.

DGX B200 Technical Specification

Parameter Value
GPU 8× NVIDIA Blackwell
GPU Memory 1.4 TB (1,440 GB) HBM3e, 64 TB/s
Performance (training) 72 petaFLOPS FP8
Performance (inference) 144 petaFLOPS FP4
CPU 2× Intel Xeon Platinum 8570 (112 cores, up to 4.0 GHz)
RAM 2 TB (configurable up to 4 TB)
NVLink Throughput 14.4 TB/s
Storage Pure Storage FlashBlade//S500 – certified for DGX B200 SuperPOD
Network Nvidia Quantum 2 – Infiniband
NVIDIA Spectrum Ethernet
8× ConnectX-7, 2× BlueField-3 DPU (up to 400 Gb/s)
Software NVIDIA AI Enterprise / NVIDIA NIM
Mission Control
DGX OS / Ubuntu

Available articles