The MGX server is a modular design for remote virtualization and supercomputing at the edge. The reference design enables quicker deployment and supports CPUs from NVIDIA, AMD, and Intel.
AI Platform, High-Performance Computing Server, High Converged Server, AI Training Server, AI Inference Server
X Series
ARM Server
ARM Server
2U
NVIDIA Grace
1
1000 W
NVIDIA-Certified Systems™ - NVIDIA ARM CPU Systems
--
--
NVIDIA Superchip
MGX Server FAQ
Why can the MGX platform handle so many different workloads simultaneously?
Because MGX is modular. The same platform can combine powerful CPUs, multiple GPUs, high-speed GPU-to-GPU communication, fast networking, and DPUs. This flexibility allows customers to run AI training, inference, simulation, rendering, and visualization on a single scalable architecture rather than separate systems.