The MSI EdgeXpert MS-C931 is an ultra-compact desktop AI supercomputer designed for high-performance local computing. It features the NVIDIA® GB10 Grace Blackwell chip, combining CPU and GPU processing power within a unified architecture, alongside 128 GB of LPDDR5x memory and a 4 TB SSD.
With a footprint of just 1.2 litres, the MS-C931 delivers outstanding performance, capable of developing, fine-tuning, and deploying AI models locally with up to 200 billion parameters, or 405 billion when two units are linked together.
Product customisation
Don't forget to save your customization to be able to add to cartarrow_drop_downRécapitulatif
Total :
The MSI EdgeXpert MS-C931 represents a new generation of computing power dedicated to local artificial intelligence. This ultra-compact desktop AI supercomputer condenses datacenter-level performance into just 1.2 litres, offering researchers, developers, and businesses a self-contained, high-efficiency platform for training and deploying advanced AI models directly on-site.
At the heart of the MS-C931 lies the NVIDIA® GB10 Grace Blackwell processor — a cutting-edge architecture that combines a 20-core ARM Grace CPU with a Blackwell GPU linked via NVLink™ C2C, providing a data bandwidth far exceeding traditional PCIe 5.0.
This unified design allows the CPU and GPU to share a common memory space, eliminating bottlenecks and maximizing AI computing efficiency. The system features 128 GB of LPDDR5x unified memory and a high-speed 4 TB NVMe SSD, ensuring seamless performance across the most demanding workloads.
Thanks to its integrated design, the MSI EdgeXpert MS-C931 can train and execute AI models locally with up to 200 billion parameters, and up to 405 billion when paired with a second unit.
Its remarkable energy efficiency and silent cooling make it ideal for research environments, development labs, universities, and enterprises that require high-performance AI computing without relying on external cloud infrastructures.
The MS-C931 is built on the NVIDIA DGX™ ecosystem, running NVIDIA DGX OS and the full NVIDIA AI software stack (CUDA®, cuDNN™, TensorRT™, PyTorch®, TensorFlow®, and more).
This native compatibility streamlines complex AI workflows — from model training and inference to deep learning, image processing, predictive analytics, and intelligent automation.
With a comprehensive range of connectivity options — 10 GbE, USB-C, HDMI™ 2.1a, Wi-Fi 7, and Bluetooth® 5.4 — the MS-C931 integrates effortlessly into professional and academic infrastructures.
Designed for on-premises and sovereign AI computing, the EdgeXpert MS-C931 enables sensitive data processing without reliance on external cloud services. This approach preserves confidentiality, reduces technological dependency, and ensures full control of the computing environment.
The system operates reliably around the clock with optimized power consumption and long-term stability, making it suitable for mission-critical or data-sensitive applications.
Compact, silent, and robust, the MS-C931 reflects MSI’s expertise in thermal management and electronic reliability. Its high-density chassis, energy-efficient components, and network scalability (horizontal expansion) make it a strong foundation for next-generation AI infrastructures.
The MSI EdgeXpert MS-C931 embodies the convergence of extreme performance and accessibility. By bringing supercomputer-class computing power to the desktop, it enables advanced AI deployment directly at the data source.
This Edge Computing approach delivers faster, more efficient, and more sovereign digital ecosystems. With EdgeXpert, MSI paves the way for a local, scalable, and sustainable AI infrastructure serving innovation across science, industry, and technology.

|
Architecture |
NVIDIA® Grace Blackwell |
|
GPU |
NVIDIA® Blackwell Architecture |
|
CPU |
20-core Arm Processor (10 Cortex-X925 + 10 Cortex-A725) |
|
CUDA Cores |
Blackwell Generation |
|
Tensor Cores |
5th Generation |
|
Tensor Core Support |
Yes |
|
Supported Data Formats |
TF32, FP16, BF16, INT8, FP8, FP6, FP4 |
|
RT Cores |
4th Generation |
|
Tensor Performance |
1 petaFLOP AI performance (FP4, Sparse) |
|
System Memory |
128 GB LPDDR5x unified system memory (CPU/GPU shared) |
|
Storage |
4 TB NVMe M.2 |
|
USB Ports |
4 × USB 3.2 Type-C (up to 20 Gb/s) |
|
Wired Network |
1 × RJ-45 connector 10 GbE |
|
Network Interface Card |
ConnectX-7 Smart NIC |
|
Wi-Fi |
Wi-Fi 7 |
|
Bluetooth |
Bluetooth 5.4 |
|
Audio Output |
Multichannel audio output via HDMI™ |
|
Video Outputs |
1 × HDMI™ 2.1, 4 × DisplayPort 1.4a via USB-C |
|
System Weight |
1.2 kg |
|
System Dimensions |
151 mm (L) × 151 mm (W) × 52 mm (H) — Volume: 1.2 L |
|
Operating System |
NVIDIA® DGX™ OS |
Data sheet