NEW

MSI EdgeXpert MS-C931 NVIDIA Grace Blackwell GB10 Desktop AI Supercomputer Mini PC – New | EDGEXPERT-34SAU

In Stock
1 units available
SKU
EDGEXPERT-34SAU-NEW
MPN
EDGEXPERT-34SAU
A$7,829
The MSI EdgeXpert MS-C931 is a desktop AI supercomputer built on the NVIDIA DGX Spark platform and powered by the GB10 Grace Blackwell Superchip. It pairs a 20-core Arm CPU with a Blackwell-architecture GPU to deliver up to 1,000 AI TOPS (1 petaFLOP FP4). With 128GB of LPDDR5x coherent unified memory and a 4TB PCIe Gen5 NVMe SSD, it lets developers prototype, fine-tune and run inference on AI models with up to 200 billion parameters at the desk. Integrated NVIDIA ConnectX networking lets two units be clustered to reach models of up to 405 billion parameters, all in a compact mini-PC chassis running NVIDIA DGX OS.
In Stock
Platform NVIDIA DGX Spark
Superchip NVIDIA GB10 Grace Blackwell
CPU 20-core Arm (10x Cortex-X925 + 10x Cortex-A725)
GPU NVIDIA Blackwell, 6,144 CUDA cores
AI Performance Up to 1,000 AI TOPS (1 petaFLOP FP4)
Interconnect NVLink-C2C

Request a quote or add to cart — pricing available on enquiry for refurbished stock.

For international customers, please contact us for personalised support and assistance.

180-Days Warranty
Secure Checkout
Free Delivery AU
Dispatch in 1–2 Business Days
MSI · EdgeXpert MS-C931
NVIDIA Grace Blackwell GB10 Desktop AI Supercomputer New
Data-centre-class AI on the desktop — 1 petaFLOP, built on NVIDIA DGX Spark.
Overview
The MSI EdgeXpert MS-C931 brings the NVIDIA DGX Spark platform to the desktop, driven by the GB10 Grace Blackwell Superchip that unites a 20-core Arm CPU and a Blackwell-architecture GPU over 128GB of coherent LPDDR5x unified memory. Purpose-built for AI developers and researchers, it delivers up to 1 petaFLOP of FP4 AI performance to prototype, fine-tune and run inference on large models locally — and two units can be clustered over NVIDIA ConnectX networking to tackle models of up to 405 billion parameters.
Key Features
  • NVIDIA GB10 Superchip: Grace Blackwell architecture uniting CPU and GPU with NVLink-C2C coherent memory (up to 5x PCIe 5.0 bandwidth).
  • AI Performance: Up to 1,000 AI TOPS (1 petaFLOP FP4) for demanding generative and agentic workloads.
  • 20-Core Arm CPU + Blackwell GPU: 10x Cortex-X925 plus 10x Cortex-A725, with a 6,144-CUDA-core GPU.
  • 128GB Unified Memory: LPDDR5x coherent memory shared by CPU and GPU at 273 GB/s — run models up to 200B parameters.
  • 4TB PCIe Gen5 Storage: Self-encrypting NVMe M.2 SSD with sequential reads up to 10,000 MB/s.
  • ConnectX Networking: Cluster two EdgeXpert units to scale to models of up to 405B parameters.
  • Compact Mini-PC Design: Runs NVIDIA DGX OS out of the box and is backed by a 3-year warranty.
Specifications
General
Device Type Desktop AI supercomputer (mini PC)
Platform NVIDIA DGX Spark
Superchip NVIDIA GB10 Grace Blackwell
Operating System NVIDIA DGX OS (Linux)
Compute
CPU 20-core Arm (10x Cortex-X925 + 10x Cortex-A725)
GPU NVIDIA Blackwell, 6,144 CUDA cores with 5th-gen Tensor & 4th-gen RT Cores
AI Performance Up to 1,000 AI TOPS (1 petaFLOP FP4)
Interconnect NVLink-C2C (up to 5x PCIe 5.0 bandwidth)
Memory & Storage
Memory 128 GB LPDDR5x coherent unified memory
Memory Bandwidth Up to 273 GB/s
Storage 4 TB M.2 NVMe PCIe Gen5 SSD (self-encrypting, up to 10,000 MB/s)
Networking
Wired Networking 10 GbE (RJ-45)
SmartNIC NVIDIA ConnectX
Wireless Wi-Fi 7, Bluetooth 5.3
I/O
USB 4x USB 3.2 / USB4 Type-C
Power & Support
Power 240 W
Warranty 3 years

Product information is automatically compiled from trusted manufacturers and data providers. Please. report any errors.

Write Your Own Review
You're reviewing:MSI EdgeXpert MS-C931 NVIDIA Grace Blackwell GB10 Desktop AI Supercomputer Mini PC – New | EDGEXPERT-34SAU
Your Rating

Common questions about the MSI EdgeXpert MS-C931 NVIDIA Grace Blackwell GB10 Desktop AI Supercomputer Mini PC – New | EDGEXPERT-34SAU.

It is a desktop AI supercomputer (mini PC) built on the NVIDIA DGX Spark platform, powered by the NVIDIA GB10 Grace Blackwell Superchip. It brings data-centre-class AI compute — up to 1 petaFLOP (1,000 AI TOPS) — to the desktop for developers and researchers.
The GB10 Grace Blackwell Superchip combines a 20-core Arm CPU (10x Cortex-X925 + 10x Cortex-A725) with an NVIDIA Blackwell-architecture GPU featuring 6,144 CUDA cores, plus next-generation Tensor and RT Cores, linked by NVLink-C2C.
Up to 1,000 AI TOPS (1 petaFLOP) of FP4 AI performance — enough to prototype, fine-tune and run inference on AI models with up to 200 billion parameters on a single unit.
Yes. Using NVIDIA ConnectX networking, two EdgeXpert MS-C931 units can be linked together to run models of up to 405 billion parameters, such as Llama 3.1 405B.
128GB of unified, coherent LPDDR5x memory shared between CPU and GPU (up to 273 GB/s), plus a 4TB M.2 NVMe PCIe Gen5 SSD with self-encryption and sequential reads up to 10,000 MB/s.
The MS-C931 provides 4x USB 3.2 / USB4 Type-C ports, a 10GbE RJ-45 connector and NVIDIA ConnectX high-speed networking, along with Wi-Fi 7 and Bluetooth 5.3.
It runs NVIDIA DGX OS (Linux) with the full NVIDIA AI software stack, and is backed by a 3-year warranty.