MSI
MSI EdgeXpert NVIDIA DGX GB10 AI Supercomputer 128GB 4TB
The MSI EdgeXpert NVIDIA DGX GB10 AI Supercomputer is a desktop AI platform powered by the NVIDIA Grace Blackwell Superchip, built for developers and researchers who need to run large language models and AI workloads locally instead of relying on cloud infrastructure. It delivers 1 petaFLOP of AI performance in a compact form factor, combining an Arm 20-core CPU with NVIDIA Blackwell GPU architecture to handle model development, fine-tuning, and inference at scale.
Why this product
1 petaFLOP AI performance in a desktop
Delivers 1000 AI TOPS (FP4) tensor throughput via NVIDIA Grace Blackwell, enabling real-time processing of complex AI models without cloud dependency.
128 GB unified system memory
LPDDR5x coherent memory shared between CPU and GPU via NVLink-C2C interconnect, five times the bandwidth of PCIe 5.0 for fast data throughput during model operations.
Scales to 512 GB total VRAM
Stack up to four units via ConnectX-7 networking to handle models reaching 405 billion parameters, with switchless ring topology for 3-node or star topology for 4-node clusters.
4 TB encrypted NVMe storage
Self-encrypting M.2 drive for storing datasets, model checkpoints, and inference data with integrated security.
Complete hardware and software stack
Ships with NVIDIA DGX OS and all software components for agents, multimodal inference, data science, and edge AI development—no separate software licensing required.
- AI Tensor Performance1 petaFLOP (FP4)
- System Memory128 GB LPDDR5x unified
- Storage4 TB NVMe.M2 encrypted
- ProcessorNVIDIA Grace Blackwell (GB10) + Arm 20-core
- Dimensions151 × 151 × 52 mm
- Weight1.2 kg
- Power240W PSU (GB10 TDP 140W)
- SKUMSI-7381
- EAN4711377403511
- Runs NVIDIA DGX OS with full support for CUDA, NVIDIA TensorRT, and standard AI frameworks (PyTorch, TensorFlow).
- Connects via 1× RJ-45 10 GbE Ethernet, 4× USB 3.2 Type-C, HDMI 2.1a, and WiFi 7 / Bluetooth 5.3.
- Scales via ConnectX-7 Smart NIC and NVIDIA NVLink for multi-unit clustering without requiring external Ethernet switches for 3-node setups.
- 4-node clustering requires a 200 GbE QSFP Ethernet switch; 2 and 3-node clusters use direct QSFP DAC cables.
- Not compatible with standard x86 or x64 software—requires ARM-native or containerised applications designed for NVIDIA DGX platforms.
Best for
- Large language model inference and fine-tuning
- Multimodal AI research and development
- Edge AI deployment without cloud latency
- Data science and model experimentation locally
- Organisations scaling from single-unit to multi-unit AI clusters
Trade-offs
- Existing x86 software stacks — runs Arm-native only; legacy applications built for x64 or x86 will not run without recompilation or containerisation.
- Graphics-intensive or gaming workloads — optimised exclusively for AI tensor compute; no consumer GPU drivers or gaming support.
- Teams without GPU development expertise — full-stack AI platform requires knowledge of CUDA, DGX OS, and containerised deployment patterns.
- Budget-constrained projects — enterprise-grade AI supercomputer positioned at the high end of desktop AI performance.
MSI EdgeXpert is a desktop AI supercomputer built on the NVIDIA DGX Spark platform, powered by the NVIDIA GB10 Grace Blackwell Superchip. It delivers up to 1,000 AI TOPS (FP4) of performance in a compact 151 x 151 x 52 mm chassis weighing 1.2 kg, letting developers run large AI models locally instead of in the cloud.
EdgeXpert is designed from the ground up to build and run AI. Powered by the NVIDIA GB10 Grace Blackwell Superchip, it is a complete local AI platform for agents, multimodal inference, fine-tuning, data science, and edge development, bringing the power of NVIDIA Grace Blackwell to every desk. Its compact design facilitates seamless integration and deployment, enabling powerful AI performance for innovators who demand excellence.
- NVIDIA® Grace Blackwell (GB10) Architecture: NVIDIA Blackwell GPU & Arm 20-core CPU, NVIDIA® NVLink®-C2C CPU-GPU memory interconnect
- 64 GB or 128 GB LPDDR5x coherent, unified system memory
- 1000 AI TOPS (FP4) AI performance
- Full stack solution, hardware & software, designed for AI developers
- ConnectX-7 networking: Scales out from 2 to 4 EdgeXpert systems, expanding total VRAM up to 128 GB~512 GB — 3-node cluster: ring topology via 3 QSFP DAC cables (switchless); 4-node cluster: star topology via a 200 GbE QSFP Ethernet switch
The MSI EdgeXpert AI Supercomputer transforms the landscape of desktop AI performance, offering petaflop-scale power via the next-generation NVIDIA® GB10 Grace Blackwell Superchip—the same engine driving the NVIDIA DGX™.
- NVIDIA® Grace Blackwell Architecture — NVIDIA Blackwell GPU & Arm 20 core CPU streamlines data preprocessing and orchestration to speed up model optimisation and deliver real-time performance.
- NVLink®-C2C Technology — Delivers a unified CPU+GPU memory architecture with up to five times the bandwidth of PCIe 5.0, enabling lightning-fast data throughput.
- Unmatched AI Computing Power — 1000 AI TOPS (FP4) Tensor Performance provides exceptional speed to seamlessly handle demanding AI workloads at scale.
- 128 GB LPDDR5x, unified system memory — Offers expansive memory capacity for seamless model development, fast experimentation, and efficient inference performance.
- Stack via NVIDIA ConnectX — High-speed NVIDIA ConnectX networking links two MSI systems to handle AI workloads with models reaching up to 405 billion parameters.
Processor & Memory
| Brand | MSI |
|---|---|
| CPU | 20-core Arm (10 Cortex-X925 + 10 Cortex-A725) |
| GPU Architecture | NVIDIA Blackwell Generation |
| Tensor Cores | 5th Generation |
| RT Cores | 4th Generation |
| System Memory | 128 GB LPDDR5x, unified system memory |
| Memory Interface | 256-bit |
| Memory Bandwidth | 273 GB/s |
| CPU-GPU Interconnect | NVIDIA NVLink-C2C |
Performance & Storage
| AI Tensor Performance | 1 petaFLOP AI Performance (FP4, Sparse) |
|---|---|
| Storage | 4 TB NVMe.M2 with self-encryption |
| Video Encoding / Decoding | 1× NVENC | 1× NVDEC |
Connectivity
| Ethernet | 1× RJ-45 10 GbE |
|---|---|
| Smart NIC | ConnectX-7 |
| USB | 4× USB 3.2 Type-C |
| Display | 1× HDMI 2.1a |
| Wi-Fi | WiFi 7 |
| Bluetooth | BT 5.3 |
| Audio Output | HDMI multichannel audio output |
Physical & Power
| Dimensions | 151 × 151 × 52 mm |
|---|---|
| Weight | 1.2 kg |
| Power Supply | 240W |
| GB10 TDP | 140W |
| Operating System | NVIDIA DGX OS |
Shipping
All our products are shipped from the United Kingdom within 24 to 48 hours of purchase. See our Shipping and Returns page for full details.
Returns
- Tech Syndicate offers a full money-back guarantee on all purchases made in our store. If you are not satisfied with your purchase, we will provide a full refund, no questions asked.
- Refunds are available within 45 days of delivery.
- After the 45-day period you are no longer eligible for a refund, so we encourage you to try the product within the first few weeks to make sure it fits your needs.