MSI EdgeXpert MS-C931 Personal AI Supercomputer NVIDIA GB10 - 128GB / 4TB
PRODUCT DESCRIPTION
MSI EdgeXpert MS-C931 Personal AI Supercomputer, NVIDIA GB10 Grace Blackwell Superchip, 20-Core Arm CPU, 128GB Unified Memory, 4TB Self-Encrypting NVMe SSD, ConnectX-7 200GbE, 10GbE, Wi-Fi 7 (9S6-C9311-61S) - 3 Year Local Warranty
MSI EdgeXpert MS-C931 4TB — Petaflop GB10 AI Supercomputer in a 1.2-Litre, 1.2 kg Desktop
The MSI EdgeXpert MS-C931 (part number 9S6-C9311-61S) is MSI's build of the NVIDIA DGX Spark platform: a personal AI supercomputer powered by the NVIDIA GB10 Grace Blackwell Superchip, pairing a 20-core Arm CPU with an integrated NVIDIA Blackwell GPU and 128 GB of coherent unified LPDDR5x memory. MSI rates it at 1 petaFLOP of FP4 AI performance with sparsity (1,000 AI TOPS), enough to develop, fine-tune and serve large language models of up to 200 billion parameters locally, with two units stacked over NVIDIA ConnectX-7 reaching 405 billion parameters. This configuration carries a 4 TB M.2 NVMe SSD with self-encryption, the largest drive MSI offers on the 128 GB model.
For Singapore developers, data scientists, research labs, universities and enterprise AI teams, the EdgeXpert puts DGX-class tooling on the desk with no cloud GPU bill and no data leaving the premises. MSI's distinguishing features are thermal: a patented FanDock airflow design, a smart fan and a copper-nickel plated SSD vapor chamber that MSI says keep the system above 200 W of sustained power without downclocking. SourceIT, an authorised IT reseller in Singapore, supplies it with a full GST invoice, 3-year local warranty and enterprise procurement support.
NVIDIA GB10 Grace Blackwell Superchip — 20 Arm Cores, Blackwell GPU, 5th-Generation Tensor Cores
The GB10 Superchip combines a 20-core Arm CPU — ten Cortex-X925 performance cores and ten Cortex-A725 efficiency cores — with an NVIDIA Blackwell architecture GPU on one package, joined by NVLink-C2C with, per MSI, up to five times the bandwidth of PCIe 5.0, so CPU and GPU address one coherent memory space instead of copying data across a bus. The GPU carries Blackwell-generation CUDA cores, fifth-generation Tensor Cores supporting TF32, FP16, BF16, INT8, FP8, FP6 and FP4, fourth-generation RT Cores for rendering and simulation, and one NVENC encoder and one NVDEC decoder for vision and media pipelines. The 1 PFLOP figure is a theoretical FP4 peak with sparsity, the same NVIDIA quotes for the DGX Spark platform.
128 GB Unified Memory at 273 GB/s — 200B-Parameter Models Locally, 405B Stacked
Discrete workstation GPUs with 16–32 GB of VRAM stop well before a 70B-parameter model loads; the EdgeXpert's 128 GB of LPDDR5x sits on a 256-bit interface at 273 GB/s and is shared by CPU and GPU as one pool, so Llama, Qwen, DeepSeek, Gemma, Mistral, gpt-oss and Nemotron-class models of up to 200 billion parameters run in full on one unit, and NVIDIA's platform figures add fine-tuning up to 70 billion parameters. MSI's scale-out guidance runs from two to four EdgeXpert systems, expanding total unified memory to as much as 512 GB; two stacked units handle models up to 405 billion parameters, and NVIDIA puts four linked GB10 systems at 700 billion. The memory is onboard and fixed at 128 GB.
4 TB Self-Encrypting NVMe SSD, ConnectX-7 200 GbE, 10GbE RJ-45, Wi-Fi 7
Storage on 9S6-C9311-61S is a 4 TB M.2 M-key NVMe SSD with self-encryption; MSI's US specification sheet for the 4 TB EdgeXpert lists it as a PCIe Gen5 drive. Networking is what sets a GB10 system apart from an ordinary mini PC: an NVIDIA ConnectX-7 SmartNIC at 200 Gbps exposes two QSFP ports for stacking a second EdgeXpert and moving datasets at line rate, while a separate 10GbE RJ-45 port serves the office or lab network and is the right port for pulling model weights. Wi-Fi 7 with two antennas and Bluetooth cover wireless; MSI notes that Wi-Fi 7 band availability depends on local regulatory approval.
Four USB 3.2 Type-C 20 Gbps Ports, HDMI 2.1, DisplayPort 1.4a over USB-C, Four Displays
The rear panel carries four USB 3.2 Type-C ports at 20 Gbps with DisplayPort 1.4a Alt Mode, plus one HDMI 2.1 port with multichannel audio output; MSI rates the system for four independent displays. One Type-C port doubles as the power input from the 240 W USB-C adapter, and MSI's user guide lists a Kensington lock slot for securing the unit on a shared desk. A TPM controller is fitted for platform security.
FanDock Cooling, SSD Vapor Chamber, 240 W USB-C Power, 151 × 151 × 52 mm
MSI's thermal design is the EdgeXpert's signature. A patented FanDock airflow path and smart fan are claimed by MSI to sustain over 200 W of system power without downclocking, yielding 10% faster token throughput and running up to 15 °C cooler under heavy load, with rear chassis temperatures reduced by up to 15 °C and top-surface heat by 9.1 °C; a copper-nickel plated vapor chamber lowers SSD temperatures by 9 °C, per MSI. Power comes from an external 240 W, 48 V USB PD adapter on 100–240 V AC. The chassis measures 151 × 151 × 52 mm (1.2 litres) and weighs 1.2 kg, and MSI rates it for 0–35 °C operation at 10–90% non-condensing humidity, with FCC Class B, CE, UKCA, VCCI and BSMI certification.
NVIDIA DGX OS, NemoClaw and the NVIDIA AI Software Stack
The EdgeXpert ships with NVIDIA DGX OS, an Ubuntu-based arm64 Linux build with the NVIDIA AI software stack preinstalled — CUDA, TensorRT, NVIDIA NIM and Blueprints, PyTorch, Jupyter and Ollama — and MSI names the NVIDIA Isaac, Metropolis and Holoscan frameworks for robotics, vision and edge work. As a GB10 system it runs the platform software NVIDIA publishes for DGX Spark: NemoClaw, NVIDIA's open-source reference stack that wraps OpenClaw agents with OpenShell policy-based privacy and security guardrails; NVIDIA AI Enterprise for DGX Spark; and the DGX Spark Playbooks. Because it shares the DGX software foundation, MSI states that models scale from the desktop to NVIDIA DGX Cloud or any accelerated data centre.
Scaling — Two, Three or Four EdgeXpert Units
MSI publishes three cluster layouts over the ConnectX-7 QSFP ports: two units linked directly with a single QSFP direct-attach cable; a three-node ring using three QSFP DAC cables with no switch; and a four-node star through a 200 GbE QSFP Ethernet switch. MSI's own cable is the optional 400G QSFP112 direct-attach cable, part 956-C931-101, 400 mm long with 112G-per-lane PAM4 signalling, sold separately; the NVIDIA QSFP112 400G DAC cable, 40 cm (X0101G00400A) that SourceIT stocks is the equivalent cable used across the DGX Spark family. The 10GbE RJ-45 port stays free for site networking throughout.
Available in Singapore Through SourceIT
SourceIT is an authorised IT reseller in Singapore and supplies the MSI EdgeXpert MS-C931 4TB (9S6-C9311-61S) with a full GST invoice, 3-year local warranty and enterprise, education and research-grant procurement assistance. For trillion-parameter work on the same MSI platform family, see the MSI XpertStation WS300 (NVIDIA DGX Station GB300). Browse every GB10 and GB300 system we stock in the NVIDIA DGX Spark & DGX Station collection and the wider AI workstations and AI servers collection.
MSI EdgeXpert MS-C931 4TB (9S6-C9311-61S) Datasheet — Technical Specifications
General
| Product Name | MSI EdgeXpert MS-C931 Personal AI Supercomputer — 4TB |
| SKU / Part Number | 9S6-C9311-61S |
| Category | Personal AI supercomputer / AI developer desktop (NVIDIA DGX Spark GB10 platform) |
| Form Factor | 1.2-litre mini desktop, 151 × 151 mm footprint |
| Colour / Finish | Not published by MSI |
| Manufacturer Warranty | MSI's US specification sheet lists 1 year; this listing is supplied with a 3 Year Local Warranty |
| Local Warranty (Singapore) | 3 Year Local Warranty — supplied by SourceIT with GST invoice and local support |
Processor & GPU — NVIDIA GB10 Grace Blackwell Superchip
| Architecture | NVIDIA Grace Blackwell — CPU + GPU on one superchip, NVLink-C2C interconnect (up to 5× PCIe 5.0 bandwidth, per MSI) |
| CPU | 20-core Arm: 10 × Cortex-X925 performance cores + 10 × Cortex-A725 efficiency cores |
| GPU | NVIDIA GB10 Blackwell architecture GPU (integrated), Blackwell-generation CUDA cores |
| Tensor Cores | 5th generation |
| RT Cores | 4th generation |
| Media Engines | 1 × NVENC, 1 × NVDEC |
AI Performance
| AI Performance | 1 petaFLOP (1,000 AI TOPS) FP4 with sparsity |
| Precision Support | TF32, FP16, BF16, INT8, FP8, FP6, FP4 |
| Model Support (single unit) | AI models up to 200 billion parameters, per MSI; fine-tuning up to 70 billion parameters per NVIDIA GB10 platform figures |
| Model Support (stacked) | Two units: models up to 405 billion parameters, per MSI; up to four GB10 systems: models up to 700 billion parameters, per NVIDIA |
Memory
| System Memory | 128 GB LPDDR5x unified system memory, shared by CPU and GPU |
| Memory Interface / Bandwidth | 256-bit, 273 GB/s |
| Upgradeability | Onboard — not user-upgradeable |
Storage
| SSD | 4 TB M.2 M-key NVMe SSD with self-encryption |
| Interface | PCIe Gen5 NVMe, per MSI's US specification sheet for the 4 TB EdgeXpert |
| Platform Range | 1 TB, 2 TB or 4 TB NVMe M.2 with self-encryption on the 128 GB model, per MSI |
Networking
| High-Speed NIC | NVIDIA ConnectX-7 SmartNIC, 200 Gbps; 2 × QSFP ports |
| Ethernet | 1 × 10GbE RJ-45 |
| Wi-Fi | Wi-Fi 7, 2 antennas (band availability subject to local regulatory approval, per MSI) |
| Bluetooth | MSI's product page specification table lists Bluetooth 5.4; MSI's US specification sheet lists Bluetooth 5.3 |
| Clustering Cable | Sold separately; MSI optional 400G QSFP112 DAC cable, 400 mm, part 956-C931-101 |
I/O & Display
| USB | 4 × USB 3.2 Type-C, 20 Gbps, DisplayPort 1.4a Alt Mode; one port is the USB-C power input |
| Video | 1 × HDMI 2.1; DisplayPort 1.4a over USB-C; up to 4 independent displays |
| Audio | HDMI multichannel audio output |
| Front Panel | Power button |
| Security | Kensington lock slot (per MSI user guide); TPM controller |
Software
| Operating System | NVIDIA DGX OS (Ubuntu Linux based), arm64 |
| AI Stack (preinstalled) | NVIDIA AI software stack — CUDA, TensorRT, NVIDIA NIM and Blueprints, PyTorch, Jupyter, Ollama; NVIDIA Isaac, Metropolis and Holoscan frameworks named by MSI |
| GB10 Platform Software | NVIDIA NemoClaw with OpenShell, NVIDIA AI Enterprise for DGX Spark, DGX Spark Playbooks (per NVIDIA) |
Cooling, Power, Physical & Certifications
| Cooling | MSI patented FanDock airflow design, smart fan, copper-nickel plated SSD vapor chamber; sustains over 200 W system power without downclocking, per MSI |
| Power Adapter | 240 W external USB-C PD adapter, 48 V / 5 A output; input 100–240 V AC, 50/60 Hz |
| Superchip TDP | 140 W (CPU + GPU), per NVIDIA GB10 platform specification |
| Dimensions (W × D × H) | 151 × 151 × 52 mm (1.2 L) |
| Weight | 1.2 kg (2.65 lb) |
| Operating Environment | 0–35 °C, 10–90% non-condensing; storage 0–60 °C |
| Acoustics | Not published by MSI |
| Certifications | EMC: FCC Class B, CE, UKCA, VCCI, BSMI; Safety: BSMI |
| In the Box | EdgeXpert MS-C931 unit, USB PD adapter, power cord, quick start guide |
Compatibility & Software Support
What Runs on the MSI EdgeXpert
Anything in the NVIDIA and open-source AI ecosystem published for arm64: Hugging Face Transformers and Diffusers, vLLM, Ollama, llama.cpp, LangChain and LlamaIndex RAG pipelines, NVIDIA NIM and Blueprints, NeMo, TensorRT-LLM, ComfyUI, RAPIDS, and the NVIDIA Isaac, Metropolis and Holoscan frameworks. NVIDIA's DGX Spark Playbooks apply directly because the EdgeXpert is the same GB10 platform.
Arm64 Linux — Plan Your Software Accordingly
The GB10's CPU is Arm-based and the EdgeXpert ships with NVIDIA DGX OS, so the operating system, drivers and container images are arm64 (aarch64) builds. Every major AI framework and NVIDIA NGC container is published for arm64, but x86-only proprietary desktop software will not run natively and Windows is not supported. Most teams drive the EdgeXpert from a laptop over SSH, Jupyter or VS Code Remote and keep Windows applications there. For a Windows 11 Arm desktop in the same collection, see the ASUS Ascent QN10; for a Windows x86 large-memory alternative, see the GMKtec EVO-X5 Pro.
Stacking Two, Three or Four EdgeXpert Units
Two EdgeXpert units connect back-to-back over their ConnectX-7 QSFP ports with one QSFP112 direct-attach cable — MSI's optional 400G cable 956-C931-101 or the NVIDIA QSFP112 400G DAC cable, 40 cm (X0101G00400A) SourceIT stocks — pooling 256 GB of unified memory for models up to 405 billion parameters. MSI also documents a switchless three-node ring over three DAC cables and a four-node star through a 200 GbE QSFP switch, scaling total unified memory to 512 GB.
Where the EdgeXpert Sits in the GB10 Family
The EdgeXpert MS-C931 is one of the GB10 desktops SourceIT stocks alongside the NVIDIA DGX Spark itself, with ASUS, Lenovo, Dell, HP, GIGABYTE, Acer and xFusion builds. All share the GB10 Superchip, 128 GB unified memory and 1 PFLOP FP4; the EdgeXpert is distinguished by MSI's FanDock cooling and SSD vapor chamber, its published three- and four-node cluster topologies, a Kensington lock slot and TPM, and the 3-year local warranty on this listing.
Package Contents & Ordering Information
What Ships in the Box
- MSI EdgeXpert MS-C931 (9S6-C9311-61S) with NVIDIA GB10 Grace Blackwell Superchip, 128 GB unified memory and 4 TB self-encrypting M.2 NVMe SSD
- 240 W USB PD adapter and power cord
- NVIDIA DGX OS preinstalled with the NVIDIA AI software stack
- Quick start guide
Monitor, keyboard and mouse are not included; MSI's specification sheet lists no peripherals. The QSFP cable for stacking units is sold separately.
MSI EdgeXpert Range at SourceIT Singapore
- MSI EdgeXpert MS-C931 4TB — 9S6-C9311-61S (this listing), 128 GB unified memory, 4 TB self-encrypting NVMe
- NVIDIA QSFP112 400G DAC Cable, 40 cm — X0101G00400A (links two units)
Other NVIDIA GB10 and DGX Station Systems at SourceIT
- NVIDIA DGX Spark 4TB — NVIDIA's own first-party GB10 unit
- ASUS Ascent GX10 — 1TB / 2TB / 4TB
- Lenovo ThinkStation PGX — 1TB / 4TB
- Dell Pro Max with GB10 — 2TB / 4TB
- HP ZGX Nano G1n — 4TB
- GIGABYTE AI TOP ATOM — 1TB / 4TB
- Acer Altos BrainSphere GB10 F1 — 4TB
- xFusion FusionXpark GB10 — 1TB / 2TB / 4TB
- MSI XpertStation WS300 (NVIDIA DGX Station GB300) — step up to 748 GB coherent memory and 20 PFLOPS for trillion-parameter models
Browse the full NVIDIA DGX Spark & DGX Station collection and the AI workstations and AI servers collection. For volume, project, research-grant or education pricing, GST invoicing and current lead times, email sales@sourceit.com.sg quoting 9S6-C9311-61S.
Datasheet, Downloads & Setup Guide
Official MSI and NVIDIA documentation for the EdgeXpert MS-C931 and the GB10 platform:
- MSI EdgeXpert MS-C931 — product and specification page
- MSI EdgeXpert — overview and specifications
- MSI EdgeXpert datasheet (PDF)
- MSI EdgeXpert MS-C931 user guide (PDF)
- NVIDIA DGX Spark (GB10) — official platform page (Singapore)
- NVIDIA DGX Spark Playbooks — step-by-step recipes for LLMs, fine-tuning, RAG and agents
- NVIDIA DGX Spark Setup and Enterprise Deployment Guide — SourceIT (applies to the EdgeXpert)
Frequently Asked Questions — MSI EdgeXpert MS-C931 4TB
What is the MSI EdgeXpert MS-C931?
MSI's personal AI supercomputer built on the NVIDIA DGX Spark GB10 platform: a 151 × 151 × 52 mm, 1.2 kg desktop with the NVIDIA GB10 Grace Blackwell Superchip, 128 GB of unified memory and 1 PFLOP of FP4 AI performance with sparsity, running NVIDIA DGX OS for local LLM inference, fine-tuning, data science and agentic AI. This configuration has a 4 TB self-encrypting NVMe SSD.
What is the difference between the MSI EdgeXpert and the NVIDIA DGX Spark?
Same NVIDIA GB10 Superchip, 128 GB unified memory, 1 PFLOP FP4 and ConnectX-7 networking, so AI performance is identical. The EdgeXpert is MSI's build with FanDock cooling and an SSD vapor chamber, HDMI 2.1 with DisplayPort 1.4a over USB-C, a Kensington lock slot, published three- and four-node cluster layouts and a 3-year local warranty on this listing; the DGX Spark is NVIDIA's first-party unit.
How does it compare with the other GB10 desktops and the DGX Station GB300 systems?
Against the other GB10 units, compute, memory and networking are identical; the differences are cooling, ports, storage tiers and warranty. The DGX Station GB300 systems such as the MSI XpertStation WS300 are a different class, with 748 GB of coherent memory and up to 20 PFLOPS FP4 for trillion-parameter models.
What AI models can the EdgeXpert run?
MSI states models up to 200 billion parameters on one unit, and NVIDIA's platform figures add fine-tuning up to 70 billion parameters — Llama, Qwen, DeepSeek, Gemma, Mistral, gpt-oss, Nemotron and FLUX-class image models in practice. Two stacked units handle up to 405 billion parameters per MSI; four GB10 units reach 700 billion per NVIDIA.
Can I link two EdgeXpert units together?
Yes. Two units connect back-to-back over the ConnectX-7 200 GbE QSFP ports with MSI's optional 400G QSFP112 cable (956-C931-101) or the NVIDIA QSFP112 400G DAC 40 cm cable. MSI also documents a three-node switchless ring and a four-node star through a 200 GbE QSFP switch.
Does the EdgeXpert run Windows or x86 software?
No. It ships with NVIDIA DGX OS (Ubuntu-based, arm64) and is a Linux AI development system. Mainstream AI frameworks and NVIDIA NGC containers are published for arm64; proprietary x86-only desktop applications will not run natively.
What is NemoClaw and does the EdgeXpert support it?
NemoClaw is NVIDIA's open-source reference stack that adds OpenShell policy-based security and privacy guardrails around OpenClaw agents and runs open models locally. NVIDIA publishes it for the DGX Spark GB10 platform and DGX OS includes a streamlined NemoClaw installation, so it applies to the EdgeXpert.
Does it need special power or cooling?
No. It runs from a standard socket through the supplied 240 W, 48 V USB-C PD adapter, and MSI's FanDock airflow design and smart fan are built to hold sustained AI load on a desk; MSI rates it for 0–35 °C operation. MSI publishes no acoustic figure.
Can I upgrade the memory or storage?
Memory is onboard and fixed at 128 GB. MSI offers the 128 GB EdgeXpert with 1 TB, 2 TB or 4 TB self-encrypting NVMe, chosen at order; this listing is the 4 TB tier. Choose the capacity you need at purchase so the warranty is unaffected.
Who is the MSI EdgeXpert for?
AI developers and ML engineers prototyping and fine-tuning before scaling to DGX Cloud; university labs and research groups; data scientists running large-memory analytics; robotics and computer-vision teams; and Singapore enterprises in finance, healthcare, legal and government that must keep models and data on-premise.
What warranty does the EdgeXpert come with in Singapore?
This listing carries a 3 Year Local Warranty in Singapore, supplied through SourceIT. Every unit comes with a GST invoice and local support.
How do I buy the MSI EdgeXpert in Singapore?
Order on this page or email sales@sourceit.com.sg quoting 9S6-C9311-61S. SourceIT is an authorised IT reseller in Singapore and supplies the EdgeXpert with GST invoice, 3-year local warranty and the QSFP clustering cable.
