The MSI EdgeXpert is a compact desktop AI supercomputer for developers, researchers and data science teams who want to build, fine-tune and run AI models locally. It combines the NVIDIA GB10 Grace Blackwell platform with NVIDIA DGX OS and a preconfigured AI software stack for local development.

Its 128 GB of unified LPDDR5x memory is shared between the CPU and GPU, supporting workloads that benefit from a large coherent memory pool. The system delivers up to 1 petaFLOP of AI performance and includes 4 TB of self-encrypting NVMe M.2 storage. ConnectX-7 networking supports high-speed links between systems for larger clustered workloads, while the compact chassis fits on a desk or in an edge deployment.

Key features

• NVIDIA Grace Blackwell architecture combines a Blackwell GPU with a 20-core Arm CPU.
• 128 GB of unified LPDDR5x system memory uses a 256-bit interface.
• 273 GB/s memory bandwidth supports data-intensive AI work.
• Up to 1 petaFLOP AI performance for supported FP4 sparse workloads.
• Preconfigured NVIDIA DGX OS includes the NVIDIA AI software stack.
• Self-encrypting 4 TB NVMe M.2 storage provides local capacity for models and project data.
• ConnectX-7 Smart NIC enables high-speed networking and multi-system scaling.
• Four USB 3.2 Type-C ports and Wi-Fi 7 support peripheral and network connections.

Compatibility

The system runs NVIDIA DGX OS, an Ubuntu Linux-based operating system for Arm64, and is intended for AI development and inference workflows. Video output is provided through HDMI 2.1 and DisplayPort over USB-C; it also has a 10 GbE RJ-45 port, Bluetooth 5.4 and HDMI multichannel audio output. Good to know: this is an Arm-based AI developer system with unified CPU and GPU memory, so check software and peripheral support for DGX OS and Arm64 before buying.

In the box

• MSI EdgeXpert

Specifications

For local AI work in a small footprint, EdgeXpert brings its compute, memory, storage and NVIDIA software platform together in one desktop system.