Key information
MSI announced a new 64 GB configuration of its EdgeXpert desktop AI supercomputer, built on the NVIDIA DGX Spark platform. The system joins the existing 128 GB tier to provide enterprises with a cost-effective option for scaling AI agents across various deployments.
Powered by the NVIDIA DGX Spark platform, which delivers 1 petaFLOP FP4 AI compute, the 64 GB model is capable of running open-source models of up to 100B parameters locally behind a firewall. This enables applications such as real-time defect detection in manufacturing, fraud investigation in banking and legal, and clinical documentation in healthcare.
Eunice Huang, Vice President of MSI IPC, stated that while the 128 GB EdgeXpert serves development teams working with larger models, the 64 GB model focuses on deployment-first organizations seeking unit cost efficiency. A standard enterprise rollout often begins with 128 GB units at headquarters before expanding to fleets of 64 GB units at branch offices and factory floors.
The hardware features 64 GB of LPDDR5x unified memory, thermal engineering for sustained workloads, and a 1.2-liter, 1.2-kg compact design with whisper-quiet cooling. Connectivity options include 10GbE networking and a 200GbE NVIDIA ConnectX-7 NIC for clustering, alongside pre-installed NVIDIA DGX OS and the NVIDIA AI software stack.
The MSI EdgeXpert 64 GB model is scheduled to become available on October 23, 2026.
