MSI Launches EdgeXpert 64GB Based on GB10 Platform for On-Premises AI Agents
MSI has announced a new 64GB configuration of its EdgeXpert desktop AI supercomputer. Built on the NVIDIA DGX Spark™ GB10 platform, this model is designed to accelerate the deployment of autonomous agentic AI workflows directly on-premises, allowing enterprises to scale AI capabilities without relying on cloud infrastructure.
Scaling AI Deployment with Cost Efficiency
The new 64GB model complements the existing 128GB tier, providing a more cost-effective path for enterprises to scale AI agents across multiple locations. According to Eunice Huang, Vice President of MSI IPC, while the 128GB version serves development teams working with up to 200B parameter models, the 64GB model targets deployment-first organizations. This allows companies to deploy fleets of units at branch offices, factory floors, or retail sites where unit cost efficiency is a primary driver.
Powered by the NVIDIA DGX Spark platform (delivering 1 petaFLOP FP4 AI compute), the MSI EdgeXpert 64GB can run open-source models with up to 100B parameters entirely behind a corporate firewall. This ensures total data sovereignty, making it suitable for highly regulated industries such as banking, legal services, and healthcare.
Key Technical Specifications
Designed for sustained, 24/7 workloads, the compact 1.2-liter, 1.2-kg desktop unit includes several enterprise-grade features:
- 64GB of LPDDR5x unified memory;
- High-speed connectivity including 10GbE and 200GbE NVIDIA ConnectX-7 NIC for system clustering;
- Turnkey software solution with pre-installed NVIDIA DGX OS and the full NVIDIA AI software stack;
- Management capabilities via NVIDIA Sync to monitor CPU and GPU activity from a single application.
The MSI EdgeXpert 64GB is scheduled to become available on October 23, 2026.
Гика
