Taipei – – October 02, 2026 -- GIGABYTE will launch a 64GB unified memory version of its AI TOP ATOM desktop AI system on October 23, expanding its existing 128GB offering and giving enterprise teams a lower-entry option for on-premises AI development.
GIGABYTE adds a second memory tier to its NVIDIA DGX Spark-based desktop platform
The new configuration retains the hardware design of AI TOP ATOM, which is built on the NVIDIA DGX Spark platform, and targets developers, researchers, and enterprise teams that need to match memory capacity to specific model sizes and multitasking workloads. The 64GB and 128GB versions now form a two-tier product lineup covering use cases from early proof-of-concept testing to practical deployment.
Built-in ConnectX-7 networking allows clustering of up to 4 units
Developers can use NVIDIA Sync to link up to four AI TOP ATOM units, pooling memory and compute capability for larger workloads. The system is designed to run model inference, prototype development, and data analysis locally in offices, laboratories, and educational settings, reducing dependence on cloud computing resources for data-sensitive projects.
Software stack pairs NVIDIA CUDA with GIGABYTE AI TOP Utility for local AI workflows
The integrated software ecosystem supports model downloading, inference, and retrieval-augmented generation (RAG), enabling users to test AI assistants or build knowledge-based question-and-answer applications from their own documents. GIGABYTE is also testing agentic AI use cases, having integrated NVIDIA Nemotron open models with the NVIDIA NemoClaw open agent blueprint in a multi-node scientific computing demonstration linking research hypothesis generation to simulation workflows.
Product availability, sales channels, and pricing for the 64GB version will vary by region, with details to follow through GIGABYTE's authorized local distributors.