Gigabyte announced the launch of a new 64 GB unified memory version of the AI TOP ATOM, expanding its existing 128 GB offering. Available starting October 23, the new configuration retains the hardware design of AI TOP ATOM, based on the NVIDIA DGX Spark platform, giving developers, researchers, and enterprise teams greater flexibility to select the configuration that best fits their AI workloads and memory requirements. The new 64 GB version, together with the existing high-capacity 128 GB version, creates a more comprehensive product lineup.
Users can evaluate and select the memory capacity that best matches their model sizes, workflows, and multitasking requirements, enabling them to develop AI applications on the same platform and progress from early proof-of-concept development to practical deployment. With built-in Connect X-7 networking, developers can cluster up to 4 units with NVIDIA Sync for larger memory pool and compute capability. On the software side, AI TOP ATOM integrates NVIDIA CUDA accelerated AI software ecosystem with Gigabyte AI TOP Utility, providing capabilities including model downloading, inference, and retrieval-augmented generation (RAG).
Gigabyte is also continuing to explore the potential of AI TOP ATOM for agentic AI applications. In a multi-node scientific computing demonstration, Gigabyte integrated NVIDIA Nemotron open models with the NVIDIA Nemo Claw open agent blueprint to connect research hypothesis generation with simulation workflows. With the addition of the 64 GB unified memory version, AI TOP ATOM is now available in both 128 GB and 64 GB configurations, addressing the needs of users across different stages of AI development and application scales.
Gigabyte will continue to expand its AI TOP product portfolio and software ecosystem, helping individuals and enterprise teams build on-premises AI capabilities and bring AI from technology exploration into everyday applications. The Gigabyte AI TOP ATOM 64 GB unified memory version is officially available starting October 23, 2026.
Source: TechPowerUp News · Summarized by HeadlinesBriefing