GIGABYTE AI TOP ATOM
Key Features
AI Supercomputer
Empower Advanced AI with GIGABYTE AI TOP ATOM. AI TOP ATOM is your personal AI supercomputer, delivering petaflop-scale performance in a compact, energy-efficient form factor—perfect for local AI development right at your desk.
NVIDIA® GB10 Grace Blackwell Superchip
Supports up to 200B parameter models
1 petaflop of FP4 computing power
128GB of coherent unified system memory
Up to 4 TB of NVMe storage
NVIDIA® ConnectX®-7 SmartNIC
NVIDIA AI Software stack
Compact desktop form factor
From prototyping and fine-tuning to inference, data science, and edge AI—AI TOP ATOM delivers the desktop power and memory you need to run today’s most demanding AI workloads.
At its core, the NVIDIA ® GB10 Grace Blackwell Superchip and fifth-generation Tensor Cores drive cutting-edge performance and reliability in a compact, developer-friendly design.
Easily handle large-parameter generative AI models through high-bandwidth, low-latency communication enabled by the ConnectX® 7 SmartNIC. By connecting two AI TOP ATOM systems, you can scale up your AI workloads and tackle even more demanding models with seamless interconnect performance.
GIGABYTE AI TOP ATOM supports the full NVIDIA AI software stack and includes exclusive access to the GIGABYTE AI TOP Utility, enabling users to easily to train, inference and deploy AI workloads locally. It also offers flexibility to scale across more powerful local or on-prem accelerated infrastructures for advanced AI applications.
Delivering the power of an AI supercomputer in a desktop-friendly size, GIGABYTE AI TOP ATOM is ideal for AI developer, enthusiast, student, researcher, and data scientist workloads.
The AI TOP Utility brings essential AI capabilities into a compact, intuitive platform, enabling users to explore and deploy AI models locally with ease. With a simplified workflow and streamlined tools, it allows developers, researchers, and enthusiasts to start their AI projects without complex setup.
Current Capabilities:
- Model Download
- Inference
- Retrieval-Augmented Generation (RAG)
- Machine Learning

* The terms HDMI, HDMI High-Definition Multimedia Interface, HDMI Trade dress and the HDMI Logos are trademarks or registered trademarks of HDMI Licensing Administrator, Inc.