Skip to Content

Gigabyte - AI TOP ATOM

Get the power of a supercomputer right at your workplace.

Request a quote

Download data sheet

GIGABYTE AI TOP ATOM kaufen
Delivery time: 4-7 business days

Factsheet



NVIDIA® GB10 Grace Blackwell Superchip

This superchip combines the groundbreaking Blackwell GPU architecture with a powerful 20-core Arm CPU (10x Cortex-X925 and 10x Cortex-A725). Thus, the system delivers an impressive AI performance of up to 1 PetaFLOP in a compact design.


128GB Coherent Unified System Memory

The system is equipped with 128 GB of fast LPDDR5x RAM , which functions as coherent, unified memory. With a 256-bit interface, this enables efficient data exchange between the CPU and GPU for complex calculations.


Support for Models up to 200B Parameters

Due to the high memory capacity and computing power, local AI models with a size of up to 200 billion parameters can be processed. For even larger requirements, two AI TOP ATOM units can be connected together.


NVIDIA® ConnectX®-7 SmartNIC

For maximum connectivity and low latency, an NVIDIA® ConnectX®-7 network card is integrated. This supports, among other things, a 10 GbE RJ-45 Ethernet port for professional network environments.


Compact Desktop Form Factor

With dimensions of just 150 mm x 150 mm x 50.5 mm and a weight of 1.5 kg , the device is a true "Personal Supercomputer" for the desk. Despite its small size, it offers full performance with moderate power consumption of 240W.


NVIDIA AI Software stack

The device runs on the NVIDIA DGX™ OS and offers full access to the entire NVIDIA AI software stack. Together with the GIGABYTE AI TOP Utility, this enables a seamless workflow from model download to local training and deployment.

GIGABYTE AI TOP ATOM

Local AI development with petaflop performance


Accelerate data science workflows


Ideal for prototyping and fine-tuning domain-specific AI models


Supports edge applications in the fields of robotics, vision, and multimodal AI.

Ready for your personal AI supercomputer?

Bring the power of a data center directly to your workspace. Request a customized quote for the GIGABYTE AI TOP ATOM now and start your local AI development without compromises.


GIGABYTE AI TOP ATOM: Petaflop Performance Right at Your Desk The GIGABYTE AI TOP ATOM brings the computing power of a data center directly to your workspace. With up to 1 PetaFLOP of AI performance and the NVIDIA GB10 Grace Blackwell Superchip , the system is aimed at developers and researchers who want to implement their AI projects locally, securely, and energy-efficiently. Thanks to the 128 GB coherent unified system memory , even huge AI models with up to 200 billion parameters can be executed smoothly.

Concrete Everyday Use Cases

  • Prototyping and Fast Iterations:

    The local environment makes developing AI prototypes particularly straightforward and fosters a fast, uninterrupted culture of innovation.

  • Fine-Tuning with Sensitive Data:

    Adapting pre-trained models – for instance, for company-specific chatbots – happens completely locally on the device. This ensures high speed and guarantees the protection of confidential data.

  • Secure Inference:

    For testing models or running personal AI assistants permanently, the system delivers the necessary computing power for fast and secure inference directly from the desktop.

  • Data Science Workflows:

    The entire process, from initial data exploration to final prediction, is handled locally, giving data scientists maximum speed and privacy while working.

  • Powerful Edge Applications:

    Thanks to its compact form factor , the system is ideally suited for deployment directly at the source of data generation (edge), for example, for demanding robotics, machine vision, or multimodal AI.


Software That Simplifies the Start A major advantage of the system is the seamless interaction of hardware and software. The AI TOP ATOM supports the full NVIDIA AI software stack. To get started right away, the device comes with the GIGABYTE AI TOP Utility. This platform bundles essential tools for model downloading, machine learning, RAG (Retrieval-Augmented Generation), and inference in an intuitive interface. This allows developers to start their projects directly without complex setups