NVIDIA Official Embedded Compute Dıstributor (MEA)
GIGABYTE AI TOP ATOM – 1 TB Version
Compact Desktop AI Supercomputer Powered by the NVIDIA GB10 Grace Blackwell Superchip and NVIDIA DGX™ Spark
GIGABYTE AI TOP ATOM is a next-generation high-performance AI platform that brings advanced artificial intelligence development processes directly to the desktop without cloud dependency. Built on the NVIDIA GB10 Grace Blackwell Superchip architecture, the system is optimized for model development, fine-tuning, inference, data science, and large-scale generative AI workloads.
With AI processing power reaching 1 PFLOP at FP4 precision and 128 GB of unified system memory, it reduces data transfer bottlenecks when working with large-scale parameter models and delivers high efficiency. The system creates smoother AI workflows through a high-bandwidth shared memory architecture between the CPU and GPU.
The AI TOP ATOM, designed in a desktop form factor, offers a powerful yet accessible solution for developers, researchers, data scientists, and teams looking to build in-house AI infrastructure. It is an ideal platform for local LLM development, RAG applications, model prototyping, post-training optimization, and custom AI workloads.
On the high-speed connectivity side, the NVIDIA ConnectX-7 SmartNIC, 10GbE network connection, Wi-Fi 7, and USB 3.2 Gen 2×2 Type-C interfaces provide expansion and scalability options. Two systems can be connected together to support larger models and more intensive AI scenarios. On the storage side, it offers 1 TB NVMe SSD support.
(The product also has a 4 TB version. To view it: ()
• NVIDIA® GB10 Grace Blackwell Superchip
• 1 PFLOP FP4 AI compute performance
• 128 GB LPDDR5x unified system memory
• 273 GB/s memory bandwidth
• Support for models up to 200 billion parameters
• NVIDIA® ConnectX®-7 SmartNIC
• 1 TB NVMe storage support (4 TB version also available)
• 10GbE + Wi-Fi 7 + Bluetooth 5.4 connectivity
• NVIDIA DGX™ OS / Ubuntu Linux support
High-End AI Performance
Powered by the NVIDIA® GB10 Grace Blackwell Superchip, the system delivers 1 petaFLOP of AI performance and 128 GB of memory for fine-tuning models with up to 200B parameters.
Developer-Focused Platform
Designed for AI researchers and developers, it provides full-stack compatibility and optimized workflows.
Scalable Architecture
It enables high scalability by allowing two devices to work together via the ConnectX-7 network.
Advanced Thermal Engineering
A specially designed cooling system ensures sustained high performance and reliability even in this ultra-compact form factor.
Ideal for labs, offices, and edge environments; it delivers high-density computing power in a compact form factor.
End-to-End AI Solution
It provides a complete AI solution covering the entire workflow—from development to deployment—powered by the GB10 platform and the NVIDIA AI software stack.
Small Footprint, Big Power
| CPU | |
|---|---|
| Architecture | NVIDIA Turing GPU Architecture |
| Tensor performance | 1 petaFLOP |
| RAM | 128GB |
| Memory Interface | 256-bit |
| Memory bandwidth | 273 GB/s |
| Power consumption | 240 W |
| OS | Linux® Ubuntu, NVIDIA AI Software Stack, NVIDIA DGX™ OS |
| Networking | 1x GbE 1x 10GbE |
| WiFi | Wi-Fi 7 |
| Bluetooth | Bluetooth 5.4 |
| NIC | NVIDIA® ConnectX®-7 NIC |
| Audio | HDMI multichannel audio output |
| Weight | 1.2 kg |
| Dimension | 50,5 x 150 x 150 mm |
| Other I/O | 1 × USB 3.2 Gen 2×2 Type-C (PD input supported, 180W EPR PD3.1 SPEC), 3 × USB 3.2 Gen 2×2 Type-C (20Gbps, sub-mode with DisplayPort 2.1 support), Power button |
| Display | 1x HDMI 2.1 |
| LAN | 1 x 10G LAN |
OpenZeka is an NVIDIA Embedded Compute distributor in the MENA region, an Elite Partner in Visualization, and a partner for NVIDIA DGX compute systems.
+90 312 266 2055
Universiteler Mah. Sehit Mustafa Tayyarcan Cad.Tepe Binasi No:5, Ic Kapi No:315 06800 Cankaya, Ankara, Turkiye
OpenZeka is an NVIDIA Embedded Compute distributor in the MENA region, an Elite Partner in Visualization, and a partner for NVIDIA DGX compute systems. Supported by Microsoft® BizSpark. © 2009-2026 NVIDIA Corporation. NVIDIA, the NVIDIA logo, 4-Plus-1, Kepler, Tegra, Turing, Tesla, Ampere, Jetson TK1, Jetson TX1, Jetson TX2, Jetson Xavier, Jetson Nano, VisionWorks, and CUDA are trademarks and/or registered trademarks of NVIDIA Corporation in the United States and other countries. Other company and product names may be trademarks of the companies with which they are associated.
