GIGABYTE NVIDIA GB300 Grace™ Blackwell Ultra Desktop Superchip (W775-V10-L01)

​

Enterprise AI Workstation for Large Language Models, Generative AI, and High-Performance Computing

The GIGABYTE W775-V10-L01 is a next-generation AI workstation powered by the NVIDIA GB300 Grace™ Blackwell Ultra Desktop Superchip. It combines a 72-core NVIDIA Grace™ Arm Neoverse V2 CPU and an NVIDIA Blackwell Ultra GPU connected through NVIDIA NVLink-C2C technology, delivering data-center-class AI performance directly at the desktop. It is designed for AI training, AI inference, large language models (LLMs), machine learning, data science, and high-performance computing workloads.

Featuring up to 748GB of coherent unified memory, including 496GB LPDDR5X and 252GB HBM3E, the system enables developers and enterprises to build, fine-tune, and deploy massive AI models locally without relying on external cloud infrastructure. Built on NVIDIA DGX Station™ architecture, the W775-V10-L01 delivers up to 20 PFLOPS of AI performance for advanced AI development environments.

Key Features of GIGABYTE W775-V10-L01

​

NVIDIA GB300 Grace™ Blackwell Ultra Desktop Superchip

Powered by a 72-core NVIDIA Grace™ Arm Neoverse V2 CPU and NVIDIA Blackwell Ultra GPU connected through NVLink-C2C, enabling high-speed CPU-GPU communication and shared memory architecture for accelerated AI computing.

Up to 748GB Unified Coherent Memory

Includes:

• 496GB LPDDR5X ECC Memory
• 252GB HBM3E GPU Memory
• Up to 748GB Total Unified Memory

Ideal for running large AI models, LLM fine-tuning, generative AI applications, and memory-intensive analytics workloads.

Up to 20 PFLOPS AI Performance

Delivers up to 20 petaFLOPS of AI computing power, enabling enterprise-scale AI training, inference, simulation, and next-generation generative AI development directly from a desktop workstation.

NVIDIA Blackwell Ultra GPU with HBM3E

Equipped with a Blackwell Ultra GPU featuring 252GB HBM3E memory and up to 7.1TB/s memory bandwidth for accelerated AI workloads, simulation, rendering, and scientific computing applications.

Advanced Liquid Cooling System

Features a closed-loop liquid cooling solution engineered for continuous AI workloads, ensuring reliable thermal performance and system stability during intensive processing.

Enterprise Expansion Capabilities

Supports:

• 2 x M.2 PCIe Gen6 x4
• 2 x M.2 PCIe Gen5 x4
• PCIe Gen5 x16 Slot
• PCIe Gen5 x8 Expansion Slots

Providing flexible scalability for storage expansion and additional acceleration hardware.

High-Speed Data Center Networking

Includes:

• 2 x 400Gb/s QSFP Ports via NVIDIA ConnectX-8 SuperNIC
• 1 x 10GbE LAN

Designed for AI clusters, distributed AI training, and enterprise-scale networking environments.

NVIDIA DGX OS and AI Software Stack Support

Compatible with NVIDIA DGX OS, NVIDIA AI Enterprise, CUDA, TensorRT, RAPIDS, NVIDIA NIM, and leading AI frameworks for accelerated development and deployment workflows.

Ideal Applications

​

• Large Language Models (LLM)
• Generative AI Development
• AI Training & Fine-Tuning
• AI Inference Workloads
• Deep Learning
• Machine Learning
• Data Science
• Scientific Computing
• Simulation & Modeling
• Healthcare AI Research
• Financial Analytics
• Autonomous Systems Development
• Enterprise AI Labs
• Private AI Infrastructure

Why Choose GIGABYTE W775-V10-L01

​

The GIGABYTE W775-V10-L01 brings NVIDIA’s GB300 Grace™ Blackwell Ultra Desktop Superchip architecture into a powerful enterprise workstation platform. With up to 748GB unified memory, 20 PFLOPS AI performance, advanced liquid cooling, and NVIDIA DGX Station-class capabilities, it enables organizations to train, fine-tune, and deploy large AI models locally while maintaining data security, reducing cloud dependency, and accelerating AI innovation.