Giga Computing Announces New AI Deskside Supercomputer to Build and Run AI; On Display at CES 2026

Server

New Data Center Solutions Scaling for AI

Jan 06, 2026

January 6, 2026 - Giga Computing, a subsidiary of GIGABYTE and an industry innovator and leader in AI hardware and advanced cooling solutions, today announced a new deskside supercomputer solution based on NVIDIA DGX station and AI technology, GIGABYTE W775-V10, to power advanced AI development from the desktop. The W775 accelerated by NVIDIA GB300 Grace Blackwell Ultra Desktop platform features a massive 775GB of coherent memory to accelerate large AI training and inference workloads locally. This system is now on display at the GIGABYTE booth at CES 2026. Also at CES, Giga Computing shows high-performance systems scaling from a mini PC to server to rack scale infrastructure and deployment.

GIGABYTE W775-V10: Liquid Cooled AI Desktop

This is the ultimate deskside AI supercomputer designed to build and run AI. It features the NVIDIA GB300 Grace Blackwell Ultra Desktop platform and a massive 775 GB of coherent memory—delivering an unprecedented amount of compute performance for developing and running large AI training and inferencing workloads locally. Preconfigured with the NVIDIA AI software stack, AI developers, researchers, and data scientists can rapidly prototype, fine-tune, and inference large AI models from the deskside and seamlessly deploy to the data center or cloud. Combining state-of-the-art system capabilities with NVIDIA AI software stack, this GIGABYTE W775-V10 is purpose-built for teams that demand the best deskside AI development platform.

Highlights at the GIGABYTE Booth:

GIGAPOD: Rack-scale AI Solutions. These customizable solutions incorporate 32 GIGABYTE GPU servers with networking, storage, and more to deliver excellent performance for LLMs and other demanding workloads. At CES, Giga Computing shows liquid cooling solutions using AMD EPYC™ + AMD Instinct™ MI355X GPUs in the GIGABYTE G4L3-ZX1, as well as a solution using Intel Xeon and based on an NVIDIA HGX B300 system in the GIGABYTE G4L4-SD3. Also, a management rack is needed for successful rack deployment, so the GIGABYTE N120-LS0 is on display. This server simplifies management of infrastructure components (CDU, PDU, and RDHx) that are responsible for power and cooling. Learn more about GIGAPOD: https://www.gigabyte.com/Solutions/giga-pod-as-a-service
NVIDIA MGX Platforms: Two tailored and balanced solutions accelerate time to market for AI factories and accelerated workloads. These two modularized servers are built following a 2:8:5 topology: two Intel Xeon CPUs are paired with either eight NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs or eight NVIDIA RTX PRO 4500 Blackwell GPUs. The five networking components are composed of a single NVIDIA BlueField-3 DPU and four NVIDIA ConnectX-8 SuperNICs. Models: air cooled GIGABYTE XL44-SX2 and a liquid-cooled version.
NVIDIA GB300 NVL72: A powerful rack-scale platform that integrates 72 NVIDIA Blackwell Ultra GPUs and 36 Arm^®-based NVIDIA Grace CPUs. This ready-made architecture allows Giga Computing to quickly get a powerful system for demanding workloads up and running for the needs of hyperscalers, cloud providers, and AI service companies. Learn more about NVIDIA GB300 NVL72: https://www.gigabyte.com/Solutions/nvidia-blackwell

With the introduction of the GIGABYTE W775-V10 and its expanded portfolio of advanced AI systems, Giga Computing continues to push the boundaries of what is possible for AI innovation, from the deskside to the data center. By integrating cutting-edge NVIDIA technologies with high-density server design and precision-engineering thermal solutions, Giga Computing empowers organizations of all sizes to accelerate development, streamline deployment, and scale AI workloads with confidence.

Visit GIGABYTE booth #8519 in the Las Vegas Convention Center North Hall at CES.

For queries or more information, please contact sales.