Giga Computing, a subsidiary of GIGABYTE and a leader in accelerated computing and infrastructure solutions, today unveiled at Computex 2026 to showcase live demos and the latest in AI solutions powered by NVIDIA from creating tokens on a lightweight, compact PC, to deskside supercomputers, to rapidly deploying a containerized data center. GIGABYTE booth #K0802 features a comprehensive portfolio of NVIDIA-powered AI infrastructure solutions to accelerate next-gen computing across enterprise, research, and industrial applications.

Featured innovations include the groundbreaking NVIDIA Vera Rubin NVL72 platform for large-scale AI training & inference and agentic AI. NVIDIA Vera Rubin NVL72 supercharges the era of agentic AI - delivering inference with 10x more throughput per watt, 1/10th the token cost and training with one-fourth the GPUs. Visitors can see firsthand all the components that go into the 3rd generation rack-scale platform that make it truly amazing, including compute and networking. Rack-scale solutions are achieving far greater performance gen on gen and becoming more power dense. A new 800 VDC solution for next-gen power distribution is also displayed.

Real-to-Sim-to-Real
Adjacent to the rack-scale solution is a real-to-sim-to-real solution, spanning digital twin simulation, model training, and real-time edge control. GIGABYTE servers architected for NVIDIA RTX PRO 6000 Blackwell (XLS4-SX2) are used to capture real-world data to recreate a simulated environment. At the second stage, a server based on NVIDIA HGX Rubin NVL8 system (G2L4-SD4) is used to train and optimize AI models, as well as handle reinforcement learning. This NVIDIA HGX platform scales AI factories for enterprise, accelerating AI and high-performance computing (HPC) workloads. That last stage is deploying the AI models into physical systems such as robots and smart factories. To do this a system built on NVIDIA Jetson Orin NX (QN-RNX16GH-A1) is used. Throughout the process, GIGABYTE solutions are paired with NVIDIA software for a fluid and seamless ecosystem. Some of the NVIDIA software used includes: NVIDIA Omniverse, NVIDIA Cosmos, NVIDIA Isaac Lab, and NVIDIA Isaac GR00T N1.6.

Real World Physical AI
To show a real-world application of physical AI, two robotic arms with dexterous hands install a GPU into a server. By using robotic systems, GPUs and other components can be precisely installed, wherein increasing assembly consistency, reduce damage, and overall better-quality control to support 24/7 manufacturing. Powering this demonstration is NVIDIA technologies: NVIDIA Isaac Sim, Isaac Lab, and Isaac GR00T.

Deskside AI Supercomputer
The enterprise hardware at the GIGABYTE booth isn't just for data centers. Powerful desktop solutions are also on display. GIGABYTE W775-V10 is a new closed loop deskside AI supercomputer. designed to build and run AI. Featuring the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip, Ubuntu with NVIDIA AI Developer Tools, and a massive 748 GB of coherent memory, it is the ultimate deskside solution for local AI development, inferencing, agentic AI, and data science workloads.

GIGABYTE deskside systems powered by the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip will also be coming to Windows later this year, empowering enterprise teams to build, run, and connect always-on, frontier AI agents to Windows applications and workflows.

AI Agents on a Mini PC
Coming in at nearly 1L, the GIGABYTE AI TOP ATOM is ideal for developers, researchers, and data scientists, offering full access to the NVIDIA AI software stack and ecosystem. This solution offers the ideal desktop platform for autonomous AI agent development and deployment, providing the compute, large GPU memory and power efficiency required for always on agents.

AI TOP ATOM - Live Demos: Scenario 1: Run concurrent multi-model inference. Scenario 2: Run autonomous workflows with NVIDIA NemoClaw.

AI Infrastructure is Ready
NVIDIA DSX is an AI factory-scale platform, unifying design, simulation, operations, and ecosystem technologies to help build AI factories optimized for lowest token cost per watt. As a reference design and platform for AI Factory design, simulation, and operations, DSX defines how AI factories are architected, built, and optimized across the full stack, from hardware to software infrastructure, facilities, and partner technologies.

Giga Computing is adopting the NVIDIA DSX platform to accelerate deployment of AI factories using their rack-scale systems in row-scale and data hall modular designs with DSX partners. As an NVIDIA AI infrastructure partner, Giga Computing is also adopting NVIDIA DSX SimReady (the asset standard for AI factory digital twins) across its rack-scale platforms, converting CAD geometry into validated OpenUSD assets with embedded power, thermal, and connection-point metadata. These SimReady assets plug directly into the NVIDIA Omniverse DSX Blueprint and the broader DSX ecosystem, including Cadence Reality Digital Twin Platform and other authoring platforms, enabling hardware to be designed, verified, and optimized in simulation before physical build-out. This positions Giga Computing to deliver pre-validated AI factory hardware to gigawatt-scale customers and compress data center bring-up timelines.⁠

Building on this foundation, Giga Computing is taking its AI infrastructure strategy to the next level with two key milestones. First, GAIFA (GIGABYTE AI Factory Accelerator), a new AI Factory launching in Taiwan, scheduled to come online by end of Q3. It is designed to enable rapid AI deployment at scale, with NVIDIA DSX playing a critical role in streamlining integration, scalability, and operational readiness. Second, the newly introduced GIGABYTE Accelerated Deployment Unit (GADU) offers a modular, containerized approach to deploying scalable AI infrastructure in virtually any environment. Integrating IT nodes with dedicated power and cooling modules, GADU delivers plug-and-play AI infrastructure with a minimal physical footprint, often described as an "AI factory in a box." Together, GAIFA and GADU reflect Giga Computing's commitment to delivering agile, factory-scale AI deployments that are ready to perform from day one.⁠

Portfolio Growth
In addition to NVIDIA DSX, Giga Computing continues to support a growing portfolio of NVIDIA AI technologies and infrastructure platforms, further accelerating the delivery of scalable, efficient, and production-ready AI factory solutions for customers worldwide. Including: