At NVIDIA GTC 2026, Compal Electronics took center stage by unveiling its latest AI server innovation. The Compal HGX Rubin NVL8 solution marks a significant step towards advancing next-generation AI supercomputing infrastructure. As the demand for powerful data center solutions rises, Compal positions itself at the forefront with a robust, scalable platform tailored to today’s evolving workloads.
Compal Showcases High-Density AI Servers at GTC 2026
During the event, Compal introduced its high-density server built on the NVIDIA HGX Rubin NVL8 platform. This new offering aligns closely with NVIDIA’s “Six New Chips – One AI Supercomputer” initiative. The server leverages NVIDIA Vera Rubin architecture, ensuring exceptional computational power and seamless integration. Attendees witnessed Compal’s engineering readiness with a live demonstration of their SG231-2-L1 model, specifically designed for demanding AI and data center applications.
Key Features of the Compal HGX Rubin NVL8 Solution
The Compal SG231-2-L1 brings several advanced capabilities to the table, including:
- High-Density Accelerated Architecture: Eight NVIDIA Rubin GPUs fit within a 2U chassis for maximum density.
- Advanced Inference Performance: Achieves up to 400 petaFLOPS (NVFP4) to support LLM training and generative AI workloads.
- High-Bandwidth GPU Interconnect: Utilizes NVIDIA NVLink 6 for up to 28.8TB/s GPU-to-GPU bandwidth.
- Scalable Memory Architecture: Supports up to 2.3TB GPU memory and 176TB/s bandwidth for intensive applications.
- Liquid Cooling System: Direct liquid-cooling handles up to 24kW system power, ensuring stability.
- Rack-Level Deployment Readiness: Built for modular expansion from single servers to full-scale data centers.
Compal’s Strategic Role in Evolving AI Data Center Ecosystems
Compal’s advancements go beyond hardware. The company showcases support for the NVIDIA Vera CPU HPM module and the new NVIDIA RTX PRO 4500 Blackwell Server Edition, highlighting its commitment to integrated solutions. According to Alan Chang, Vice President of Compal, future data centers must focus on deployment efficiency and total ecosystem alignment. Compal aims to provide not just high-performance compute, but also the load-bearing and expansion capabilities needed for sustainable, scalable AI infrastructure.
In summary, Compal HGX Rubin NVL8 stands out as a high-density, high-performance server solution designed for the future of AI workloads. By combining powerful NVIDIA technology with innovative engineering, Compal strengthens its position in the rapidly evolving world of AI data centers. As demand for robust infrastructure grows, Compal’s integrated approach signals a promising path forward for organizations looking to harness next-generation computing power.
Don’t miss our latest Startup News: Rogo Boosts Financial Workflows With Strategic Offset Acquisition