LITEON and QCT Showcase NVIDIA Vera Rubin NVL72 Platform at 2026 OCP Global Summit

ADVERTISEMENT

LITEON Technology (2301.tw) and Quanta Cloud Technology (QCT), a leading provider of data center and AI solutions, jointly showcased an AI rack-level infrastructure solution supporting the NVIDIA Vera Rubin NVL72 platform at the 2026 OCP Global Summit. The demonstration features LITEON’s 110kW Power Shelf integrated within a rack-level power architecture, highlighting how advanced power delivery and system integration can support the growing token-processing requirements driven by agentic AI while improving computing performance and energy efficiency for AI factories.

As AI computing density continues to increase, power infrastructure has become a critical factor in determining overall data center performance. In modern AI racks where GPUs, networking equipment, and thermal management are tightly integrated, power conversion efficiency, dynamic load management, and rack space utilization directly impact deployable compute capacity and operational efficiency. Excessive power system footprint can reduce space available for computing resources, while power conversion losses generate additional heat that increases cooling requirements and operating costs. As a result, next-generation AI infrastructure requires power architectures optimized not only for power density, but also for efficiency, responsiveness, and system resilience at the rack level.

LITEON showcases its 110kW Power Shelf designed to support NVIDIA MGX architecture. Delivering high power density within a compact footprint, the solution provides the rack-level power required by NVIDIA Vera Rubin NVL72 systems while maximizing space available for compute resources. Equipped with high-efficiency power conversion, intelligent load balancing, and digital power management capabilities, the solution is designed to accommodate rapid power fluctuations associated with AI training, inference, and Agentic AI workloads, ensuring reliable operation while reducing energy loss and thermal burden. Combined with QCT’s expertise in AI platforms, system design, and data center integration, the two companies are working closely on platform-level co-design and validation to help customers accelerate deployment of high-density AI infrastructure built on the NVIDIA Vera Rubin NVL72 platform.

 “AI is evolving from generative AI toward agentic AI, driving unprecedented growth in token processing and redefining the role of data center infrastructure,” said Anson Chiu, President of LITEON Technology. “The challenge is no longer simply delivering more power, but ensuring every watt contributes to greater AI performance and business value. LITEON continues to extend its leadership in high-efficiency power technologies from shelf to rack- and system-level energy architectures. Together with QCT’s expertise in AI platforms and system integration, we are helping customers achieve the optimal balance of performance, energy efficiency, and scalability while accelerating the development of next-generation AI factories.”

“The rise of agentic AI is driving AI infrastructure evolution from standalone servers to rack-level and even POD-level systems,” said Mike Yang, President of QCT. “As AI workloads become increasingly complex and require uninterrupted operation, compute, networking, storage, and power systems must be designed in closer coordination. By combining QCT’s comprehensive AI platform and system integration capabilities with LITEON’s expertise in high-efficiency power delivery and rack-level energy architecture, we help customers improve overall system efficiency and accelerate large-scale AI infrastructure deployment.”

With decades of experience in cloud data center power solutions, LITEON continues to expand its technology portfolio beyond server power supplies into Power Shelves, Power Racks, and system-level energy management. The company is driving the transformation of power infrastructure from component optimization to holistic infrastructure optimization. Through close collaboration with ecosystem partners such as QCT on platform-level integration and co-design, LITEON helps customers build AI infrastructure that delivers high performance, reliability, and scalability while addressing the growing demands for computing power, energy efficiency, and system resilience, enabling the continued expansion of AI factories worldwide.