Compal Debuts Vera Rubin Server at GTC 2026
On March 16, 2026, at NVIDIA's GTC conference, Compal Electronics (2324) provided an early look at its next-generation AI server. The high-density system is built on NVIDIA's forthcoming Vera Rubin architecture, integrating the new Vera CPU and Rubin GPU. This reveal demonstrates that key hardware partners are already engineering solutions for NVIDIA's next major platform, preparing for the transition beyond the current Blackwell systems.
Compal's showcase centered on its "ONE Integrated Solution," a rack-level design for compute, power, and liquid cooling. This approach directly addresses the escalating power and density requirements of advanced AI, pushing data center construction away from single-node designs and toward holistic infrastructure planning.
Agentic AI Drives CPU Market to Projected $60 Billion
Compal's announcement aligns with a major infrastructure shift driven by the rise of "agentic AI." These complex workflows, which orchestrate multiple AI agents to perform tasks, demand substantial general-purpose compute power, placing renewed importance on the CPU. The trend has led Bank of America to forecast the data center CPU market will more than double from $27 billion in 2025 to $60 billion by 2030.
NVIDIA's strategy directly targets this need. While its GPUs remain central for model training, the new Arm-based Vera CPU is optimized for the high-speed data processing and orchestration required to "feed" the powerful Rubin GPUs, preventing bottlenecks in agentic systems. This focus on single-threaded performance for orchestration contrasts with the higher core counts of general-purpose x86 CPUs from competitors Intel and AMD.
"Quiet Supply Crisis" Looms with CPU Prices Up 10%
The surging demand for CPUs is already straining the supply chain, creating what some analysts call a "quiet supply crisis." Reports indicate CPU prices have increased by more than 10%, and delivery lead times have stretched to six months. While NVIDIA claims its supply chain is robust, competitors like AMD and Intel have acknowledged supply pressures.
In this competitive landscape, NVIDIA holds just 6.2% of the server CPU market as of Q4 2025, trailing Intel (60%) and AMD (24.3%). However, its strategy of creating an integrated CPU-GPU platform and opening its NVLink interconnect technology to third parties allows it to benefit regardless of who fills the demand. This positions NVIDIA as a central player in the evolving AI hardware ecosystem, set to capitalize on the industry's architectural shift.
