Nvidia(US:NVDA) EBSCN·2026-03-02 08:15Investment Rating - The report maintains a "Buy" rating for the high-end manufacturing industry, indicating an expected investment return exceeding 15% over the next 6-12 months compared to the market benchmark index [5]. Core Insights - The introduction of Language Processing Units (LPU) by Nvidia is expected to significantly increase the demand for PCB equipment and drilling needles due to the enhanced requirements for PCB area and materials [2][3]. - The LPU technology, designed for AI inference, complements GPUs in AI workflows, marking a shift from general computing to dedicated inference, which is anticipated to drive growth in the PCB sector [2][3]. - The anticipated increase in PCB area usage and the need for advanced materials will lead to a substantial rise in demand for PCB micro-drilling and processing equipment [3]. Summary by Sections LPU Technology Impact - LPU's low latency and high bandwidth characteristics make it suitable for real-time AI interactions, outperforming GPUs in specific tasks while still relying on GPUs for large-scale model training [2]. - The deployment of multiple LPUs will require significantly larger PCB areas compared to traditional GPU architectures, leading to increased material demands and complexity in PCB manufacturing [3]. PCB Equipment Demand - The integration of LPU technology is expected to create a surge in demand for high-precision drilling and exposure equipment, as well as advanced PCB assembly devices [4]. - Companies to watch in the PCB manufacturing sector include Dazhong CNC, Inno Laser, and others involved in high-precision drilling and assembly processes [4]. Advanced Packaging Requirements - Nvidia's proposed Prefill-Decode Disaggregation technology aims to optimize the deployment of GPUs and LPUs, which will necessitate advanced packaging solutions and higher precision in PCB assembly [3]. - The combination of GPU and LPU architectures will likely increase the demand for high-precision assembly equipment in the PCB sector [3].