Bank of China International: Artificial Intelligence Enters the Era of "Physical AI", Supply Chain Inventory is Expected to Speed Up

China International Bank released a research report stating that at the 2026 CES exhibition, Huang Renxun announced the arrival of the "Physical AI" era. Siasun Robot&Automation and autonomous driving are expected to become ideal carriers for "Physical AI". The Rubin platform has entered full production, with core raw materials expected to reach a key milestone from "01", and supply chain stocking is also expected to accelerate. The bank believes that AI has gradually extended from the digital world to the real world, with Siasun Robot&Automation, autonomous driving, and edge AI hardware expected to become the market carriers for AI, officially entering the era of commercialization. The main points of China International Bank are as follows: Artificial intelligence enters the "Physical AI" era According to TechWeb's report on January 6th, Huang Renxun delivered a speech at the 2026 International Consumer Electronics Show (CES), announcing that artificial intelligence has entered the "Physical AI" era. The concept of "Physical AI" emphasizes that AI needs to integrate real physical dynamics such as gravity and friction to precisely execute complex tasks. Its core pillars include the Newton physics engine, the Cosmos foundation model platform, and the GPU+LPU hybrid architecture. Rubin platform goes into full production According to Tencent Technology, NVIDIA has revealed complete data for the Rubin platform for the first time. Compared to Blackwell, Rubin GPU NVFP4's inference performance has increased to 50PFLOPS (5X), NVFP4 training performance has increased to 35PFLOPS (3.5X), HBM4 memory bandwidth has increased to 22TB/s (2.8X), and single GPU NVLink interconnect bandwidth has increased to 3.6TB/s (2X). The Rubin architecture uses a collaborative design of Vera CPU, Rubin GPU, NVLink 6, ConnectX-9, BlueField-4, and Spectrum-6, thereby achieving a revolutionary decrease in inference costs. According to TechWeb, Rubin's order size has reached 300 billion US dollars, Rubin has entered full production stage, and products based on this platform will be launched through partners in the second half of 2026. Siasun Robot&Automation and autonomous driving are ideal carriers for "Physical AI" According to TechWeb reports, Huang Renxun referred to Siasun Robot&Automation as "the ultimate form of AI" and announced a deepening cooperation with Tesla to provide the AI brain for the Optimus humanoid Siasun Robot&Automation. Optimus has completed over 90% of training through the Omniverse digital twin platform, with a self-driving rate of 85%, far exceeding the industry average. Its production plan includes: achieving 50,000 units in the first quarter of 2026, with costs below $20,000; by the end of 2026, capacity will increase to 100,000 units, covering all processes from parts sorting, assembly, to quality inspection. At the same time, Huang Renxun believes that autonomous driving is the trend of the automotive industry, with NVIDIA releasing the Alpamayo series AI models to support L4 autonomous driving. Core raw materials for AI are expected to reach a key milestone from "01" The bank estimates that the Compute Tray/Switch Tray/Midplane/CPX PCB and CCL solutions corresponding to the NVIDIA Rubin server will be upgraded to M8/M8.5/M9/M9 solutions, with the M9 solution possibly using a material combination of high-frequency high-speed resin+HVLP4/5 copper foil+Q cloth, while the M8.5 solution may use a material combination of high-frequency high-speed resin+HVLP4 copper foil+Low-Dk second-generation cloth. The bank believes that the Rubin Ultra server is expected to adopt a M9 resin+high-end HVLP copper foil+Q cloth orthogonal backplane solution. The bank expects the upstream supply chain of Rubin servers to start stocking up in the first half of 2026, at which point the core raw materials of M8.5 and M9 PCB/CCL are expected to reach a key milestone from "01". Supply chain stocking is expected to accelerate The bank believes that AI has gradually extended from the digital world to the real world, with Siasun Robot&Automation, autonomous driving, and edge AI hardware expected to become the market carriers for AI, ushering in the official commercialization stage. At the same time, "Physical AI" requires substantial computational power, data, and other resources, which presents higher requirements for the construction of AI computational infrastructure. The mass production of the Rubin platform means a further improvement in AI computational power, data transfer efficiency, and inference costs, and is expected to stimulate the progress of the upstream supply chain. Investment recommendations The bank recommends focusing on NVIDIA's supply chain: [PCB] Victory Giant Technology, Wus Printed Circuit, Shennan Circuits; [CCL] Shengyi Technology; [Q cloth] Hubei Feilihua Quartz Glass, Sinoma Science & Technology; [resin] Sichuan Em Technology. Key risks facing the rating Overheating demand in the AI market leading to industry bubbles. Excess capacity on the supply side leading to price declines. Technological changes leading to the obsolescence of existing products.