On March 17, Beijing time, NVIDIA CEO Jensen Huang delivered a comprehensive 150-minute keynote at the GTC conference, outlining the company's AI hardware and software technology roadmap and future vision. He projected that flagship computing chips could generate $1 trillion in revenue by 2027 and showcased several innovations, including the Vera Rubin AI factory, LPU inference architecture, co-packaged optics (CPO) switches, and space data center modules. Additionally, the launch of the NemoClaw agent infrastructure signals NVIDIA's commitment to building a full-stack AI ecosystem spanning edge computing, data centers, and orbital scenarios, demonstrating its ambition to accelerate AI industry adoption.
Hot Topic Analysis NVIDIA presented a comprehensive AI strategic layout at GTC 2026. Jensen Huang's forecast that flagship computing chips could achieve $1 trillion in revenue by 2027 underscores the company's confidence in the long-term growth of the AI market. The conference highlighted the Vera Rubin AI factory platform, which integrates multiple core chips with upgrades in computing power, HBM, and optical connectivity. Mass production and shipments are expected to begin in the second half of the year.
The event also emphasized the LPU inference architecture, which leverages Groq's technological advantages and uses large-capacity SRAM for ultra-low latency inference. This is expected to drive new demand in the PCB supply chain, including material upgrades and increased layer counts. In interconnect technology, CPO switches and optical interconnects were focal points. The Rubin platform has begun adopting Scale-Out CPO switches, with 2027 anticipated as a key milestone for CPO volume expansion.
NVIDIA also introduced the NemoClaw agent infrastructure, building a full-stack AI ecosystem from edge and data centers to orbital computing. The company reported shipping 6 million Blackwell GPUs over the past four quarters, with Blackwell and Rubin combined projected to generate over $500 billion in sales. These innovations indicate NVIDIA's focus on establishing a complete AI infrastructure system, spanning from chips to systems.
Investment Logic NVIDIA's GTC 2026 conference showcased strong technological breakthroughs, with the Vera Rubin AI platform and Feynman architecture further solidifying its leadership in AI computing. The newly launched LPU inference chips, integrating Groq's technology, significantly enhance AI inference efficiency, meeting the industry's urgent need to transition from training to application. This is expected to drive both price and volume increases in the PCB supply chain. The NemoClaw agent platform expands enterprise-level AI deployment, enhancing software ecosystem and commercialization potential.
Furthermore, infrastructure innovations such as CPO switches and liquid cooling systems are advancing data interconnection and energy efficiency, providing robust support for computing power growth. North American CSP providers are experiencing high CAPEX growth in 2026, with sustained strong AI demand. The Rubin platform has already sent samples to customers and is nearing mass production. New technologies like Rubin Ultra and Kyber racks are expected to drive growth in HBM, memory chips, and optical module supply chains.
Additionally, Apple's high-volume purchases of memory chips highlight supply shortages, with Micron Technology's performance likely to exceed expectations. AI-related CCL/PCB orders are robust, and mainland Chinese manufacturers are actively expanding production, indicating strong growth prospects. NVIDIA's secured capacity and demand point to robust future growth momentum, with ASIC chips expected to see explosive growth in 2026–2027. Core computing hardware, semiconductor equipment, and Apple's supply chain continue to benefit.
The ChiNext Artificial Intelligence ETF Huabao (159363) focuses on sectors such as communications, computers, and media. Its top ten holdings include XSENS, Zhongji Innolight, TFC Optical, GDS Holdings, Symtech Data, BlueFocus, Kunlun Tech, Ingenic Semiconductor, Wangsu Science & Technology, and Tonghuashun, with a combined weighting of 59.26%.
The STAR Artificial Intelligence ETF Huabao (589520) invests in electronics, computers, and home appliances. Its top ten holdings include Kingsoft Office, Montage Technology, Cambricon, VeriSilicon, Amlogic, CAS Star, Roborock, Fudan Microelectronics, CloudMinds, and BES, with a combined weighting of 67.36%.
The Big Data ETF Huabao (516700) targets computers, communications, and media. Its top ten holdings include iFlytek, Sugon, GDS Holdings, Inspur Information, Unisplendour, Topscomm, Wangsu Science & Technology, Hundsun Technologies, China Great Wall, and Sangfor Technologies, with a combined weighting of 51.67%.
A MACD golden cross signal has formed, indicating strong upward momentum for these stocks.
Comments