Copper light parallel has become an industry consensus, Longbridge is laying out in the AI computing power field
I'm LongbridgeAI, I can summarize articles.NVIDIA CEO Jensen Huang proposed the consensus of "copper light parallelism" in the field of AI computing power at Computex, advocating for the priority use of low-cost copper cables. EWPT has already laid out in this field, mass-producing high-speed cables, connectors, and other AI data center infrastructure hardware, and has acquired equity in Weixian Technology to enhance its competitiveness
On June 2, 2026, during the Computex exhibition in Taipei, Taiwan, Jensen Huang, CEO of NVIDIA, and Matt Murphy, CEO of Marvell, jointly proposed a new logic for the AI industry: as AI models move towards a massive "Agent" era, AI competition enters its second half. After breakthroughs in computing power and memory bottlenecks, "connectivity" will become the decisive battlefield. The core principle in choosing connectivity solutions is "use copper where possible, and only use optical when necessary." Companies with the capability for integrated copper and optical solutions are expected to occupy a more advantageous position in the industry chain competition.
According to reports, the listed company EWPT (300115.SZ) has achieved mass production and delivery of hardware for AI data center infrastructure, with products mainly including high-speed cables, connectors, optical module housings, splitters, quick connectors, and other core components for liquid cooling.
Copper and optical parallelism becomes industry consensus, high-speed copper cables open up incremental space
Regarding the industry's keen interest in the timeline for the transition from copper cables to optical fibers, Jensen Huang stated, "Use copper cables where possible, and only use optical devices where necessary."
He further explained that copper cables have physical limits in bandwidth and transmission distance. Before breaking this boundary, copper cables are a simple, low-cost, and practical choice; once the critical point is crossed, optical fibers will take over the expansion needs between racks, data centers, and across data centers. Huang predicts that in the next 5 to 10 years, copper cables will still be used extensively, along with a massive amount of optical devices.
According to statistics and analysis from Frost & Sullivan, the global precision manufacturing market for AI data center infrastructure is expected to grow from RMB 24.1 billion in 2021 to RMB 193.3 billion in 2025, with a compound annual growth rate (CAGR) of 68.3%. With the continued penetration of AI applications, the global precision manufacturing market for AI data center infrastructure is projected to grow from RMB 265.2 billion in 2026 to RMB 885.9 billion in 2030, with a CAGR of 35.2%.
In recent years, EWPT has been optimistic about the AI server industry and the high-speed communication module business based on copper connections, actively laying out in AI computing power-related fields. In 2025, the company acquired 51% of Shenzhen Weixian Technology Co., Ltd., enriching its high-speed copper connection module product line and providing products for leading overseas connector customers and AI server clients.
According to the Hong Kong stock prospectus, EWPT's high-speed server cables have a direct transmission rate of over 96%, ensuring that the vast majority of signals experience no significant attenuation during transmission, thereby ensuring the consistency of data transmission quality. In terms of high-speed backplane connectors, as bandwidth demands increase, product designs support upgrades to 112G. Additionally, the company's MCIO connectors meet the signal integrity requirements of the PCIe 6.0 standard, providing leading industry standard high-speed interfaces for connection processors, storage, and network components used in server mass production.
In terms of customer expansion, EWPT has been certified by leading North American and Taiwanese manufacturers, providing server data transmission copper cables for AI brands. In the core components for liquid cooling, it supplies core parts (such as splitters and quick connectors) for leading temperature control system companies, with products ultimately applied in the data centers of globally renowned AI companies According to reports, Changying Precision's AI data center connection components and structural parts will start generating revenue in 2025, while the core liquid cooling components will begin to generate revenue in 2026.
According to the Q1 2026 report, Changying Precision has been working on many new projects this year, continuously increasing investment in the field of embodied intelligence, and actively expanding new businesses such as server liquid cooling quick connectors and high-speed copper cables. Against the backdrop of "copper light parallel" becoming an industry consensus, the company's emerging technology hardware business is expected to contribute to performance growth.
Intelligent Agents Reshape Industry Ecology: AIPC and Embodied Intelligence Open New Opportunities
On June 1, at the Taipei GTC conference, NVIDIA CEO Jensen Huang officially announced the launch of the first RTX Spark super chip designed specifically for personal intelligent agents. He introduced that based on the RTX Spark and the new PC with Microsoft Windows, users only need to state their needs, and the rest can be autonomously completed by the computer.
RTX Spark integrates NVIDIA's CUDA, RTX, and NVIDIA AI platforms into a single super chip, allowing local intelligent agents, cutting-edge models, creative workflows, and RTX games to run smoothly on a single laptop, making it more portable, thinner, and with better battery life.
In other words, in the future, those who can work with computers will not only be humans but also AI intelligent agents. This marks the reshaping of the PC ecosystem by Agent intelligent agents, ushering in a brand new era of personal AI computing.
In terms of hardware implementation, ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI will launch lightweight Windows laptops equipped with RTX Spark this fall, featuring all-day battery life and high-end screens, along with compact desktop hosts. Related products from Acer and Gigabyte will be launched subsequently.
As AI PCs and other AI intelligent agent terminals accelerate their implementation and enter daily office and life scenarios, the upstream of the industrial chain is also welcoming structural development opportunities.
In the field of precision components for intelligent terminals and electronic products, Changying Precision has established long-term stable cooperative relationships with leading industry brands. The company seizes the opportunity of AI hardware iteration, continuously developing key components for laptops and wearable products based on AI technology, especially focusing on the development of metal and non-metal structural parts that meet the requirements of lightweight and high strength, achieving significant breakthroughs. Key projects have successfully entered mass production with domestic and overseas clients, and new projects are continuously being introduced. As an industry chain enterprise with technical reserves and customer base, it is expected to be among the first to share the growth dividends brought by this round of AI hardware transformation.
Additionally, it is worth mentioning that in the physical extension of AI intelligent agents—embodied intelligence, Changying Precision has comprehensive capabilities from parts to complete machine assembly. The company's products cover core high-precision components such as dexterous hand transmission parts, core transmission parts of rotary actuators, core transmission parts of lead screws, large-size trunk transmission parts, and skeletal structural parts. The company has also established testing and verification capabilities for embodied intelligence, and through the operational data and working condition feedback accumulated during assembly and testing, it can effectively feed back into the design optimization and manufacturing process improvement of components, thus forming a positive cycle of "components → assembly → data feedback → component iteration." In 2025, Changying Precision delivered approximately 689,000 precision components for humanoid robots, with overseas customers accounting for about 80% of the shipments. From January to April 2026, the company's total shipment of robot components has exceeded 380,000 units, continuing a good growth trend.
As the end applications of humanoid robots extend from industrial manufacturing to commercial services, home assistance, and other scenarios, as well as the accelerated commercialization of quadruped robots and industrial collaborative robots, the company is expected to continue benefiting from the rapid expansion of the embodied intelligence industry.
Currently, Changying Precision has established a global production network covering China, Southeast Asia, North America, and Europe, with flexible global delivery capabilities and regional supply chain response advantages, providing solid capacity support for deeply serving leading customers both domestically and internationally and undertaking incremental orders. In the future, with the rapid development of embodied intelligence and emerging technology fields, the demand for core global components is expected to continue to be released, and Changying Precision is likely to take this opportunity to further expand its product matrix, increase market share, and continue to benefit from AI hardware upgrades and the growth of the robotics industry
