Ubtech, a humanoid robotics company valued at 60 billion yuan last year, has decided to make significant investments in Shanghai. This year, the company will launch six projects in the city, including a commercial humanoid robot smart manufacturing base, an embodied intelligence research institute, a global brand headquarters, a pilot zone for integrated data and computing, a startup incubation platform, and an industrial fund. This strategic choice is not coincidental. Data from the Municipal Commission of Economy and Informatization reveals the city's strong momentum in developing new quality productive forces: by 2025, 394 large-scale artificial intelligence enterprises in Shanghai achieved an industrial scale exceeding 637 billion yuan, representing a year-on-year growth rate of 39.5%. The city's industrial investment increased by 20% year-on-year, surpassing the national average by 17.4 percentage points. In 2025, Shanghai's regional GDP reached 5.67 trillion yuan, contributing over 4% of the national GDP while occupying less than 0.1% of the country's land area. For the current year, Shanghai has set its GDP growth target at around 5%. Where does this confidence originate? A key term首次 included in this year's government work report—intelligent economy—may provide the answer, serving as both a critical move in transitioning between old and new growth drivers and a natural convergence point of years of strategic planning.
The race for computing power requires frequent communication and continuous decision-making between drivers and teams. On the Tianmen Mountain circuit with its 99 turns, the fastest lap time for a human driver is 7 minutes and 38 seconds, while AI's best time is 16 minutes and 10 seconds. How can AI close this gap of over 8 minutes? Stanford Professor and "AI godmother" Fei-Fei Li stated that the next era of AI is "spatial intelligence," meaning AI must enter the physical world to unlock greater value. Recently, the global physical intelligence platform Hitch Open announced a partnership with Zhiyuan Robot to host a robot table tennis challenge, issuing an open invitation to the global automotive and robotics industries to tackle ultimate physical scenarios at Tianmen Mountain, thereby advancing embodied intelligence. The difficulty of AI in the physical world lies in its nature; it is not a statistical "fill-in-the-blank" exercise but requires efficiently calculating a unique, stable, and correct "solution" under complex constraints. Not long ago, Shanghai-based company Shanshu Technology, which specializes in AI decision engines and intelligent computing, established itself in Shanghai. The company has sought "optimal solutions" in physical world scenarios involving millions of variables and hundreds of thousands of constraints, such as China Southern Airlines' "in-flight engine shutdown" incidents and China Southern Power Grid's peak summer demand management. Having provided decision optimization for hundreds of leading enterprises across more than 20 industries, they aim to foster a related AI decision-making industry in Shanghai.
As the intelligent industry surges, how will the computing power bottleneck be addressed? Shanghai has plans in this area as well. Nine years ago, MIT PhD Shen Yichen founded Xizhi Technology in Shanghai, focusing on optical computing and optical interconnection chips. Shen believes the industry is on the verge of a breakthrough, predicting that within the next 3 to 5 years, all major chip manufacturers nationally and globally will adopt optical interconnection, as only "light" can meet future computational demands. Over nine months, engaging in 672 negotiation rounds with enterprises—averaging nearly 20 rounds per week—and undertaking more than 20 investment promotion trips abroad, averaging one trip every two weeks: this was the work schedule of Wang Tao, Deputy General Manager of Shanghai Zhangjiang Embodied Intelligence Robot Co., Ltd., since April of last year. Currently, Pudong has gathered over 100 key enterprises spanning the entire embodied intelligence industrial chain, with nearly half located within the 4.2 square kilometer Zhangjiang Robot Valley. In January of this year, the world's first fully bionic embodied intelligence robot, "Moya," debuted in Zhangjiang Robot Valley. Breaking away from the traditional "steel image" of robots, Moya's body is covered in silicone, with a temperature close to that of the human body and a delicate texture. Internally, 25 precision drive components work in coordination, combined with micro-expression algorithms, allowing for the natural display of emotions like joy, anger, sorrow, and happiness. This "response with warmth" is transforming human-robot interaction from "tool usage" to "companion coexistence."
At the 2026 Appliance & Electronics World Expo, the A1 robot, independently developed by Tashizhihang, set a new Guinness World Record for "the most sub-millimeter wire harness assemblies completed by a robot in one hour," achieving 105 effective assemblies. Wire harness assembly, often called the "Goldbach conjecture" of industrial automation, has long relied on manual labor in sectors like 3C electronics and automotive manufacturing. The A1 robot, leveraging the company's self-developed general embodied large model, repeatedly performed high-precision assembly tasks during 60 minutes of continuous operation, meeting precision and force control requirements each time. While it's difficult to predict the exact moment of a "nuclear explosion point" in innovation, the convergence of resource elements is a prerequisite for transformative breakthroughs. Morgan Stanley recently raised its 2026 sales forecast for humanoid robots in China, expecting sales to reach 28,000 units this year, a 133% year-on-year increase. As the city that nurtured the world's leading company in humanoid robot shipments, Shanghai, shouldering a major role, is expected to see a strong "dragon head rise" in its robot sales curve this year.
When the intelligent agent architecture OpenClaw, nicknamed "Crayfish," surpassed the 35-year-old Linux kernel to top GitHub, the world's largest software project hosting platform, in just 100 days, it signaled that a paradigm shift in the information age has arrived, suggesting that all software is worth redeveloping with intelligent agents. Amid this wave of intelligent agents, cost-effective "Shanghai-born" large models are gaining prominence. In the first week of March, weekly usage of Chinese large models surged 34.9% month-on-month, reaching 419 billion tokens, once again exceeding usage in the United States. During the same period, Shanghai AI companies Xiyu Technology and Jieyue Xingchen ranked in the top three on both the daily and monthly usage charts for OpenClaw. "The rapid rise of Shanghai's AI sector is inseparable from its early anticipation and forward-looking layout regarding global AI technology trends," said Zhong Junhao, Secretary-General of the Shanghai Artificial Intelligence Industry Association. The built-in AI process interpretation capability of "Shanghai-born" large models leads the nation, specifically adapted to the development needs of intelligent agents. Price has always been a key factor driving product adoption. When one-third of the nation's AI talent combines with an input cost of $0.3 per million tokens, a "chemical reaction" is inevitable, encouraging many innovative individuals to explore and experiment, leading to the落地 of more new applications. The prediction made years ago by OpenAI CEO Sam Altman—that one person plus ten thousand GPUs could build a billion-dollar "unicorn"—is now materializing in Shanghai's various industrial parks. A large number of "one-person companies" (OPCs) are gathering, with many founders yet to even graduate. Through systematic and forward-looking support services, Shanghai enables these individuals to command "armies" under the spotlight, creating entrepreneurial legends befitting this era.
At the start of the 15th Five-Year Plan period, Shanghai has charted a new course focusing on five new growth areas—low-altitude economy, commercial aerospace, embodied intelligence, bio-manufacturing, and intelligent terminals—collectively targeting a scale in the trillions of yuan. This is not merely a numerical sum but an intelligence-guided effort to weave point breakthroughs into a cohesive network, allowing innovative elements to undergo fission through collision and爆发 through integration.
Comments