Tashi Zhihang CEO Says China Is Moving Embodied AI Into Large-Scale Deployment

The session drew more than 220 entrepreneurs, representatives of industry organizations and investment institutions from 24 countries and regions. Chen shared Tashi Zhihang's practice and thinking on using embodied intelligence to empower global industrial transformation.

Global industry is shifting from traditional automation to physical AI. Labor force changes and process complexity leave auto final assembly, precision wire harness assembly and fine parts feeding highly dependent on manual work, making them a 'last mile' that automation upgrades have struggled to cross. At the same time, the manufacturing sector's large volume of real work data and strict engineering standards provide a foundation for embodied intelligence to enter the real economy.

Chen said embodied intelligence cannot remain in the laboratory and must be tested on real production lines. 'China has the world's richest manufacturing scenarios and application demand, which provides irreplaceable soil for the training, validation and iteration of robot models,' he said. He added that China's long-term industrial ecosystem is allowing the country to push its embodied intelligence into large-scale deployment. Shanghai, with top interdisciplinary AI talent, a precision supply chain and an open cooperation ecosystem, is an important highland that can carry and amplify this industrial advantage.

Born in Shanghai and facing the global market, Tashi Zhihang says it has completed a three-stage leap covering foundation model breakthroughs, hardware iteration and optimization, and large-scale deployment.

The company says it created a human-centered data paradigm and collected more than one million hours of high-quality real human operation data. It independently developed the AWE embodied intelligence foundation model, which it says solves industrial problems in fine manipulation, flexible object handling, long-horizon task planning and generalization, giving robots a 'general brain.' Relying on the Yangtze River Delta's 'one-hour industrial circle,' it built hardware and bodies 'born for AI,' including its self-developed A-series wheeled robot, T-series biped robot and 21-degree-of-freedom quasi-direct-drive DexHand. With support from Shanghai's special policies, it accelerated large-scale production line validation of embodied intelligence technology, allowing robots to 'really work.'

Shanghai's leading AI talent advantage has provided core support for Tashi Zhihang's rapid growth, according to the article. The company has assembled a top R&D team in which more than 80% of members hold master's or doctoral degrees. Its five-member founding team consists of founder and CEO Chen Yilun, chairman Li Zhenyu, co-founder and chief scientist Ding Wenchao, co-founder and chief architect Chen Tongqing and co-founder and chief strategy officer Vincent. Team members generally have rich experience in autonomous driving and artificial intelligence. They bring leading capabilities in embodied model building, body R&D, software-hardware integrated product mass production and engineering implementation of large complex systems, as well as deep insight into target industry pain points, the ability to identify high-value application scenarios and deliver complete solutions.

Chen said the company will continue to expand its R&D team, accelerate the layout of production bases and comprehensively improve robot delivery capacity. It will also actively build a 'China R&D, global-facing' business layout, work with partners at home and abroad to build an industry ecosystem, contribute more 'embodied intelligence solutions' to global industrial upgrading and transformation, and strive to become 'an important enabler of global industrial intelligent transformation.'

Editor's Summary Tashi Zhihang CEO Chen Yilun used a Shanghai event to argue that China's manufacturing data and supply chain can move embodied AI from laboratories to factory floors. The company says it has completed foundation-model, hardware and production-line validation steps, and it plans to expand R&D and production capacity. The account is based on company-provided material published by QbitAI.