Tianjin Robot Town Bets on Domestic Chip Base for Embodied AI
The usual script for a robotics industrial park is to recruit tenants, hand out land, bring in a few complete-machine makers, and wait for output figures to climb. Tianjin's Robot Town took a different route.
At a recent site visit by 160 robotics full-machine and component companies from the Beijing-Tianjin-Hebei region, one disclosed development stood out: Hebei University of Technology is leading a joint laboratory for Xinchuang embodied-intelligence robots, working with Sugon, Hygon, Phytium, Kylin and Tianxing Technology.
The lab aims to become China's first Xinchuang embodied-AI innovation service platform. A robot built on domestically developed Xinchuang products is expected to be released this October.
Regional competition in robotics is shifting from assembly capacity to positioning in underlying chips and operating systems. Whoever first gets a domestic Xinchuang base running with a robot body will hold the entry point to the next phase of the industrial ecosystem. Tianjin chose the harder path.
Tianjin's August action plan for intelligent robotics sets a target of more than 20 billion yuan in core output value by the end of 2028, with annual growth above 15% and a Beijing-Tianjin-Hebei local supply rate above 60%. From January to July this year, the city's above-scale intelligent robotics enterprises posted 7.89 billion yuan in industrial output, up 15% year on year.
Tianxing Technology, a model company inside the town, has already mass-produced embodied-intelligence robots, with monthly capacity of 50 dual-arm units, orders booked through year-end, and a second phase of its 800-million-yuan embodied-intelligence complex under construction.
The town's three functional zones have distinct roles. The innovation ecosystem park focuses on humanoid robots, covering R&D incubation, design and production, pilot testing, data collection, maintenance and talent training. The scenario integration zone spans smart buildings, parks, transportation, campuses, manufacturing and warehousing. The university achievement-transfer base, planned at 20,000 square meters with 10,000 already in use, works with nine nearby universities.
Together the numbers make Tianjin's intent clear: use a Xinchuang base to address dependence on foreign chips and software, use pilot testing and scenario validation to bridge the gap between lab results and mass production, and use university transfer bases to secure talent and projects.
The hardest link is Xinchuang adaptation. The presence of Sugon, Hygon, Phytium and Kylin on the lab's co-building list means domestic chips and operating systems must move from servers and PCs into robots' real-time control and perception-decision loops. The adaptation work is substantial, but once it runs, the robotics supply chain gains an alternative option.
Tianjin's route offers more long-term value than chasing assembly capacity, though it pays off slowly. Whether Xinchuang robots can match mainstream international solutions will get a first answer from the October product. Tianxing's 50 units a month is modest nationally, but orders through year-end show real demand in niche scenarios. A 15% annual growth target for 20 billion yuan is not aggressive by robotics standards, suggesting Tianjin knows the combination of Xinchuang, pilot testing and scenario validation needs time to ferment.
The town's progress is essentially building industrial infrastructure: an exhibition center for trading, the innovation park for incubation, the scenario zone for validation, and the university base for transfer. Whether this infrastructure grows competitive robotics companies depends on the speed of Xinchuang adaptation and the depth of scenario validation. After the October launch of the Xinchuang-based robot, Tianjin Robot Town faces its first stress test.