Chengwu Robotics raises angel round from Hecheng Technology to scale industrial embodied-AI for Foxconn and beyond
What's the deal? Chengwu Robotics, a Shenzhen-based industrial robotics startup founded in 2025, has closed an angel round of financing. The strategic investment came from Hecheng Technology, a Taiwanese leader in industrial automation and intelligent robot solutions. Huajun CapitalDealroom has a profile for this one. Try Dealroom → served as exclusive financial adviser.
The company builds customised robot solutions for factories, combining hardware, software, AI models, and deployment services into a single package. It has already completed over a dozen industrial projects and counts Foxconn among its clients.
In the first half of 2025, cumulative revenue surpassed 20 million yuan (roughly $2.8M). The team has also secured top-tier authorisation from eight leading robot body manufacturers to provide scenario-specific customisation.
Why now? Industrial robotics is shifting from rigid, pre-programmed systems to AI-driven "embodied intelligence" — robots that can perceive, reason, and adapt. But factory floors are full of non-standard environments that demand bespoke solutions, creating a gap between general-purpose robot hardware and real-world deployment.
Chengwu is positioning itself in that gap. Rather than building robot bodies from scratch, it integrates existing hardware with its own end-effectors, algorithms, and AI models to deliver turnkey solutions quickly.
The founding team brings deep domain expertise. CEO Huang Jinlong has over a decade of experience in full-stack robot R&D and industrialisation. Co-founder Dr Shan Yuhu previously led core robotics R&D at Tencent, XPeng, and Meituan, with specialisms in multi-modal large models and 3D perception.
What could go wrong? Building a scalable business on customised industrial solutions is inherently tricky — each deployment risks becoming a one-off project rather than a repeatable product. The company is betting that its standardised "technical product package" and its Vision-Language-Action (VLA) large model can solve this, but vertical AI models for manufacturing remain largely unproven at scale.
Data is another bottleneck. Chengwu has developed its own data collection hardware and teleoperation system to gather training data from real production lines, but competing with better-funded rivals in the embodied AI space will require sustained investment.
The signal: Chengwu's early traction reflects two converging trends. First, China's manufacturing giants — Foxconn chief among them — are actively seeking AI-powered automation to handle complex, irregular tasks that traditional robots can't manage. Second, a new wave of startups is emerging that treat the robot body as a commodity and focus on the intelligence layer instead.
The company's ambition to build a "one brain, multiple forms" model — a single AI system that generalises across different robot types and tasks — mirrors a broader industry bet that foundation models will do for robotics what large language models did for software. If Chengwu can prove this works in live factories, it could carve out a valuable niche between hardware makers and end users.
Read more: eu.36kr.com
