Unitree Says It Won’t Become a Robot Integration Company

Key Takeaways

Show all

• Unitree industrial solutions lead Xie Yipeng says the company does not intend to become an engineering contractor handling project delivery, according to Gasgoo’s account of his September 15 remarks.
• His comments suggest a division of work: Unitree develops robots and core components, while partners adapt them to particular applications.
• That approach leaves a crucial question for factory adoption: who makes the complete system reliable and economical enough to use every day?

Unitree wants its robots to find industrial uses, but it does not intend to take on every part of putting them to work.

Speaking at a Gasgoo industry seminar on September 15, Xie Yipeng, Unitree’s head of industrial scenario solutions, said the company would not become an engineering company or undertake project delivery, according to Gasgoo’s report. He described an approach open to third-party tools, sensors, communications and additional power systems, while Unitree concentrates on developing robots and their core components.

The remarks draw a boundary around its ambitions. Supplying a robot and delivering a working factory installation are different businesses, with different responsibilities.

Who turns the robot into a working system?

Our reading is that Xie is describing a role for application partners and integrators. That does not amount to a newly announced partner program, nor does it establish how responsibilities would be divided in individual contracts.

For an industrial customer, the machine itself is only part of the purchase. Someone must connect it to the workflow, select its tools, establish safe operating procedures and demonstrate that it can complete the job consistently. The commercial question is who owns that work—and who supports the installation when conditions change.

That distinction fits the challenges Unitree CEO Wang Xingxing described at the World Robot Conference. Wang said the company’s robots could perform simple assembly, but remained slower than people and required retraining for new tasks. Those limitations had held back large-scale factory rollout.

His comments concerned technical readiness. Xie’s address adds a business question: how will the work of turning a capable platform into a useful application be distributed?

Hardware progress still needs an application

The recently launched G1+ shows Unitree continuing to improve the platform. It adds a moving neck, stronger motors and revised sensing hardware, with a listed base price of US$15,000 before tax and shipping. Those changes do not, by themselves, demonstrate a production-ready factory workflow.

Xie also pointed to a data problem: information collected by factories over time does not necessarily have the formats and relationships needed to train AI. He nevertheless forecast faster humanoid development over the next two to three years. These were assessments of the industry’s direction, rather than a dated commitment to scaled deployment.

For prospective customers, the next milestone is therefore more specific than another hardware launch: a complete application that meets their requirements for output, reliability and cost, with clear responsibility for keeping it working. Xie’s remarks help explain where Unitree expects to focus its own efforts within that process.