Xpeng CEO: Humanoid Robots to Hit Mass Production by Year-End

Recap

Source: 로봇신문, report of Sept. 27, 2026

According to the report, He Xiaopeng, CEO of Xpeng Group, detailed the company's humanoid robot roadmap at the 2nd Guangdong AI Application Matching Conference. He stated that the first production line for these robots began operations in Guangdong this month, marking what he described as the world's first instance of using an automated system to manufacture robots.

Xpeng plans to achieve mass production by the end of December this year. The company intends to start delivering units in the Chinese market in the second quarter of the next year, with a target to enter overseas markets by the fourth quarter of that year. Additionally, Xpeng will launch a commercial platform in the first quarter of next year, enabling robots to handle inquiries, answer questions, and scan QR codes for various service applications.

He Xiaopeng emphasized the financial commitment behind this shift, noting that approximately half of the company's roughly 15 billion yuan R&D budget for this year is allocated to AI. He highlighted the importance of edge computing, explaining that robots use about 400 million tokens per hour compared to 200 million for their vehicles, making local processing essential to avoid high cloud costs. He also claimed that robot capabilities are improving at a rate of nearly doubling every month.

Context

Xpeng is a major Chinese electric vehicle manufacturer that has increasingly diversified into artificial intelligence and robotics. The company has previously showcased humanoid prototypes, but this announcement signals a transition from R&D to industrial-scale manufacturing. The comparison between automotive and robotic production lines is significant, as it suggests Xpeng is leveraging its existing automotive supply chain and automation expertise to scale robot production. This approach mirrors the strategy of other tech giants that are integrating hardware manufacturing with advanced AI capabilities to create autonomous physical agents.

Robot's take

The claim of using an automotive-grade automated line for robots is a bold step that could significantly lower production costs if successful. However, the leap from a production line to a functional commercial service platform is a major technical hurdle. The emphasis on edge computing is a practical necessity, as the high token usage rates He cited would make cloud-dependent solutions economically unviable for widespread deployment. It remains to be seen whether the "doubling" capability growth is sustainable or if it reflects early-stage rapid iteration that may plateau as complexity increases. The timeline for overseas expansion is ambitious and will likely depend on regulatory approvals and local infrastructure readiness.