Xpeng Robotics Raises $900 Million to Build Better Humanoid Robots

Xpeng Robotics has secured $900 million in funding to develop advanced humanoid robots. This investment, led by IDG Capital, includes participation from tech giants Tencent and Alibaba. The valuation of the company has soared past $6.3 billion, marking it as the largest private funding deal in China's "embodied AI" industry. This sector focuses on AI integrated within physical robots, as opposed to purely digital AI.

For context, just last month, TARS Robotics secured $455 million, which was previously the largest private funding in the same sector. The $900 million will be allocated across five key areas. These include enhancing the physical design and capabilities of robots, training AI using real-world data, gathering human movement data for learning purposes, constructing mass-production factories, and expanding the business globally beyond China.

Xpeng's first humanoid robot, named IRON, is set for mass production by the end of 2026. The company intends to deploy these robots in its own facilities, including retail stores for assisting customers and industrial campuses for factory tasks. Commercial sales and deliveries are expected to commence in China and other international markets starting in 2027. CEO He Xiaopeng personally spearheads the robotics initiative, signaling its importance to the company, alongside its electric vehicle business.

This significant investment in humanoid robotics can be seen as a natural progression from Xpeng's expertise in electric vehicles. The company already possesses the necessary technological foundation and supply chain, enabling it to scale up production of complex hardware and manage related software. With this move, Xpeng is positioning itself to potentially become the first to commercially sell humanoid robots at scale, potentially revolutionizing the operations of stores and factories.

However, the real test will be to demonstrate the affordability and practicality of these robots for businesses, and ultimately, to have them tested with actual human operators.