ResNet Author Ren Shaoqing Launches Embodied AI Startup — Already a Unicorn, Still at NIO

QbitAI has learned and confirmed that Ren Shaoqing, the renowned AI scientist, ResNet co-author, and NIO's senior vice president and head of intelligent driving, has formally founded a company to pursue embodied intelligence robotics. According to LatePost, NIO CEO William Li announced the news at an internal all-hands meeting for the intelligent driving business.

Ren's new company focuses on physical AI foundation models and embodied intelligence, and NIO has confirmed a strategic investment plus business cooperation. The company has reportedly completed registration with a unicorn-level valuation (above $1 billion), though its name and financing details remain undisclosed.

Notably, Ren's entrepreneurial model is not yet fully clear — although the robotics venture is confirmed, he remains employed at NIO. People close to NIO's management said the automaker has long focused on frontier AI technology, balancing in-house R&D with open cooperation, and continues to support innovative companies in embodied intelligence, algorithm models, smart hardware, new energy, and compute infrastructure.

NIO's internal view is that intelligent driving is one of the most mature large-scale deployments of AI in the physical world, sharing continuous underlying capabilities with embodied intelligence in perception, prediction, planning, control, world models, and reinforcement learning — a key reason talent has been flowing from autonomous driving into robotics over the past two years.

NIO's deliberate ambiguity around the move is telling: as the architect of NIO's intelligent driving transformation, Ren leaving outright would be bad news for the momentum the company currently enjoys; but incubating the venture as an internal innovation would expose NIO, which has just reversed its finances, to the high-cost, long-cycle nature of robotics.

NIO has long signaled its robotics inclination. In 2025, when financials were under pressure, both Li Bin and Ren told QbitAI that NIO's next act in AI-plus-automotive leans toward robots rather than Robotaxi. In July 2024, Ren unveiled NIO's world model NWM (NIO World Model) at NIO's tech day; it began mass deployment in 2025 and was upgraded this year to a world model plus supervised fine-tuning and closed-loop reinforcement learning architecture.

Ren has repeatedly argued that world models are a shared foundational paradigm for autonomous driving and robotics, and that the spatiotemporal cognition, long-horizon reasoning, and closed-loop optimization accumulated on the vehicle side can transfer directly to physical AI and embodied scenarios — the technical foundation for his new venture.

Born in 1988 in Bengbu, Anhui, Ren earned his PhD in 2016 under a joint program between the University of Science and Technology of China and Microsoft Research Asia, supervised by Sun Jian. He is one of the four authors of ResNet, which won the CVPR 2016 Best Paper Award and has been cited more than 340,000 times, and the first author of Faster R-CNN. After co-founding Momenta, he joined NIO in 2020, and since September 2025 he has also served as a chair professor and director of USTC's Institute of General Artificial Intelligence.

What to watch: the new company's name, funding, and product roadmap remain undisclosed, and how Ren's entrepreneurship-while-employed model plays out inside and outside NIO will determine whether embodied intelligence gains another world-class AI scientist.