Dr. Shaoqing Ren, a legendary AI researcher and author of ResNet, has launched a robotics startup focused on embodied intelligence and physical AI foundation models that achieved a valuation exceeding $1 billion on its first day of registration. The move marks a significant moment in the race to commercialize physical AI, as one of deep learning's most influential scientists pivots from autonomous driving to robots.

Who Is Ren Shaoqing and Why Does His Move Matter?

Ren Shaoqing is no ordinary entrepreneur. His 2016 paper on ResNet fundamentally solved the problem of gradient vanishing in deep neural networks, making it possible to train extremely deep AI models. That work has been cited over 340,000 times and won the Best Paper Award at CVPR 2016, one of computer vision's most prestigious conferences. He is also the first author of Faster R-CNN, an object detection algorithm widely used in autonomous driving, security, and medical imaging.

Before launching his robotics venture, Ren served as Senior Vice President and Head of Intelligent Driving at NIO, the Chinese electric vehicle manufacturer. He won the 2023 Future Science Prize and currently directs the Institute of General Artificial Intelligence at the University of Science and Technology of China. His departure from pure autonomous driving to embodied AI signals where the industry's top talent believes the next frontier lies.

What Makes This Startup Structure Unusual?

The arrangement is unconventional. Ren has founded an independent company to focus on physical AI foundation models and embodied intelligence, yet he remains employed at NIO. NIO CEO Li Bin announced the move at an internal all-hands meeting, confirming that NIO will make strategic investments and pursue business cooperation with Ren's new venture. The company name and specific financing details remain undisclosed, though the valuation has already crossed the $1 billion unicorn threshold.

This structure resembles what one source colorfully described as "getting a divorce but still living together." The ambiguity appears intentional. On one hand, losing Ren entirely would damage NIO's momentum in intelligent driving. On the other hand, spinning out the robotics business as a separate entity shields NIO from the high investment costs and long development cycles that characterize embodied AI. The approach mirrors Xpeng Motors' recent strategy of spinning out its robotics unit, IRON, which raised $900 million at a $6.3 billion valuation.

How Do World Models Connect Autonomous Driving to Robotics?

The technical bridge between Ren's autonomous driving work and his new robotics focus is world models. A world model is an AI system that learns to predict how the physical world behaves, enabling both vehicles and robots to plan and act intelligently. NIO began developing world models internally in the second half of 2023. In July 2024, Ren presented NIO's World Model (NWM) at the company's Tech Day event. The model entered mass production in 2025 and was upgraded this year to incorporate supervised fine-tuning and closed-loop reinforcement learning, a technique that lets AI systems improve through trial and error.

Ren has emphasized repeatedly that world models represent a shared foundational paradigm for both autonomous driving and robotics. The spatiotemporal cognition, long-sequence reasoning, and closed-loop optimization capabilities developed for vehicles can transfer directly to physical AI and embodied intelligence scenarios. This technological continuity explains why Ren is well-positioned to launch a robotics company and why NIO views the move as a natural extension of its AI strategy.

What Are the Key Factors Behind This Timing?

• Talent Migration: Over the past two years, more and more researchers and engineers from autonomous driving and intelligent driving fields have moved into robotics, drawn by the technology transfer opportunities and the massive market potential of embodied AI.
• NIO's Financial Turnaround: NIO faced financial pressure in 2025 but has recently achieved a financial turnaround. Both CEO Li Bin and Ren made clear that for the next phase of AI plus automotive business, NIO prefers investing in robots rather than Robotaxi services.
• Manufacturing Synergies: NIO's existing business scenarios and supply chain manufacturing capabilities are naturally suited to the closed-loop implementation of robots, giving the new venture a competitive advantage.

What Does This Mean for the Broader Physical AI Race?

Ren's move underscores how the embodied AI industry is attracting top-tier talent and capital. Just days before this announcement, Xpeng's robotics unit raised $900 million from IDG Capital, Gaorong Ventures, Tencent, and Alibaba, valuing IRON at $6.3 billion. That round represents China's largest single-round private financing in embodied AI. The IRON humanoid robot features 76 degrees of freedom across its body and 21 in each hand, runs on three Turing AI chips delivering up to 2,250 TOPS (tera operations per second), and targets mass production by the end of 2026.

"I believe XPENG will not only build one of China's most valuable humanoid robotics companies, but also become a global leader in physical AI," stated He Xiaopeng, Xpeng's Chairman and CEO.

He Xiaopeng, Chairman and CEO at Xpeng

The convergence of these moves suggests that the physical AI market is entering an execution phase. Ren's decision to launch a dedicated robotics startup, backed by NIO's strategic support, indicates that autonomous driving expertise is now viewed as a direct pipeline into embodied intelligence. His track record in deep learning and his leadership of NIO's world model development give his new venture credibility in an increasingly crowded field.

However, the market remains cautious about robotics valuations outpacing actual deployments. Xpeng's stock fell 7 percent on August 24 after the company missed Q2 revenue expectations and guided Q3 revenue below Wall Street consensus, even as the IRON robotics round was announced. This suggests investors are weighing near-term automotive performance against long-term robotics optionality.

What Should Investors and Industry Observers Watch?

Ren's startup will likely focus on developing and commercializing physical AI foundation models that can power multiple embodied AI applications. The company's success will depend on translating NIO's world model research into practical robotics systems. Initial deployments may occur within NIO's ecosystem before expanding to external customers. The startup's ability to raise additional funding, attract top talent, and achieve meaningful robot deployments will signal whether world models truly represent the breakthrough that Ren and others believe they are.

The broader trend is clear: the most accomplished AI researchers are moving from software and autonomous systems into physical robotics. Ren's unicorn-status startup on day one reflects both the massive capital flowing into embodied AI and the conviction among leading scientists that the next era of AI progress will be measured not in benchmark scores but in robots that can perceive, reason, and act in the real world.