Wang Xingxing and Jensen Huang Discuss Physical AI, Robots Become a Key Deployment Scenario

Cailian Press Financial News Agency

Open APP

STAR Market Daily, Song Ziqiao, Editor: Ji Sheng, 2026-10-22 12:24

① Wang Xingxing said that the core challenge for embodied intelligence to pass the ChatGPT moment is the alignment between the input and output of AI models and real robots; ② Wang Xingxing said that robots will be deployed in commercial services faster than in homes and industrial fields; ③ Robots are an important deployment scenario for physical AI.

STAR Market Daily, October 22 (Editor: Song Ziqiao) — At the 2026 World Physical AI Conference and Unitree Partner Ecosystem Conference, Wang Xingxing, founder of Unitree Robotics, said that almost everyone in the world, including himself, does not attach enough importance to and have enough faith in AI. With the release of GPT-6, physical AI has now shown signs of being broken through, very likely in the next 1-2 years or 2-3 years; the core challenge for embodied intelligence to pass the ChatGPT moment is the alignment between the input and output of AI models and real robots.

He believes that deployment in commercial services will be faster than in homes and industrial fields.

In his view, the continuous leap in foundation model capabilities will drive a simultaneous improvement in the effectiveness of the self-evolution loop of physical AI robot models. As foundation models continue to iterate and upgrade every month and every year, the capabilities of the self-evolution system itself will also strengthen, forming a virtuous cycle.

Once this closed-loop system operates smoothly, it will greatly improve robot development efficiency and the speed of iterative evolution. After running continuously for a period of time, it is expected to truly achieve a substantial leap in robots' self-evolution capabilities.

Huang Renxun has stated on multiple occasions that "the next wave is physical AI," that is, the era of robots. According to NVIDIA's explanation, physical AI refers to models that use motor skills to understand and interact with the real world; they are usually encapsulated in autonomous machines such as robots or self-driving cars. Using physical AI, autonomous machines can perceive, understand, and perform complex operations in the real world.

On October 21, Huang Renxun delivered a keynote speech at GTC Berlin, saying that AI factory deployment could be shortened to as fast as 14 days, and physical AI will accelerate its deployment.

On Wednesday, US Eastern Time, Musk said at Tesla's earnings call that he firmly believes the entire humanoid robot sector will become the world's largest industry, and that in the future the number of humanoid robots on Earth may even exceed the number of humans. He also expects that by the end of next year, a series of issues including robot AI, electromechanical design, reliability, and service life are expected to be resolved. "By that point, we can deliver Optimus to companies other than Tesla."

Open Source Securities pointed out that physical AI is expected to become the market with the largest demand space after coding. Taking Optimus as an example, its deployment path is first factories, then commercial use, and then homes. Optimus has now clearly entered the mass production stage. By the end of October 2026, upstream core components are expected to reach a supply capacity of 2,000 to 2,500 robots per week, and in 2027, a volume of 100,000 units is conservatively expected for the full year. Most general-purpose frontier model companies such as OpenAI, Anthropic, and DeepSeek are preparing to enter embodied intelligence and physical AI.

Guojin Securities stated that world models, by learning environmental states and their changes over time and actions, establish internal environmental representations for robots that can be used for simulation. VLA maps visual and language inputs to robot actions, while world models provide future state prediction and outcome evaluation for action selection. As model capabilities extend to the physical world, competition in physical AI will cover the complete technology chain, including spatial data, physical simulation, robot bodies, training compute, and real-world scenario feedback, creating larger commercialization scenarios for AI.

CITIC Securities said that the progress of GPT-6 and Gemini 4 in robot planning, spatial perception, and Real-to-Sim modeling is expanding the space for general large models to participate in robot development and training. On hardware, domestic complete-machine manufacturers and Tesla's Optimus mass production ramp-up are continuously accelerating, and Tesla is expected to release Optimus V3 in the fourth quarter, bringing the development of the humanoid robot industry into a new stage. Based on this, it is optimistic about the mass production and application prospects of the robot industry and recommends paying attention to core manufacturers in the industry chain.

Artificial Intelligence

Robots

Cailian Press statement: The content of this article is for reference only and does not constitute investment advice. Investors who act on it do so at their own risk.

Business Cooperation

Popular Unlocks

[Institutional Research] This company is one of the few domestic enterprises with end-to-end full-process foundry capabilities for optical interconnection

This company is positioned in the NPO/CPO next-generation high-speed optical interconnection track, has opened up full-size production capacity, and provides scarce full-chain optical interconnection solutions.

[Big Shot Holdings Tracking] Rare earths + earnings pre-increase, third-quarter net profit is expected to increase by up to 1.3 times quarter-on-quarter, can achieve full separation of fifteen rare earth elements, oxide purity can reach above 99.999%, this company's annual 5,000-ton rare earth separation and processing project has been transferred to formal production and operation

Rare earths + earnings pre-increase, third-quarter net profit is expected to increase by up to 1.3 times quarter-on-quarter, can achieve full separation of fifteen rare earth elements, oxide purity can reach above 99.999%, the annual 5,000-ton rare earth separation and processing project has been transferred to formal production and operation, and this company has solved the low-cost defluorination "bottleneck" problem in the rare earth separation industry.

Related Reading

Eight departments: Strengthen R&D layout of general large models and industry large models in the industrial foundation field, and promote the integration of industrial mechanisms and agent decision-making models

Cailian Press 2026-10-22 10:16

① The Ministry of Industry and Information Technology and seven other departments issued the "15th Five-Year Plan for Improving Industrial Foundation Capabilities"; ② Promote AI empowerment of industrial foundation innovation, and accelerate the implementation and application of artificial intelligence in all links such as product R&D, design simulation, pilot testing, and production operations.

Cailian Press, Liu Rui, 2026-10-22 08:48

① Meta's total revenue in the third quarter of 2026 was US$64.6 billion, up 26% year-on-year, with earnings per share of US$3.83; ② The company expects Reality Labs' losses to narrow to about US$15 billion next year, and full-year capital expenditure to narrow to US$135-145 billion.

Tesla Q3 earnings released: doubling down on AI and robots while the core automotive business comes under pressure, capital expenditure far below expectations!

Cailian Press, Huang Junzhi, 2026-10-22 08:07

① Tesla's third-quarter revenue was US$28.05 billion, down 0.2% year-on-year; adjusted earnings per share were US$0.48, estimated at US$0.44; ② Tesla's third-quarter capital expenditure was US$4.73 billion, estimated at US$7.95 billion; ③ The above funds will be used for the Optimus production line, AI computing, etc.