Unitree Shares Surge on Hopes That Chatbot AI Can Power Robots

Preface

Unitree Robotics has attracted renewed investor attention as optimism grows that artificial intelligence developed for chatbots could soon help robots operate more effectively in the physical world. Since Oct. 15, the Chinese robotics company’s shares have risen by more than 60%, bringing its market value to 271.8 billion yuan ($40.8 billion). The movement reflects a broader investment thesis: advanced AI models may give robots the ability to understand instructions and carry out practical tasks, rather than simply demonstrate preprogrammed actions.

That possibility received a boost when Unitree founder Wang Xingxing discussed the timeline for progress in physical AI at a conference on Oct. 22. JPMorgan also upgraded the stock on Oct. 26, citing a reassessment of the company’s prospects. This article examines the factors behind the rally, the potential role of large models, and the opportunities and concerns surrounding Unitree and the wider humanoid-robot industry.

Lazy bag

Unitree shares have climbed sharply as investors consider whether chatbot-style AI can enable robots to perform useful tasks in real-world settings. Founder Wang Xingxing said physical AI could make a breakthrough within one to three years, while JPMorgan upgraded the stock and increased its price target. Unitree already sells humanoid robots, including the G1, which costs roughly $16,000. However, the company’s valuation also depends on overcoming practical challenges, including software capability, reliable operation, and concerns about data security.

The central question is whether advances in AI can turn robot hardware into dependable, commercially useful machines.

Main Body

Unitree Robotics’ recent share-price rally illustrates how expectations about artificial intelligence can influence valuations in the robotics sector. Shares have risen more than 60% since Oct. 15, and the company’s market value reached 271.8 billion yuan (roughly $40.8 billion). The rise is not solely a judgment about current robot sales. It also reflects investors’ belief that improvements in AI could make robots more capable and expand the range of tasks they can perform.

The technology at the center of this investment case is often described as physical AI: AI systems that interpret information and act in the physical world. Chatbots respond to text or other digital inputs, while a physical AI system must connect perception and reasoning to a machine’s movements. A robot asked to tidy a room or fold laundry, for example, would need to understand the request, identify relevant objects, plan a sequence of actions, and adjust its movements as circumstances change.

This is a more demanding task than producing a convincing answer in a conversation. Real environments are unpredictable. Objects may be arranged differently, instructions may be unclear, and a robot must carry out movements safely and consistently. The prospect that large models could help meet these challenges has therefore become a significant part of the market’s interest in companies that manufacture robots and their components.

Unitree founder Wang Xingxing added to that interest at the World Physical AI Conference in Hangzhou on Oct. 22. The event was the company’s first partner conference in a decade of operation. Wang said large models—chatbot-style AI systems trained on vast amounts of data—could make a breakthrough in physical AI within a few years. He described the possible timeline this way: “If everything goes well, it will take one to two years, otherwise it will take two to three years.” His comments offered investors a time frame for when more capable systems might emerge, while also acknowledging that progress depends on how development proceeds.

The source also reports a claim that OpenAI’s GPT-6 Astra can already direct robots with zero training. Such a capability would be relevant to the broader debate about whether general-purpose AI models can guide machines without requiring extensive task-specific preparation. However, the claim does not, by itself, establish how reliably or widely such a system works in practical settings. The commercial significance would depend on real-world performance, the tasks supported, and whether the system can operate safely and consistently.

Trading activity showed the strength of investor response. Unitree shares reached the 20% daily limit on Friday, the maximum amount a stock may rise in one session, and closed at 624 yuan ($93.55). On Monday, they gained another 7.69%, closing at 672 yuan ($100.75). These gains helped lift the company’s market value to 271.8 billion yuan ($40.8 billion). Share-price momentum can reflect expectations and new information as well as business fundamentals, so the rally should not be taken as proof that the technology or commercial opportunity is already established.

JPMorgan upgraded Unitree to overweight from underweight on Oct. 26 and raised its price target to 995 yuan ($149.17). In investment research, an overweight rating generally means an analyst expects a stock to perform better than its benchmark or recommends holding a larger position than the benchmark weight. The bank had previously argued that Unitree relied too heavily on hardware and was weak in software. It changed its view after hearing from Wang and other executives and reviewing the company’s humanoid lineup in Hangzhou.

The rating change highlights a key issue for robotics companies: manufacturing a robot is only part of the challenge. Hardware provides the physical structure, motors, and joints, but useful performance also depends on software that can interpret instructions and coordinate actions. If AI models make robots more flexible, companies with existing manufacturing capabilities could benefit. Yet investors must still assess whether software improvements translate into products customers will buy and use at scale.

Unitree went public on Aug. 19 on the Shanghai Stock Exchange’s STAR Market, a technology-focused board. The company already sells robots, distinguishing it from some competitors that are still working toward broader production. Its G1 humanoid costs roughly $16,000, making it cheaper than rival humanoids, according to the source. A comparatively low price could make the product more accessible to researchers or other customers, but the price alone does not determine whether a robot is practical. Capability, reliability, support, and the cost of deployment also matter.

Security is another consideration. Researchers have flagged flaws involving audio, video, and spatial data being streamed to outside servers without users being notified. This raises questions about transparency, privacy, and how robotic devices handle information collected in homes or workplaces. The issue matters because robots may operate in close proximity to people and gather sensitive data as they navigate their surroundings. Addressing these concerns would be important to building trust among customers and other stakeholders.

Unitree’s position can also be considered alongside activity by other companies. Tesla has only just started producing Optimus robots at its Fremont, California, factory and expects outside deliveries around the end of 2027. Figure AI partnered with OpenAI on its humanoids back in 2024, using OpenAI’s models to power its robots. These examples show that companies are pursuing different combinations of robot hardware and AI partnerships, with timelines and commercial progress that are not necessarily identical.

The wider investment question is whether AI advances can create practical value beyond demonstrations. A robot that responds to an instruction in a controlled setting may still need substantial development before it can perform the same task reliably in an ordinary home, factory, or other workplace. Useful deployment requires the system to cope with varied objects and environments, operate safely, and deliver results at a cost customers consider worthwhile. Progress in large models may help, but it does not remove these engineering and commercial requirements.

For Unitree, the current share-price movement reflects optimism about that possibility as well as attention to the company’s existing products and manufacturing position. Wang’s remarks and JPMorgan’s upgrade have strengthened the case that physical AI deserves investor consideration. At the same time, the software questions previously raised by analysts, the security concerns identified by researchers, and the difficulty of turning robot demonstrations into reliable services remain relevant. The company’s future prospects will depend on how effectively it addresses those issues and whether customers adopt its robots in meaningful numbers.

In short, the rally is a bet on a potential transition: from robots that can perform limited, prearranged actions to machines that can understand instructions and adapt to real-world tasks. If that transition happens, companies producing robot bodies and related technology could benefit. If it takes longer than expected, or if the resulting systems are unreliable or difficult to deploy, investor expectations may not be matched by commercial outcomes. The gap between promise and practical performance is therefore central to evaluating Unitree’s rise.

Key Insights Table