1. Unitree stock rallies about 60% in under two weeks
Shares of Unitree, the Chinese robot maker listed on Shanghai's STAR Market, have gained more than 60% since October 15, closing at 672 yuan on Monday, October 26, and lifting its market value to 271.8 billion yuan, about $40.8 billion. Investors are betting that advanced chatbot technology, specifically OpenAI's GPT-6 Astra, will supply the intelligence that robots have so far lacked.
The rally has turned Unitree into something of a stand-in for artificial intelligence itself. The company sells machines, but traders are buying the stock for what AI may soon allow those machines to do. The move gathered speed after a company conference last week and was reinforced on Monday by a rating upgrade from JPMorgan.
This article explains what drove the surge, how the shares moved day by day, what the bank changed its mind about, and which risks remain for a company whose valuation now leans heavily on a technology it does not itself build.
For context, a rise of that size in under two weeks is large for a company with a market value above 270 billion yuan. Moves of this scale in established firms usually require a major shift in how investors view the business, and here that shift is attributed almost entirely to expectations about artificial intelligence.
2. What is the bet behind the Unitree rally?
The wager is that the technology behind chatbots can soon be used to control physical machines, a field commonly called physical AI. Chatbots are good at language and reasoning, but robots have historically struggled to turn instructions into reliable movement in messy real-world settings. If a general-purpose AI model can close that gap, robots could start doing household chores such as tidying rooms or folding laundry, tasks that today remain out of reach for commercial products.
In that scenario, much of the value would go to the companies that make the physical hardware: the bodies, motors and joints that an AI model would command. Unitree makes exactly that kind of hardware and already sells robots, which is why investors see it as a natural way to gain exposure to the theme.
The logic is a bet on sequencing. The software intelligence is arriving from outside, and the company with ready, affordable machines stands to benefit once the two are combined.
Put simply, the stock is priced less on what Unitree sells today than on what its machines might be able to do tomorrow. That distinction explains why a rise in an AI model's reputation can lift a hardware company's shares by double digits within days, and why the shares could react just as strongly if expectations for the technology cool.
The term physical AI refers to systems that perceive and act in the physical world, in contrast to software that only produces text or images. Robots are its most visible application, because a machine with arms, legs and sensors must translate reasoning into precise motion.
3. What did Unitree's founder say at the Hangzhou conference?
The immediate catalyst was the World Physical AI Conference in Hangzhou on October 22. It was Unitree's first partner conference in its decade of operation, and founder Wang Xingxing used it to argue that large models are close to a breakthrough in physical AI. He said the breakthrough could arrive within one to three years, and he claimed that GPT-6 Astra can already direct robots with no task-specific training.
In an interview with 36Kr, Wang gave a more detailed timeline. If things go well, he said, it will take one to two years; otherwise, two to three. The estimate is a forecast from the person who runs the company, not an independent assessment, but it supplied the narrative investors were looking for.
Statements from a founder at the company's own event carry obvious limits as evidence. Still, the combination of a specific time frame and a claim about an existing OpenAI model gave traders a concrete story to attach to the stock.
The Hangzhou event also marked a milestone for the company. Holding a first partner conference after ten years of operation signals a shift from a founder-led hardware maker toward a business courting an ecosystem of outside partners, which fits the narrative that its machines will serve as a platform for other firms' AI.
4. How have Unitree shares moved this week?
The price action has been sharp. On Friday, October 23, the day after the conference, the shares hit the 20% daily limit, the maximum gain permitted in a single session, and closed at 624 yuan, about $93.55. On Monday they rose another 7.69% to 672 yuan, about $100.75.
A limit-up close shows that buyers outnumbered sellers so heavily that the exchange's cap on daily gains was reached. The cumulative gain since October 15 is more than 60%, and the company's market value has climbed to 271.8 billion yuan.
Rapid advances of this kind often reflect momentum as much as fundamentals, since each strong session attracts more buyers who want to avoid missing out. That makes the next several trading days a test of how much of the move rests on durable conviction.
Daily price limits exist to temper speculative swings, and hitting one is a signal worth noting. Traders in a stock that closes at its limit often see heavy demand carry into the next session, which is consistent with Monday's further 7.69% rise.
5. Why did JPMorgan upgrade Unitree to overweight?
On Monday, JPMorgan upgraded Unitree to overweight from underweight, a two-step move from the bank's most cautious stance to a positive one. Overweight means the bank believes investors should hold more of the stock than its share of the market would suggest. JPMorgan also raised its price target to 995 yuan, about $149.17, roughly 48% above Monday's close.
The bank had previously argued that Unitree depended too heavily on hardware and was weak in software. It changed its view after speaking with Wang and other executives and reviewing humanoid models in Hangzhou.
A reversal from underweight to overweight within a single rating cycle is unusual, and it gave the rally an endorsement from a major global institution. It also suggests the bank sees the software gap as something that outside AI models can fill, rather than something Unitree must solve alone.
For readers unfamiliar with analyst ratings, such changes matter because large institutions and funds often refer to them when deciding allocations. A higher price target also gives the market a reference point for how much more upside the bank believes is justified if its assumptions prove correct.
The upgrade is also notable because it came after direct contact with management. Speaking with executives and examining the humanoid models in person gave the bank information beyond published financial statements, and its conclusion shifted accordingly.
6. What are Unitree's background and risks?
Unitree listed on the STAR Market, Shanghai's technology-focused board, on August 19. Its G1 humanoid costs about $16,000, which is cheaper than rival machines and a key part of its appeal. The company already has products on the market, setting it apart from firms that are still at the prototype stage.
Risks are real, however. Researchers have flagged security flaws in Unitree robots, including audio, video and spatial data being streamed to outside servers without users' knowledge. For machines that may one day operate inside homes, such findings bear directly on whether customers and regulators will accept them.
The AI premise has nonetheless pushed the valuation higher. A company priced for a future in which robots become useful household devices has little margin for disappointment if the technology takes longer than hoped or if privacy concerns slow adoption.
The mix of strengths and weaknesses explains why views on the stock are so polarized: the product is real and priced to sell, yet the data practices described by researchers and the high valuation both invite scrutiny.
7. How do Tesla and Figure AI compare?
Unitree is not alone in chasing the humanoid market. Tesla has just begun producing its Optimus robots at its Fremont, California, factory and expects external deliveries around the end of 2027. Unitree, by comparison, already sells its robots.
Figure AI, another humanoid developer, partnered with OpenAI in 2024 and uses OpenAI models to power its machines. That link shows why OpenAI's progress is closely watched across the sector: the same models that could lift Unitree are also tied to its competitors.
The picture is one of an early race in which hardware makers, software developers and carmakers are converging on the same goal, with price, readiness for sale and the quality of the AI brain all shaping who benefits.
In practical terms, Tesla's timeline places meaningful outside deliveries of its robots well in the future, while Unitree can already ship units. Whether that head start translates into lasting advantage depends on how quickly competitors with large manufacturing and software resources close the gap.
8. What happens next for Unitree investors?
The next signals will come from the stock itself and from evidence on the technology. Investors will watch whether trading holds near 672 yuan after the rapid gains, whether the price moves toward JPMorgan's 995 yuan target, and whether any demonstrations or customer results support Wang's claim that GPT-6 Astra can direct robots with zero training.
Wang's timeline of one to three years for a physical AI breakthrough sets a rough yardstick. Over that period the company will need to show that capable AI and affordable hardware can combine into products people actually buy, while addressing the data-security concerns raised by researchers.
Any reversal in sentiment toward AI stocks more broadly would also likely be felt here, given how closely the shares now track the theme.
The central question for investors is therefore one of timing: whether the improvements in AI models arrive quickly enough, and in a form reliable enough, to justify a price that already reflects optimistic assumptions about the next few years.