On the evening of July 6, the Shanghai Stock Exchange website updated Unitree Technology's IPO review status to "Registration Effective." On July 2, the China Securities Regulatory Commission approved Unitree Technology's registration application for an initial public offering and listing on the Sci-Tech Innovation Board.
From application acceptance on March 20 to registration effective, 104 days. The SSE review and approval took only 73 days, a remarkably fast pace under the Sci-Tech Innovation Board's pre-review mechanism. This Hangzhou-based "maker of robot dogs" is the first company in this wave of humanoid robotics to secure an IPO ticket after UBTech.
According to public information, Unitree plans to raise 4.202 billion yuan by issuing no less than 10% of its shares. Based on previous market expectations, Unitree's post-IPO valuation could approach the 50-billion-yuan mark.
But this is far more than just an IPO story.
A-Share's Only Humanoid Robot Complete Machine Play: What's It Really Worth?
Looking back at Unitree's R&D and commercialization path, a clear main thread emerges. Leverage extreme engineering capabilities to turn expensive legged robots into "big toys" and "productivity tools," and sell them worldwide.
In 2013, Wang Xingxing was still an undergraduate at Zhejiang Sci-Tech University. During summer break, he tinkered in the lab and created a quadruped robot called XDog. He soldered all the motors, drivers, and main control boards himself. Three years later, armed with this graduation-project-level prototype and 2 million yuan in angel investment, he registered Unitree Technology in Hangzhou.
What truly brought the industry's attention to this company was Laikago in 2017, the first publicly available high-performance quadruped robot for retail in China. This was followed by AlienGo (2019) and A1 (2020), with products iterating generation by generation and prices progressively dropping. In 2021, 24 A1 robots performed a dance for Andy Lau on CCTV's Spring Festival Gala, marking Unitree's first breakout moment.
But Unitree's pace didn't stop there. The summer of 2023 was a turning point for the company: in July, it launched Go2 with a starting price slashed to 9,997 yuan; just a month later, the full-size general-purpose humanoid robot H1 debuted, standing 1.8 meters tall and weighing 47 kilograms. Unitree, which started with quadrupeds, officially plunged into the deep waters of the humanoid robotics track.
After that, Unitree's product cadence was breathtakingly fast.
From the G1 humanoid agent launching its first salvo in May 2024 with a starting price of 99,000 yuan, to the H2 Plus featuring NVIDIA's Jetson Thor and Isaac GR00T platform unveiled in June 2026, in just two years, it relentlessly released a series of products, laying out a comprehensive product matrix ranging from consumer-grade to industrial-grade, from the 26,900-yuan dual-arm humanoid to the 3.9 million-yuan manned mech GD01.
Sandwiched between these releases were the 2025 Spring Festival Gala where H1 robots collaborated with Zhang Yimou on the viral "秧Bot" performance, and the July 2025 launch of the ultra-lightweight humanoid robot R1 priced at 29,900 yuan. To some extent, Unitree is unleashing its engineering momentum across various product lines.
This has placed Unitree far ahead in commercialization performance within the entire humanoid robotics track.
By 2023, Unitree's share of global shipments for quadruped robots exceeded 60%, with overseas revenue accounting for half. This dual-track advancement of "products everywhere and solid revenue" is the real foundation that allowed it to break from the geek circle into the capital markets.
The prospectus shows Unitree's revenue skyrocketed from 159 million yuan in 2023 to 1.71 billion yuan in 2025, an increase of more than tenfold in two years. Net profit also turned from a loss in 2023 to 288 million yuan in 2025. In 2025 alone, it shipped over 5,500 humanoid robots, while cumulative sales of quadruped robots exceeded 30,000 units.
Price is Unitree's biggest weapon, but certainly not its only one. Go2 starts at under 10,000 yuan, R1 under 30,000 yuan—just two years ago, the entry point for humanoid robots was six or even seven figures.
The overseas market contributes roughly 40% of revenue. According to media reports, starting from May this year, Japan Airlines began trial runs of Unitree's G1 robots at Tokyo's Haneda Airport, handling tasks like loading/unloading luggage, transporting cargo, and coordinating conveyor belts, with the trial period planned to last until 2028.
After Getting 42 Billion, Can Unitree Blaze a Trail No One Else Has?
For the roughly 4.2 billion yuan raised from the IPO, Unitree mentioned several key uses: strengthening robot model R&D, body (hardware) R&D, new product development, and building a new manufacturing base.
In plain English: make more, make them smarter, and drive costs down further.
But the challenges now facing this company extend far beyond "how to spend this money well." As the only A-share humanoid robot complete machine play, Unitree is indeed a promising target for investors. However, post-IPO, its biggest test is growth. How the raised money is spent will directly impact its subsequent growth narrative.
According to the "2025 Humanoid Robot Market Research Report" released by CCID Media, in 2025, China had over 140 humanoid robot complete machine enterprises, with shipments of 14,400 units (84.7% global share) and a market size of 1.55 billion yuan (53.8% global share). It is understood that over half of China's provinces included embodied intelligence and robotics in their 2026 government work reports.
Wei Kai, Director of the AI Research Institute at the China Academy of Information and Communications Technology, stated that for humanoid robots to enter real-world scenarios, a much stronger "brain" must be trained. Currently, data usable for robot operational training is extremely scarce, which is the biggest bottleneck for the evolution of the humanoid robot "brain."
The pie is indeed huge, with an industry consensus that the market will grow from hundreds of billions to trillions. But there are also more players at the table.
At this crossroads, the strategic divide between different players is becoming clearer than ever. The company most often compared to Unitree is UBTech, already listed on the Hong Kong Exchange. Although both have humanoid robot products, their underlying track DNA is diverging.
Judging from UBTech's 2025 financial report, its largest revenue stream comes from industrial manufacturing scenarios, a commercially proven path. However, surprisingly, UBTech has also newly ventured into an "Emotional Companion" track beyond this field.
On June 30, UBTech launched its full-size ultra-bionic humanoid robot series "U-World U1 Series" at a global press conference. The series focuses on family emotional companionship, available in male and female versions with heights ranging from 1.60m to 1.85m. Equipped with 88 high-DOF joints, they feature dialogue and eye-gaze interaction capabilities. Prices are tiered clearly: from the Lite upper-body version at 119,800 yuan, to the full-body high-end Pro at 169,800 yuan, to the Ultra male version at 990,000 yuan and female version at 880,000 yuan. Official disclosures state that total channel orders have exceeded 13,000 units.
In contrast, Unitree faces a large amount of homogenized competition, including domestic peers like Zhongqing, Zhiyuan, Fourier, as well as overseas players like Tesla's Optimus and Norway's 1X NEO.
Fourier has already introduced its GRx series and has the support of mature cash flow from its rehabilitation robot business line. Tesla's Optimus Gen 2 isn't in mass production yet, but Musk has anchored its price around $30,000, squarely clashing with Unitree's R1 price point. Norway's 1X NEO has begun accepting pre-orders, targeting home scenarios.
Therefore, whether Unitree can succeed next depends on three things.
First, whether it can find a "second category anchor" beyond quadrupeds and research-oriented humanoids. Unitree's robotic dogs have achieved global shipment leadership, but humanoid robots are a different story. The G1 trial at Haneda Airport is a good start. However, transitioning from "trial run" to "signed contract," and replicating from Haneda to more airports, warehouses, and logistics scenarios, tests product stability and service systems, not just price.
It also depends on whether Unitree can defend its price moat amidst fierce competition. The R1 at 29,900 yuan and G1 at 99,000 yuan currently place Unitree almost alone in this price segment. But as UBTech's U1 series pulls the consumer-grade humanoid starting price down to 119,800 yuan and Tesla's Optimus targets $30,000, the price gap is narrowing. Unitree must simultaneously achieve two things: continue cost reduction, while turning "cheap" into "cheap and effective."
The final and most crucial point is whether it can turn the embodied intelligence story into reality. The H2 Plus developed with NVIDIA, equipped with the Jetson Thor computing platform and Isaac GR00T foundational model framework, aims for large-scale training of robot operation skills in simulated environments. This is the right direction. No matter how many hardware units are sold, if the "brain" can't keep up, the ceiling arrives quickly. But the distance from the lab to the factory for embodied intelligence is farther than many imagine.
So, what path might Unitree forge?
Based on its current layout, Unitree most likely will not take UBTech's consumer-grade route, nor Tesla's "super-factory self-use" route. The former requires a seasoned consumer market expert, a type of person not prominent among Unitree's current management holding decision-making power. The latter requires its own factory production lines.
A more plausible path is one of "development tools + industry benchmarks": use low-priced products like R1 and G1 to achieve volume, allowing developers and universities to explore application scenarios; simultaneously, use high-performance platforms like H2 Plus to tie closely with NVIDIA, serving as a training foundation for embodied intelligence; then, use benchmark cases like Haneda Airport, BYD factories, and State Grid inspections to penetrate industries one by one.
The risk of this path is that achieving volume often means unattractive gross margins, and industry cases imply long cycles and customization. But if the bet pays off, if the humanoid robot market truly explodes in the next three to five years, the shipment volume, data, and scenario experience accumulated by Unitree would become assets that competitors couldn't catch up with regardless of their spending.
Looking at it now, the "Registration Effective" status is just the admission ticket. The real game is only just beginning.
This article is from the WeChat public account: Phoenix News Tech , author: Phoenix News Tech, editor: Dong Yuqing
