The lobby of Chinese robotics firm Linkerbot’s headquarters doubles as a waiting area and display room, with rows of robotic hands developed by the Beijing firm taking up half the space. Some are mounted on robotic arms to demonstrate their capabilities to visitors, like tightening a screw into a pre-drilled hole. A few are found on a quartet of full-sized humanoids dressed in traditional red Mamianqun skirts that can play pieces like Chinese composer Li Huanzhi’s “Spring Festival Overture” on the keyboard and drums.
Cofounded in 2023 by chairman and CEO Alex Zhou, Linkerbot debuted last year on Forbes Asia’s 100 to Watch list featuring Asia-Pacific’s small companies and startups on the rise. The five-fingered dexterous robotic hands the company makes are among the most crucial components in humanoids and industrial robots, enabling them to execute a range of tasks from mundane and repetitive to highly complex. They are already being used in factories across China to grasp objects and assemble parts, says Zhou, adding that one of his next goals is to make them adept at chores such as cooking and cleaning, which would transform humanoids into indispensable home helpers.
“The core of what we do is letting robotic hands have the skills of a top craftsman,” says the 40-year-old entrepreneur from his office crammed with robotic parts. “We aren’t just trying to make the hand perform a single task like cooking. We are bestowing upon it the entire skill set of a human hand.”
A band of robots fitted with Linkerbot hands performs in Shenzhen.
Zuo Dongchen/VCG via Getty Images
Reverse engineering the human hand, which is remarkably sophisticated with its capacity for fine motion control, adaptable grip and sensory feedback, is described by industry insiders as one of robotics’ greatest challenges. At the Los Angeles All-In Summit last September, billionaire Elon Musk said that while Tesla’s Optimus humanoid is good at walking, creating human-like hands, from the hardware that powers movement to software that controls action, has proved far more elusive.
Zhou acknowledges that a technology gap exists, but says his plan of combining AI with China’s manufacturing prowess will close it. The company, he discloses, is developing a database of skills to boost the hand’s versatility while it simultaneously expands production to drive down costs, giving Linkerbot a pricing edge in an already crowded field at home and overseas.
Wuhan-based brokerage TF Securities estimates there are over two dozen Chinese companies making robotic hands as of 2025. Meanwhile, the total shipment of hands in China is likely to touch 70,000 units by year-end, a nearly 300% rise from 2025, and expand more than sixfold to 430,000 units annually by 2030, according to a report by Shenzhen-based research firm Gaogong Industry Research Institute.
Investors are piling in. Backed by fintech giant Ant Group, HSG and CICC Capital, Linkerbot closed a series B+ round in April, which it said valued the startup at $3 billion. Zhou, whose 31% stake gives him an estimated net worth of $920 million, has already kicked off another financing round targeting a valuation of $6 billion, which would make him a billionaire. While media reports indicate that Linkerbot has plans to list in Hong Kong in 2027, a company representative says there is no clear timeline for an IPO.
So far, Linkerbot’s portfolio includes six models priced between 6,666 yuan and 99,999 yuan ($980 and $14,700). The 06, Linkerbot’s cheapest model, is capable of bending its five fingers to grasp and hold bigger objects like a smartphone, while the most-expensive L30 sports knuckles and fingers that can each move independently to perform high-precision tasks like using a tweezer to pick up tiny screws. The company’s bestseller is the 49,999-yuan L20, which is a notch below L30 in dexterity but still capable of holding water cup-size objects, tightening bolts and uncapping bottles.
The ability to undertake such seemingly mundane tasks is no small engineering feat, says Dermot McGrath, the Shanghai-based founder of consultancy ZenGen Labs. Robotic hands are typically made of small motors that power their motion and have specialized wires to aid movement. Sensors installed on fingertips help the hardware “feel.” Tightening a screw for instance, requires several motors to work in sync to ensure that the grip is neither too hard nor too soft. Complex sensing technologies come into play to enable robotic hands to apply just enough pressure to pick up a delicate object like an egg without cracking the shell.
A delicate gripping task needs just the right amount of force.
courtesy of linkerbot
In March, Singapore-based Sharpa, one of Linkerbot’s biggest competitors, successfully trained its Wave dexterous hand to peel an apple—a task that’s widely considered as a breakthrough in the robotics world. As compared with Wave, Linkerbot’s most advanced L30 doesn’t apply pressure as accurately because it’s equipped with fewer sensors, says McGrath. The two products, however, offer similar dexterity, and pricing might be Linkerbot’s biggest advantage as the company expands, he adds.
Unlike most of its competitors, Linkerbot makes nearly all components in-house, which gives it better control over the supply chain and helps in keeping costs down. The Wave hand, which reportedly sells for $50,000 each, is three times more expensive than the L30. Shadow Robot, a London-based maker of robotic hands, sells its most advanced model, which can use tools like a wrench, for €110,000 ($125,000) each.
“Linkerbot has one of the cheapest and best hands you can buy right now,” says McGrath. “They make their own parts and components, which means they control the cost, and they can iterate faster.”
Zhou wants to drive costs down further. With a new production line in Beijing that started in August, which brings its total factories to four across China, the company is aiming to increase production to as many as 100,000 hands annually in the next few years. Last year, it was the country’s second-largest maker of robotic hands by shipments, after fellow Beijing-based Inspire Robots, according to Gaogong.
“We aren’t just trying to make the hand perform a single task like cooking. We are bestowing upon it the entire skill set of a human hand.”
Leveraging economies of scale could bring hardware costs down to less than $100 per pair within three years, according to Linkerbot. Globally, the average hardware cost for a pair of robotic hands stands at about $6,000, while in China, the average cost is $1,200 per pair, says Ming Hsun Lee, Hong Kong-based analyst at BofA Global Research. He estimates that costs may fall by 15% annually as production increases.
Churning out higher volumes could give Linkerbot more bargaining power with suppliers of materials, adding to cost savings. Linkerbot’s cofounder, Su Yang, who oversees AI and software, was quoted last year by local media as saying that as production expands, it could help Linkerbot bring costs down to 500 yuan, or roughly $75 for a pair of robotic hands.
Fistful of Dollars The robotic hand is the most expensive single component of a humanoid robot.
Source: BofA Global Research
Besides humanoid robot makers, Linkerbot’s customers include universities, such as Stanford University in California and Tsinghua University in Beijing, which are researching more applications of robotic hands in school labs. South Korea’s Samsung and Germany’s Siemens count among its overseas customers, according to Linkerbot. The company declines to disclose its financials.
Faced with a looming shortage of workers amid a rapidly aging population, Chinese manufacturers are also deploying Linkerbot’s products across assembly lines, according to Zhou. That includes Shenzhen-listed foodservice firm Qianweiyangchu Food, which in July announced its plans to use Linkerbot’s hands for a new automated assembly line to make frozen and prepackaged meals.
But local rivals are giving chase. Shanghai-based AgiBot, which was founded the same year as Linkerbot, sells its OmniHand for $2,200 each, while Inspire Robots sells an entry-level model for $2,700. Chinese e-commerce site JD.com has models from dozens of brands on sale priced between $445 and $14,800, depending on the function and dexterity offered.
Kevin Lynch, a professor of mechanical engineering at Northwestern University in Chicago, says he has tested hands from Chinese companies including Robotera, Wuji Technology, Inspire Robots, as well as products from Sharpa, Shadow Robot and South Korea’s Wonik Robotics, but there’s not one clear winner. “Each model has its own strengths and weaknesses,” he notes.
To stand out, Chinese companies might resort to the old playbook of racing to the bottom, cautions ZenGen Labs’ McGrath. The practice, known as neijuan, or involution, has beset industries from electric vehicles to food delivery as firms seek to undercut each other in pricing to win customers.
Zhou dismisses Linkerbot’s local competitors as “followers,” pointing to their establishment over the past year amid China’s efforts to nurture high-tech industries such as robotics with tax rebates and subsidies. He says Linkerbot is pulling ahead on the software side with its proprietary database of skills called LinkerSkillNet, which the company claims is by far the largest in the world. It currently has over 500 skills, including such precision movements as chopping vegetables and threading a needle.
Linkerbot employees collect data points by wearing gloves that capture motion and also by fitting them onto robotic arms. These are then converted into blocks of code that correspond to a particular robotic movement such as bending a finger. The codes can also be combined and recombined—similar to how Lego blocks are stacked—to direct robotic hands in a series of actions that culminate in more complex tasks, like making dumplings or playing musical instruments.
Chinese humanoid robot maker Unitree Robotics, whose 36-year-old founder Wang Xingxing amassed a fortune of $12.1 billion following the firm’s blockbuster listing in Shanghai in August, recently launched its own software platform for robots. It’s similar to an app store for a smartphone, where users can download robot skills such as dancing and martial arts. But Linkerbot’s database appears to be focused solely on the hand instead of whole-body movement, says McGrath.
To stand out, Chinese companies might again resort to the old playbook of racing to the bottom.
Linkerbot is also working on an AI model, called Linker Genesis, that allows users to control the hands through verbal commands. “We aren’t just a dexterous hand company,” says Zhou. “We are also an AI model developer.” The entrepreneur plans to double the team’s size to 1,000 employees by year-end, with half of the new hires expected to work on AI-related research.
Zhou’s passion for robots goes back to his childhood when, as a voracious reader of Japanese manga series Doraemon, he became fascinated by the cat-shaped robot’s ability to pull out a seemingly endless number of gadgets from its magical pocket. Hands, the entrepreneur says, are just like that magical pocket because they are key to carrying out countless tasks.
But Zhou didn’t immediately settle on robotics after graduating from Huazhong University of Science and Technology in Hubei Province, where he secured a place at age 14 to participate in interdisciplinary studies, including AI and robotics, as part of a nationwide program for talented kids. He spent two years working in Chinese electronics giant Lenovo’s research division in Beijing, where he developed specialized software for the design of smartphones. Zhou later cofounded two startups, one in gaming and the other in autonomous driving, before selling his stakes, details of which he declines to disclose.
The proceeds were used to set up Linkerbot and fund it in the first year, Zhou says. As the company moved from the lab to production, investors came knocking, including Louis Liang, vice president of Beijing-based Cyber Creation Ventures, which invested in the startup last year. Liang says when he first met Zhou at a networking event organized by Tsinghua University in 2024, he was impressed with Zhou’s focus on lowering costs by making components in-house. “He understood from the very beginning that these products can’t just have advanced technologies,” Liang says. “Somehow they must be produced on a big scale.”
Analysts such as BofA Global Research’s Lee caution that robotic hands will take a while to become as good as human hands and moreover, the business is inextricably linked to advances in humanoid technology. Zhou, however, sees it differently, emphasizing that he’s seeking to position Linkerbot as a key components supplier to the entire robotics industry.
Another potential cloud looms. With the U.S.-China tech race intensifying, the Trump administration in late July banned imports of humanoid robot models and power inverters. The company says the restriction doesn’t cover makers of robotic hands, adding that Linkerbot’s relationship with its U.S. customers remains unaffected. However, the company could be slapped with high tariffs in the future, according to Ethan Qi, Beijing-based associate director at Counterpoint Research. If that comes to pass, component makers may opt for workarounds such as shifting production to Southeast Asia as Chinese garment manufacturers have done, he notes.
Meantime Zhou is pressing ahead. In July, at the World Artificial Intelligence Conference in Shanghai, Linkerbot demonstrated its first step toward making its own factories fully automated by unveiling an upcoming model, the O30, which has the capability to assemble robotic hands without human involvement. “We are laser focused on making the most important component,” Zhou says. “It is just like how Nvidia is developing the most crucial hardware in the age of AI.”
Get A Grip
A dexterous hand is by far the most complex and expensive component of a humanoid, accounting for nearly a fifth of the average $28,000 hardware cost of a human-shaped robot, according to an April report from BofA Global Research. Performance is typically measured by degrees of freedom (DoF), a benchmark of the number of movements a hand can make. To fold clothes or chop vegetables, for example, hand hardware needs to have at least 20 DoF. A human hand has 27 DoF.
ForbesForbes Asia 100 To Watch 2026By Rana Wehbe WatsonForbesForbes Asia 100 To Watch 2025By Forbes Asia TeamForbesForbes Asia 100 To Watch 2024By Forbes Asia Team