Dexterous Hands Become New AI Investment Hotspot: Over 20 Billion Yuan Raised in H1, Lingxin Qiaoshou Valued at $6 Billion

According to incomplete statistics from QbitAI, in the first half of 2026, financing for dexterous hand-related companies exceeded 20 billion yuan, making it one of the hottest trends in the embodied intelligence sector. Ten companies focused on dexterous hands raised nearly 8 billion yuan in new funds, with post-investment valuations totaling over 75 billion yuan. Among them, Lingxin Qiaoshou was reported in June to have a valuation as high as $6 billion (approximately 41 billion yuan), surpassing many leading humanoid robot manufacturers, prompting industry exclamations of 'too surreal.'

Behind the capital heat is a shift in customer structure. Jia Hao (pseudonym), an employee at Yinshi Robot, said that this year, traditional industrial leaders such as energy giants and home appliance manufacturing groups have proactively approached them, and procurement standards have shifted from 'can install, can move' to specific scenario needs, such as explosion-proof, anti-static, waterproof, and food-grade paint. After the Beijing Yizhuang Robot Marathon, market expectations for robots 'able to work' have risen, further boosting attention on dexterous hands.

Players are diverse, including robot manufacturers (UBTech, AgiBot, Unitree, etc.), technology background companies (Zhaowei Machinery & Electronics, Pasini, etc.), startups focused on dexterous hands (Yinshi, Aoyi, Lingxin Qiaoshou, etc.), and tech giants (Xiaomi, Tencent). AgiBot split its dexterous hand business into a new company, Critical Point (AGILINK), early this year. Qichacha shows that as of June 2026, there are over 510 existing dexterous hand-related companies nationwide.

Current high-DoF (22+ DoF) dexterous hands are expensive. For example, Sharpa's Wave five-finger dexterous hand sells for about $50,000 (over 300,000 yuan) per unit, featuring 22 degrees of freedom and integrated high-precision tactile sensors. Qin Lei (pseudonym), a business vice president at a bionic dexterous hand company, revealed that his company is about to launch a higher-DoF product with a planned price of over 500,000 yuan. In the early industry stage, due to lack of mass production, quoted prices are generally high, and actual transaction prices often have discounts, with prices around 100,000 to 200,000 yuan not uncommon.

Price differences mainly stem from core component costs. According to Qin Lei, the cost of slotless brushless coreless motors can account for 50% to 60% of the entire dexterous hand; if visuotactile capabilities are added, sensors become a new major cost. Additional features (tactile/version differences) directly create a 20% to 50% premium. For example, AgiBot's OminiHand Lingdong version without tactile sensing costs 14,800 yuan, while the version with tactile sensing costs 19,800 yuan (a 33.8% premium).

However, QbitAI's analysis of 17 manufacturers' products found that the correlation between price and degrees of freedom (DoF) is positive but weak (Pearson correlation coefficient r and coefficient of determination R² are both low). Many industry insiders believe that the overall mechanical design, transmission structure, control algorithms, models, and data capabilities are the key determinants of final performance, rather than simply stacking hardware.

Despite the capital frenzy, several leading robot manufacturers remain distant. Galaxy General, Zhiji Power, Zhiding Power, and Zhipingfang still outsource dexterous hands, while Qianxun Intelligence has developed its own three-finger dexterous hand but is still waiting on five-finger versions. The reason lies in insufficient hardware reliability: a person from an embodied robot company revealed that after testing products from multiple manufacturers, durability is generally poor, with frequent failures, and many products need replacement within a week; even a leading manufacturer's dexterous hand can only run continuously for 50 hours under a 5 kg load.

Zhang Wei, CEO of Zhiji Power, affirmed the value of dexterous hands but believes that at this stage, self-development requires answering 'whether it's worth doing and whether it adds value.' The current mainstream route for dexterous hands relies on large-scale real-world data, which demands extremely high hardware reliability; meanwhile, simulation training cannot accurately replicate complex hand contact, and hardware and algorithms constrain each other. A model engineer pointed out that most dexterous hands claiming 20+ DoF actually have narrow effective opposition space, with output control dimensions of only 6 to 10 DoF, and in most scenarios their performance is inferior to industrial grippers.

Some argue that in many industrial scenarios, two-finger or three-finger grippers can cover 90% of tasks, making five-finger designs redundant. A CEO of a Tsinghua-affiliated embodied robot startup believes that using a dexterous hand to hold a torque wrench and tighten screws is 'too redundant,' and that specialized end effectors are superior in cost, reliability, and precision.

Immature production models: Dexterous hands are still in the early industry stage, with low standardization. Many products rely on manual assembly and debugging, almost 'handcrafted,' in a 'one machine, one style' phase, far from large-scale mass production.

Beware of homogenization: Although technology routes have not converged, many companies' technical directions are highly similar. Qin Lei believes that if routes converge in the future, competition will shift to price, after-sales, and service, rather than the product itself, which will have a significant long-term impact on the industry. Lingxin Qiaoshou believes the industry is far from convergence, and the future may develop along either a fusion approach or a fundamental breakthrough.

Commercialization is a hurdle: Only a few companies have achieved true batch delivery. Dexterous hands do not exist independently; they require coordination of mechanical structure, vision, algorithms, data, and other modules. Many customers propose budgets of several million or even tens of millions of yuan, hoping to quickly deploy robot production lines, but the reality is that even Galaxy General, with billions in financing, has only achieved basic tasks like picking up snacks. There are few domestic comprehensive service providers with full-chain integration capabilities.