1. Technical Routes and Mainstream Directions of Dexterous Hands Mainstream technical route characteristics: Currently, the mainstream drive routes for dexterous hands include tendon ropes, linkages, and gears, with no obvious advantage for any single route, each having its own characteristics: tendon rope transmission has good flexibility, linkage transmission has high precision, and gear transmission has strong load-bearing capacity. In the next 10 years, there will be no unified solution for dexterous hand technology, which needs to be customized according to the scenario. For example, in factory box-moving tasks, the dexterous hand needs to have strong gripping power; for applying sealing strips, the hand needs sensitive feedback and flexible control to match the force accurately. Analysis of Tesla’s solution: The dexterous hand of Tesla Model 3 adopts a hybrid solution combining gears, linkages, and tendon ropes, aiming to balance flexibility and strength, but has not been validated. The original plan to mass-produce 5,000 units by 2025, with a unit cost below 20,000 yuan and extensive use of dexterous hands, has been shelved. Due to issues with Optimize 2 last month, the focus has shifted to Optimize 3 this month, targeting the box-moving scenario. Currently, Tesla configures different models of dexterous hands according to scenarios, one for moving items and another for scooping popcorn at shopping mall entrances (requiring flexible control), but a universal hand that accommodates multiple scenarios has not yet been launched. 2. Comparison of the Development Status of Dexterous Hands at Home and Abroad Domestic manufacturers’ technical level: Domestic manufacturers (such as Zhiyuan, Yushu, and UBTECH) mainly produce ‘showcase products’ with limited practical application value, currently only capable of ‘entering the factory to move boxes’, meeting only simple force requirements (the force can be borne). Some manufacturers (such as Pasini and the Peking University entrepreneurial team) have explored flexible grasping, for example, the Peking University entrepreneurial team developed a hand that can play Mahjong (grasping small objects), but overall, there are no successful cases in flexible object grasping and precise control in China. There are multiple technical bottlenecks: hollow cup motors lack sufficient power; human hand grip strength requires 7-8 kg, but existing motors cannot meet this power requirement; control precision stability is poor, with some manufacturers claiming control precision of ±0.2 mm, but the repeatability is only 50%-60%, and there is significant variability in stability across different actions; the grasping success rate is low, with the international top figure hand achieving a success rate of about 98.5% for grasping glass cups (dropping once in 70 operations), while the domestic level is weaker, and the overall industry is still in its early stages with limited practical application capabilities. The technical gap with Tesla: Tesla’s dexterous hand also fails to meet practical application requirements, mainly used for simple scenarios like moving boxes, and has not achieved complex labor operations like folding clothes or sweeping floors. Even if domestic manufacturers adopt the same hand structure as Tesla, due to hardware (such as silicon micro motors, transmission components, sensors, etc.) and software (insufficient control stability) gaps, the actual performance is weaker than Tesla’s. Tesla’s dexterous hand itself also has limitations, as its dexterity has not met the needs for complex operations, and both domestic and international dexterous hands have not reached practical application standards (such as insufficient grasping success rates). 3. Key Technical Bottlenecks of Dexterous Hands Analysis of hardware limitations: The hardware limitations of dexterous hands mainly manifest in the performance of core components not meeting standards and the limitations of transmission schemes. Currently, dexterous hands generally use hollow cup motors as drives, but face issues of insufficient power and precision at small sizes, needing to meet the requirements of smaller sizes, greater power, and better precision, which have not yet been achieved.In terms of transmission schemes, linkages and tendon ropes each have their advantages and disadvantages:Linkage schemes have good precision and strong load-bearing capacity, while tendon rope schemes offer better flexibility, but there is no significant technical advantage difference between the two, and the same manufacturer often tests both schemes simultaneously. Existing dexterous hands face the problem of being ‘both expensive and not user-friendly’, reflecting that the current performance of hand products has not reached practical standards. Relationship between software and hardware: Hardware is the foundation for the functionality of dexterous hands, and software cannot compensate for hardware defects. Current hardware has issues such as insufficient motor strength and precision, high sensor noise and heat generation (difficult heat dissipation during continuous operation), and under these circumstances, software algorithm optimization cannot solve the functional deficiencies. New product development should first achieve functionality through high-spec hardware, then optimize costs, but the current situation is that even with the best hardware, it is still impossible to produce products that meet basic needs. If the hardware does not meet standards, the hypothesis of attempting to optimize through software collaboration is invalid, violating the R&D logic of ‘function first, cost later’.Material and process challenges: The performance differences of tendon rope materials (metal and polymer) and the lack of long-term experimental data are major challenges. Metal tendon ropes have high strength, low creep, and durability but are heavy and complex to wind; polymer tendon ropes have good flexibility and are easy to wind but are prone to deformation under continuous stretching. Domestic materials (such as Nanshan Zhishang and Daye Co.) can be used within a standard 100-hour usage cycle, but fatigue resistance tests for 10,000 hours have not been conducted, and long-term stability is unknown. Foreign materials (such as Dutch DYNEE and SKS fibers) show no significant difference in short-term tests compared to domestic ones, and there is currently no performance verification data in large-scale production scenarios.4. Application Scenarios and Implementation ChallengesPotential application scenarios: The main application scenarios currently being explored are concentrated in fields requiring strength or precision. One is the factory box-moving scenario, which needs to meet the specifications of a 15 kg load and a maximum length of 70 cm for rectangular objects; the success rate of robot handling in this scenario is not high, and the dexterous hand needs to address the issues of force measurement and inconsistent force application; the second is the factory sealing strip application scenario, where if it can achieve the precision of manual operation, it can be widely applied. Currently, many robot companies are focusing on promoting the implementation of these two scenarios. Other scenarios, such as welcoming and guiding in commercial industries, have lower dependence on hand performance and do not require high performance from the hands. Implementation costs and efficiency: The cost disadvantage of humanoid robots replacing human labor is significant; for example, in the box-moving scenario, the current cost of a robot is about $40,000 (approximately 300,000 yuan), with labor efficiency equivalent to more than half a worker. In comparison, the monthly salary of a mover is about 8,000 yuan, with a two-year labor cost of about $20,000. The high cost and low efficiency of robots lead to their cost-effectiveness being far lower than that of human labor, making it difficult for the market to accept. Currently, even in the most basic scenarios like moving boxes or screwing, robots still perform far worse than humans, so large-scale replacement of human labor has not yet been found. The contradiction between scenarios and technology: There is currently a chicken-and-egg problem of “unclear scenario demand” and “directionless technology development”. The downstream lacks clear scenario demands, leading the upstream to be unwilling to invest resources in developing suitable hand products; while the upstream does not develop, it is difficult to meet the potential scenario demands of the downstream. Solving this contradiction requires first fixing specific scenarios, clarifying merchant needs, and then developing suitable hand products accordingly, rather than aimlessly improving performance such as precision and strength, which could lead to products being expensive and unsellable. 5. Industry Trends and Future Outlook Mainstream manufacturers’ self-research strategy: Leading manufacturers generally choose to self-develop dexterous hands, as dexterous hands are a key core part of robots, involving patents, intellectual property, and the accumulation of R&D technical data. Currently, companies related to Musk, as well as China’s Yushu and UBTECH, continue to invest in self-research but have not launched mature products that satisfy the market. Existing self-developed products are ‘not user-friendly’ and ‘expensive’, failing to meet the standards of ‘decent and satisfactory’, leaving the market in a state of ‘no one has produced particularly impressive products’ and ‘no one has confirmed procurement’. Standardization and modularization: The current progress of industry standardization and modularization is slow, with no companies actively promoting unified interfaces. The reason is the insufficient product strength, as each company focuses on its own barriers, believing that ‘no one can solve the problem now, so there is no need to unify interfaces’. The premise of standardization is ‘clear scenario implementation + large-scale shipments’; for example, a certain hand type may promote interface unification if it is accepted by the market due to its cost-effectiveness and labor capability and is widely shipped.In terms of cost, if a million-level production scale is achieved (such as Tesla’s target of 1 million units), the cost of two hands may drop to 20,000 yuan; if the production scale is 100,000 units, the cost of two hands is about 20,000 to 30,000 yuan. Currently, the cost of producing a small batch (1,000 pairs) is about 40,000 to 50,000 yuan. Future breakthrough nodes: The key breakthrough nodes in the industry may be led by Tesla. Experts predict that breakthroughs may be achieved within three years, with the main observation target being Tesla’s Optimize 3/4 versions. If Tesla achieves the implementation of the box-moving scenario (target cost of $20,000) and the product’s cost-effectiveness reaches a level that can replace human labor, the industry will accelerate its development. Tesla has abundant resources and continues to invest, making it the most likely company to break through; other companies face limitations in resources and technical accumulation, making breakthroughs less likely. If Tesla fails, the robotics industry will remain at the level of ‘performance, display, and toys’ for a long time. Q&A Q: Among the current technical routes of tendon ropes, linkages, and gears, which is most likely to become mainstream? How do these routes compare in terms of performance, cost, and reliability? A: The current mainstream technical routes are tendon ropes, linkages, and gears, with no obvious superiority of one over the others; each route has its own characteristics in terms of performance, cost, and reliability. Taking Tesla’s latest integrated solution for Model 3 as an example, it uses a combination of gears, linkages, and tendon ropes to solve the problem of combining rigidity and flexibility, but this product has not yet been released. In the future, dexterous hands will be developed according to specific scenarios, and there will not be a unified mainstream route; at least for the next 10 years, no single technical route will significantly outperform the others.Q: Will dexterous hands not form a unified standard like human hands, but instead adopt different solutions based on different scenarios? A: Currently, the development of dexterous hands has two directions: one is specialized solutions for specific scenarios, such as those for moving boxes that need to meet strength requirements, and those for applying sealing strips that need flexible pressure feedback and precise control capabilities; the second is multifunctional solutions similar to human hands, but limited by the underlying sensor perception and motion control capabilities, which are difficult to achieve within 10 years. Currently, the industry adopts a scenario-adaptive approach, where the robot body changes hands for screwing or moving tasks based on demand, but there is still no universal solution that combines strength, precision, and cost advantages. Q: If a technology emerges in the future that enables dexterous hands to possess human hand functions, could this technology become a unified mainstream solution? A: The closest solution to this goal is the gearbox + screw + tendon rope solution proposed by Musk, which combines the motor in the arm and drives the hand, balancing flexibility and strength requirements, but the final effect still needs to be observed.Q: Will Tesla’s future mass production of dexterous hands adopt a single solution or arrange different solutions based on different scenarios? A: Tesla currently has two or three different configurations of dexterous hands, and has not yet achieved a single hand that meets multiple scenario needs. The original plan to produce 5,000 units in 2024, with pump costs controlled below 20,000, and extensive use of dexterous hands, has been shelved. Due to issues with Optimize 2 last month that need rectification, the focus has shifted to Optimize 3 for redevelopment. Currently, two hands are used for different scenarios: one for moving and one for flexible control in shopping malls, with differences in configuration.Q: What specific scenarios will the screw + tendon rope + gearbox solution be applied to in the box-moving and consumer scenarios? A: The screw + tendon rope + gearbox solution is currently applied in the box-moving scenario.Q: Is the dexterous hand for consumer scenarios simpler? A: The dexterous hand for the popcorn scenario mainly deals with precise control and flexible feedback, which differs in capability from the hand launched by figure. The current hand’s precision is insufficient, leading to spillage during operation, failing to fully meet the requirements. The hand launched by Feige, which can stack towels, performs best in flexible control and is currently the optimal product in the flexible control field. Q: How is the development level of domestic manufacturers in the dexterous hand technology field? A: Some domestic manufacturers have developed hand products, but their functionality is insufficient; companies like Pasini and Lingqiao Intelligent also have relevant product layouts. Currently, the most dexterous hand is developed by a Peking University-backed entrepreneurial team, capable of grasping small objects. Domestic manufacturers have no successful cases in the field of flexible control for flexible object grasping, with technical levels concentrated in strength control fields, such as products developed by Xiaomi Cyber, Zhiyuan, and Yushu that can perform tasks like moving boxes. The industry as a whole is still in its early stages, with practical application capabilities weaker than video demonstration effects.Q: If a hand structure with tendon ropes and four linkages is adopted, will the product performance still be weaker than Tesla’s dexterous hand due to hardware or software deficiencies? A: Yes. Even if the same structure is adopted, if the hardware or software does not reach Tesla’s level, the product performance will still be weaker than Tesla’s dexterous hand. Moreover, Tesla’s dexterous hand itself has not fully met standards, currently only capable of completing simple tasks like moving boxes, and cannot perform complex operations like folding clothes or sweeping floors. Q: What are the specific characteristics of the dexterous hand solutions from leading domestic manufacturers? A: The dexterous hand solutions from leading domestic manufacturers are generally similar, mainly using hollow cup motors and gearbox control for hand movements, with the core indicator being degrees of freedom. Higher-end solutions integrate tactile sensors for end perception, while lower-cost solutions do not have this configuration. Q: In domestic robot hand design, which scheme is more widely used, linkages or tendon ropes? A: In domestic robot hand design, there is no unified standard for the application of linkages and tendon ropes; the same manufacturer usually tests both schemes simultaneously. The tendon rope scheme has better flexibility, the linkage scheme has higher precision, and the gear structure has stronger load-bearing capacity, with each scheme having advantages in different performance dimensions; currently, there is no significantly superior product. The current performance of robot hand products is poor, with high costs and low practicality, and there are no separate hand-related events in robot competitions.Q: Will manufacturers ultimately choose self-research or procurement of core components, which is more likely? A: The final choice of core components depends on who first completes scenario implementation and achieves large-scale distribution, at which point the standards for hands will be established. Currently, companies solely focused on hand development have not launched significantly advantageous products, and the hand requirements of complete machine manufacturers are also relatively general. Currently, leading complete machine manufacturers still choose to self-develop hands, as they are core components involving patents, intellectual property, and technical data accumulation, with no manufacturers clearly indicating procurement of third-party products. Q: Do most capable manufacturers choose to self-develop dexterous hands because they are key core components? A: Leading manufacturers all choose to self-develop dexterous hands, and while no market-satisfactory products have been launched, relevant R&D work continues to advance. Q: If cost is not considered, what is the main bottleneck of current dexterous hand performance limitations, hardware or control algorithms? A: The main bottleneck of current dexterous hand performance limitations is the hardware not meeting standards, which leads to poor software performance. When hardware does not meet standards, performance limitations cannot be solved through software algorithm optimization. The basic method for developing new products is to first use high-quality components to achieve functionality without considering costs, and then optimize costs through software; however, even with high-quality hardware, functionality has not been effectively realized, so hardware is the primary bottleneck.Q: Is the hardware referred to the overall components or specific parts? A: The main hardware referred to is not up to standard, with core issues concentrated in motion and perception aspects. In terms of motion, motor strength and precision are insufficient; in terms of perception, sensors have issues with high noise and heat generation, and heat dissipation is difficult during continuous operation. Motor manufacturers report that under current size constraints, it is impossible to meet the demands for smaller sizes, greater power, and higher precision. Q: What is the expected timeline for solving current hardware performance issues? A: The timeline for solving hardware issues is constrained by the mismatch of upstream and downstream demands: downstream scenario parties believe that hardware performance is not up to standard and cannot be used, while upstream manufacturers find it difficult to develop targeted products due to a lack of clear scenario demands, creating a contradiction between demand and supply. The solution path requires first determining specific implementation scenarios, with merchants proposing clear demands, and then developing according to demand direction; if performance such as precision and strength is blindly improved without specific scenarios, it is likely to lead to excessively high product costs and poor sales, making this path unfeasible.Q: In which single scenarios are dexterous hands most likely to achieve more applications and better implementation? A: The currently foreseeable main application scenarios are concentrated in two types of operations within factories: one is moving rectangular objects of different weights, which requires strength control for dexterous hands; the other is applying sealing strips, which requires precision control. Currently, most manufacturers and robot companies are focusing on promoting the implementation of these two types of scenarios, and if they can reach the level of manual operation, it will be considered initial implementation. Other scenarios, such as welcoming and guiding in commercial industries, have lower functional dependence on hands, and more complex scenarios cannot currently be realized.Q: Is the cooking robot an important potential implementation scenario? A: The cooking robot is not an independently sold product; it is essentially a linkage mechanical arm. Q: Is the main type of motor currently used a hollow cup? A: The main type of motor currently used is a hollow cup, while frameless torque motors can be used in leg application scenarios. Hollow cup motors have the characteristics of small size and precise dimensions, while frameless torque motors have the characteristic of high power output. Sine wave magnetic field motors and permanent magnet motors are rarely used in related links, with basically no practical application cases. Q: Can Tesla’s design of placing the motor on the outer side of the arm avoid using hollow cup motors and instead use other lower-cost motors? A: This design is mainly to break free from the spatial constraints of concentrating motors in the hand, allowing for larger motor sizes. Tesla’s third-generation hand has increased degrees of freedom by 22, making it closer to human hands, but the actual effect needs to be observed. Its transmission components include a gearbox, screw, and tendon rope three-level transmission, with a complex structure, and the durability and precision of the connections between transmission components need experimental verification. The long-chain design is prone to problems if any link gets stuck, and the complex structure does not conform to Musk’s principle of first principles; the mass production situation after successful experiments still needs to be observed.Q: Does the commercialization of dexterous hands have reasonable degrees of freedom? Is there a tendency to pursue excessive degrees of freedom? A: The development of commercial dexterous hands has two directions: one focuses on special scenario needs, being practical-oriented without requiring high degrees of freedom; the other theoretically, high degrees of freedom can enhance flexibility and the ability to perform advanced tasks, but high degree of freedom designs significantly increase R&D and production costs, making products difficult to commercialize. Therefore, the current trend is more towards practicality, and pursuing degree of freedom indicators has no realistic significance.Q: When degrees of freedom increase, since each degree requires a motor to drive, will this lead to insufficient space in the hand, forcing the motors to be placed externally? A: Yes, it is certain. If the standard size of the hand is maintained, increasing degrees of freedom requires more motors to drive, and there are two solutions: one is to require motors to be more miniaturized, but motor manufacturers may not be able to achieve this; the other is to place the motors on the arm, which are the only two solutions. Q: How does the performance and lifespan comparison of domestic hollow cup motors stack up against international hollow cup motors? A: Comparisons need to be made under the same cost-performance conditions: domestic low-end hollow cup motors are comparable in performance to international products; however, in high-end fields, domestic products are weaker than international brands like Maxon. In terms of transmission components, domestic manufacturers like Shuanghuan Transmission and Zhongdali De’s reducers still have gaps compared to Murnak. The specific performance is that products in the same price range are comparable, but in the high-end price range, international products have clear advantages, while domestic products have acceptable cost-performance but insufficient high-end performance. Q: Can domestic tendon rope materials meet mass production needs in high-frequency scenarios? What are the future improvement or upgrade directions? A: Currently, the mass production scale of this product is small, and material usage refers to national material standards, meeting basic requirements for tensile strength and flatness; however, 10,000-hour operational stability tests have not been conducted, and specific situations are unknown. Q: Are the main tendon rope materials used domestically and internationally more inclined towards polymers or metals, and does this vary by application scenario? A: In domestic and international tendon rope materials, both metal and polymer materials are used, with differences in characteristics: metal materials have high strength, low creep, and high flexibility but are heavy and complex to wind, with the main feature being durability; polymer materials are soft and suitable for space-constrained scenarios but are prone to deformation under continuous stretching. Domestic suppliers include Nanshan Zhishang and Daye Co., while foreign suppliers include Dutch DYNEE and SKS, both of which are applicable within standard usage cycles and have passed high-temperature, high-load, and destruction tests, but long-term performance has not been tested.Q: Who are the relatively high-quality suppliers of brushless motors domestically and internationally? A: The selection range for brushless motor suppliers is wide, known ones include Israel’s L Company, Dongmukawa, and domestic suppliers like Huichuan.Q: Is the technical difficulty barrier for brushless motors much lower than that for hollow cup motors? A: The technical development time for brushless motors is longer, and the main difference from hollow cup motors lies in the size and arrangement of application scenarios, as the arms are thicker and have larger internal spaces, making them easier to arrange. Q: If the robot control scheme based on large models evolves highly, can it compensate for the hardware deficiencies of dexterous hands to some extent? Will future technical paths lean more towards strong hardware + simple algorithms or universal hardware + strong algorithm AI? A: The industry is currently attempting to enhance robot decision-making capabilities through large models such as visual language decision models to compensate for the hardware deficiencies of dexterous hands. If large models develop maturely, they can significantly enhance decision-making capabilities and alleviate hardware limitations, but they cannot break physical laws to compensate for the lack of hardware degrees of freedom. For example, even if the model is smart enough, it is difficult to solve operational stability issues caused by hardware clumsiness, and no feasible path has been seen to solve hardware clumsiness through strong algorithms.Q: Is there a possibility for the entire domestic industry to form interface or performance standards for dexterous hands from the top down, promoting standardization and modular design? A: The premise for the industry to form standardized and modular designs is to expand order scales and for a certain hand type to be widely accepted due to its cost-effectiveness and labor capability. Currently, due to the technical barriers of each enterprise, no company is willing to promote unified interfaces, nor is there any practical action. A possible future direction is to separate hands from the body, but current technology is not mature enough to achieve widespread application.Q: Are current leading manufacturers and related companies not pushing for interface unification due to their own ideas or unwillingness to be standardized? A: The main reason for the lack of promotion for interface unification is that both hand-making companies and complete machine companies lack the technical capabilities to solve core issues. At this stage, all parties lack the motivation to promote interface unification; only when one party has strong body performance but weak hand performance, and does not comply with interface standards, will they face elimination and be forced to open interfaces. Q: How do you view the guidance given by domestic mobile phone manufacturers and others to scale up to 200,000 to 300,000 units in two to three years? A: The volume guidance proposed by companies in the industry is mostly slogan-like and needs to clarify application scenarios and actual uses to prove its rationality. For example, Musk’s initial target of 5,000 units was later adjusted to 1 million units by 2027-2028, but this has not yet been realized, and is more of a phased expectation statement. Currently, the actual implementation scenarios for humanoid robots face significant difficulties, with the core issue being insufficient product strength: for example, in the moving scenario, the cost of a single robot is about $40,000, with labor efficiency only equivalent to half a worker, while the monthly salary of a mover is about 8,000 yuan, making the cost of robots far higher than that of human labor, leading to an inability to replace human labor on a large scale, and can only be used in small batches for demonstrations. Essentially, the cost-effectiveness of products has not reached the level of human labor, and there is a lack of implementation scenarios; whether it is directing traffic, screwing, or moving, robots perform significantly worse than humans. Q: What is the cost proportion of structural components such as actuators, motors, reducers, and sensors for dexterous hands with practical functions? A: In the cost proportion of structural components for dexterous hands, actuators account for the highest proportion, about 50%-55%; if hollow cup motors are used, this proportion can increase to 58%-59%. The sensor system accounts for about 12%-13%. Mechanical components account for about 10%-18%. The electronic control system accounts for about 13%-14%. Q: If dexterous hands with practical functions achieve mass production at the million level, what is the expected level of material costs? A: According to estimates, if dexterous hands achieve mass production at the million level, the material costs are expected to drop to $18,000-$19,000; if the production scale is 100,000 units, the unit cost is expected to be over $20,000 to $30,000. Currently, the above mass production scale has not been reached.Q: When a product priced at 100,000 yuan is sold at 1 million units, and the price drops to 20,000 yuan, does this price correspond to a single hand or a pair of hands? A: 20,000 yuan corresponds to the price of a pair of hands. Q: What is the approximate price of a hand capable of moving boxes? A: When the procurement volume is 1,000 pairs, the cost is about 40,000 to 50,000 yuan for a pair, and the cost decreases as the procurement volume increases.Q: Is the main reason for the price or cost reduction of robot-related components the scale effect? A: The main reason for the price or cost reduction of robot-related components is the scale effect. For example, for hollow cup motors, the price difference between purchasing 100 and 1,000 units is significant; a single robot requires about 10 motors, while 1,000 robots require 10,000 motors, and large-scale procurement can significantly reduce unit prices. Additionally, structural components will also reduce costs after mold production. Q: In mass-produced humanoid robots, what should the ideal cost proportion of the hands be? Compared to the computing unit, which part should be prioritized for investment at the current stage? A: Prioritizing investment in hands is more valuable at the current stage. The technology related to computing units is not yet mature; for example, Nvidia’s Jetson chips and other computing devices lack compatible models, and the application of reinforcement learning technologies is still far off, making current investments of limited significance. Q: Do other leading domestic manufacturers prioritize dexterous hands? A: The industry generally believes that dexterous hands are a key area that needs to be tackled, but few manufacturers are willing to invest large amounts of money, often showing more publicity than actual R&D. Some companies attract attention by setting up experimental factories to collect data, but progress in technical breakthroughs is limited. Q: Is Tesla indeed spending money on dexterous hand R&D? A: Currently, only Tesla and Shadow are leading in the field of dexterous hand R&D. There is a difference in the tolerance for R&D investment between the US and Chinese markets: US investors can accept long-term R&D investments without results; in the Chinese market, if a company continues to invest without results and no product sales, it is likely to be abandoned by investors. Therefore, the current industry focuses more on achieving business implementation and profitability, such as through the sale of practical products like floor cleaning machines, pool cleaning robots, and hotel delivery robots, while promoting progress in the R&D of humanoid robots and other advanced robots. Q: Can domestic companies making dexterous hands be ranked into first and second tiers? A: Lingqiao Intelligent and Pasini participated in Nvidia’s press conference, and their capacitors, sensors, and other components performed well, but it is uncertain whether they will continue to participate in the future. Companies like Yushu and UBTECH, as well as Lingqiao Intelligent and Pasini, are classified as the first tier.Q: Do the functions and costs of dexterous hands made by domestic companies generally align? A: The functions and costs of dexterous hands from domestic companies are generally similar, mainly used for demonstration, and currently cannot achieve practical labor replacement, only capable of completing simple or advanced complex demonstration operations, and have not yet broken through the key threshold of replacing human labor.Q: In the field of dexterous hands, what aspects might establish future moats? A: Currently, there is no actual moat in the field of dexterous hands; the knowledge and intellectual property accumulated through independent R&D have limited effects and have not yet demonstrated actual value; the future moat lies in being the first to achieve the implementation of specific scenarios.Q: What important supplementary information is there at the current stage? A: The current industry is in a wait-and-see state, with large companies like Xiaomi Cyber choosing to accumulate rather than push forward with cutting-edge exploration, and the industry is waiting for key goals to be achieved. The core risk in the industry is whether the goal of achieving a $20,000 box-moving capability can be reached; if successful, it will drive industry development, but if it fails, robots will remain at the level of performance, display, and toys. Q: What is the expected timeline for achieving the box-moving function in general humanoid robots? A: It is expected to take about three years. Musk has ample resources and funding, and if he cannot achieve implementation, it will be even more difficult for other companies. Currently, the products, while lacking practicality, have acceptable cost-effectiveness, and we need to wait for the release of Optimize 3/4 versions. The Musk team has made significant adjustments to previous solutions, while other companies have not made significant progress.