Experts Say Robotics
In the first half of 2026, Mingxin Xuteng (605068), an automotive interior leather company, officially announced the R&D of electronic skin materials, developing electronic skin surface base materials adapted for humanoid robots and smart cockpits;
In May 2026, Qianjie Technology was newly established, focusing on the embodiment + tactile model route, supported by electronic skin hardware data collection;
In September 2026, Lushan New Material (603051) jointly announced a strategic partnership with Jinshan Intelligence and Huan Intelligence, officially deploying the integrated solution for electronic skin + bionic robots;
Saigan Technology shifted its 2026 focus to the commercialization of full-body electronic skin for humanoid robots...
As humanoid robots enter mass production and volume scaling in 2026, and tactile sensing transforms from an optional feature to a mandatory one, the electronic skin sub-sector has unsurprisingly exploded.
01. The Upstream Gets Paid First
Morgan Stanley has raised its shipment forecast for China's humanoid robots twice this year: 14,000 units in January, revised to 28,000 units in June, and the latest 2026 forecast is 50,000 units, with 446,000 units in 2030. In mid-September, the bank's China industrial team further disclosed that in the first half of 2026, Chinese suppliers accounted for over 97% of global humanoid robot shipments, with the market size growing 272% year-on-year.
While OEMs are still worrying about capacity, costs, and application scenarios, a small group of upstream players in the supply chain have already pocketed the money first.
From August 11 to 13, Fulai New Material signed three deals in three days: Daile Somatosensory placed a bulk order for flexible tactile sensors; Zhongke Guiji purchased a full set of tactile sensing systems totaling 200,000 sets; and Haocun Technology jointly developed a new generation of data collection gloves equipped with tactile modules. Adding the 100,000-set order signed with LinkerBot in February, the company's backlog has reached the scale of hundreds of thousands of sets. Yet in December 2025, it was only delivering thousands of sets.
A similar trajectory has also emerged in Wuhan. Huaweike's market share in humanoid robot electronic skin exceeds 70%. In 2025, it supported over 18,000 dexterous hands, becoming the first company in the industry to achieve mass production of tactile sensors for dexterous hands at the 10,000-unit level; in May 2026, its tens-of-millions-piece flexible sensor base broke ground in Gedian, Ezhou, and the company disclosed that this year's orders are expected to grow by 300%.
What they are selling is the robot's "skin".
02. Distinguishing the Two Types of "Skin" First
The skin on a robot is actually not a single thing; it consists of at least two layers.
The first layer is the "outer skin". It is an exterior covering made of silicone, TPE, and thermoplastic elastomers. Its task is to make the robot look less like a pile of metal, while also providing dustproof and waterproof protection, shielding joints and wiring harnesses, and preventing collisions from causing injury. The essence of this business is materials and molds, with limited technical barriers and relatively low value addition; it solves the question of "how human-like it looks".
The second layer is the protagonist of this article: electronic skin. It is a flexible sensing system attached to the robot's surface, converting physical quantities such as pressure, shear force, temperature, proximity, and even texture into continuous data that the robot can read, use for decision-making, and utilize for training. Whether the fingertips can pick up an egg without crushing it, whether the palm can anticipate before an object slips, and whether the robot can retract within tens of milliseconds upon touching a human—all rely entirely on this layer.
The difference between the two can be described as follows: the outer skin is the clothing, while the electronic skin is the nerve endings. The former makes money from material processing, while the latter makes money from semiconductors. What is truly being repriced in this industrial cycle is the latter.
03. Just How Hot Is It?
According to Grand View Research, the global electronic skin market size was approximately USD 10.9 billion in 2024, and will increase to USD 37.1 billion by 2030, with a CAGR of 23.0%. Verified Market Research predicts that the global tactile sensor market will reach USD 35.59 billion by 2032. The statistics from these two firms are relatively broad, including medical patches and wearables, with humanoid robots being just one segment.
Zooming in on the humanoid robot sub-segment, the numbers are much smaller, but the slope is much steeper. QYResearch statistics show that global robot tactile sensor sales were USD 1.263 billion in 2024 and will reach USD 4.992 billion by 2031, with a CAGR of 21.7%; China's humanoid robot fingertip array tactile sensor market was CNY 150 million in 2025 and will reach CNY 1.367 billion by 2032, with a CAGR of 32.0%. Gongyan Wang's domestic statistics are more aggressive: the humanoid robot electronic skin market was about CNY 460 million in 2024 and will reach CNY 9.05 billion by 2030, with a CAGR of 64.3%.
Although the figures from different sources vary, they all point in the same direction: an extremely small base and an extremely steep slope. The domestic market, which was less than CNY 500 million in 2024, will grow to CNY 9 billion in six years. Such a curve is rare in the manufacturing industry.
Capital has already moved first. Since 2026, financing in the tactile sector has been advancing almost on a monthly basis: in March, Pacini completed an over-CNY 1 billion Series B round with a valuation exceeding CNY 10 billion, and in August, it secured another CNY 1 billion strategic round, accumulating CNY 3.5 billion; in June, Daimeng completed a CNY 100 million Series A round, jointly invested by Inovance Industrial Investment and China Telecom; in July, Tashan Technology secured a several-hundred-million-yuan Series B round, and Qianjue Robotics secured a hundred-million-yuan round; the former revealed that its order volume in the first half of the year exceeded four times the total for the entire previous year, with monthly deliveries reaching tens of thousands of units. Yimu Technology also joined the CNY 10 billion valuation camp with an over-CNY 1 billion Series E round.
Within a year, a niche sector originally only engaged in by component manufacturers has produced two companies with valuations exceeding CNY 10 billion.
04. Why Now?
Three conditions have matured simultaneously.
First, the bottleneck has shifted from the legs to the hands. Over the past two years, the industry has been competing on who can stand steadily and who can run and jump; now, the competition is about who can actually do the work. When it comes to doing the work, vision can only provide positioning and cannot determine whether an object is hard or soft; force control can only apply rough pressure, crushing the object if too much force is applied and dropping it if too little. Figure 03's self-developed tactile sensor can detect pressure changes as low as 3 grams, working with the Helix system to make adjustments before an object slips—this is not just showing off parameters, but the dividing line between being able or unable to sort soft express delivery bags.
Second, the cost curve has come down. The piezoresistive route features mature processes and cheap materials. The introduction of roll-to-roll printing and fully printed manufacturing processes has moved tactile sensing from the laboratory to the mass production line. Huaweike reached a capacity of 2 million automotive-grade pieces in Gedian in June 2025, and its tens-of-millions-piece base broke ground in May 2026, adopting a self-developed fully printed process. After capacity expansion, the unit price has dropped from the early tens of thousands of CNY per single point to thousands of CNY per whole hand, bringing full-body coverage into the cost range acceptable to OEMs for the first time.
Third, tactile sensing is beginning to be treated as a data asset, not just a feedback signal. Daimeng open-sourced 10,000 hours of tactile data in April 2026, with over a million downloads in the first month; data collection devices such as data collection gloves and motion capture suits have themselves become a business. Fei-Fei Li's team's recent T-Rex research reminds the industry that tactile sensing is not a vassal of vision, but an independent high-frequency physical perception channel; simply splicing tactile signals into existing VLA architectures does not necessarily yield benefits. Selling hardware is just the ticket to enter; the ability to turn signals into model capabilities will be the watershed for the next round.
Supporting technological breakthroughs are also landing in a concentrated manner in 2026.
Full-body coverage moves from concept to product. At the World Robot Conference in August, Hanwei Technology, in collaboration with Zhuoyide, released a capacitive full-body electronic skin. Based on flexible sensitive fabrics, it integrates large-area pressure arrays with clothing through a double-layer fabric structure, achieving 0.1N-level sensitivity and millisecond-level response, supporting over 10 types of interactive movements.
Multimodality becomes a consensus. Single-piece thin films synchronously collecting pressure, shear force, temperature, and proximity are replacing array solutions that can only measure single points. Taishen and Aoganwei are pursuing the triboelectric and ion-capacitive routes; Lushan New Material released a triboelectric temperature-pressure dual-modal electronic skin in January this year; Seoul National University unveiled an integrated multimodal artificial skin in July, with the thickness controlled within 0.2 millimeters.
Self-healing and durability begin to be quantified. Hydrogels and self-healing polymers allow the surface layer to restore conductivity after being scratched, with lifespan indicators entering the scale of 100,000 cycles. The group standard for "Flexible Sensors for Electronic Skin" is also being advanced, setting unified thresholds for pressure channel sensitivity (≥0.15 kPa⁻¹), temperature (-20℃~80℃), and humidity (10%RH~95%RH)—the implementation of standards usually means an industry is moving from everyone speaking for themselves to a comparable stage.
05. Landscape: 50 Manufacturers with Diverse Routes
A structural breakdown of this list of 50 companies yields several interesting judgments.
Geographically highly concentrated, Shenzhen is the absolute home ground. There are 13 companies in Shenzhen, accounting for a quarter; 7 in Beijing, 5 in Shanghai, 4 in Suzhou, and 4 in Guangzhou, with the rest scattered in Hangzhou, Dongguan, Nanjing, Changzhou, Chengdu, Wuhan, and other places. This distribution highly overlaps with the OEM and dexterous hand industries—electronic skin is a business built close to customers, and delivery radius and response speed are more important than anything else.
Technical solutions are diverse. There are 9 companies explicitly pursuing the visual-tactile (optical) route, which is the most clustered group; 5 for piezoresistive, 4 for capacitive, 4 for thin-film, 3 for MEMS, and 3 for fabrics and fibers; in addition, there are niche branches such as triboelectric and ion-capacitive, electromagnetic lattices, magnetic induction, PVDF piezoelectric, event-driven, and printed curved surfaces.
Behind the divergence of routes are two sets of logic. The piezoresistive route excels in maturity, low cost, and ease of scaling, making it the main force for securing orders at present; the visual-tactile route can achieve tens of thousands of sensing points per square centimeter and outputs image data, naturally adapting to end-to-end neural networks—this is exactly the route Tesla Optimus is taking. The collective bet by 9 companies including Daimeng, Qianjue, Diedong, and Xutai shows that the industry is betting on one thing: the next generation of tactile sensing is an image problem, not a resistance problem.
The players come from three categories. First are professional tactile startups, such as Pacini, Daimeng, Qianjue, Tashan, Tujian, and Taishen, which have the most intensive financing and highest valuations, playing the game with a sensor-algorithm-data closed loop. Second are cross-border players in materials and manufacturing: Fulai New Material enters from coated composite materials, Riying Electronics from automotive electronics, and Lushan New Material from polymer materials, with advantages in capacity ramp-up and cost control. Third are extensions of sensor leaders: Hanwei Technology (with controlling subsidiary Suzhou Leanstar Electronic Technology), Huaweike (started with RFID), and Shenhao Technology, with advantages in process accumulation and existing customer channels. There are at least 8 listed or quoted companies in the list: Hanwei Technology, Fulai New Material, Riying Electronics, Lushan New Material, Xiangshan Co., Ltd., Shenhao Technology, Aodawei, and Kuangda Technology.
There is also a structural gap. Only 5 companies in the list explicitly include "full-body coverage" in their application scenarios—Hanwei Technology, Fulai New Material, Huaweike, Riying Electronics, and Mogan Technology. The tactile sensing of the vast majority of companies remains concentrated on dexterous hands, fingertips, and finger pads. The bulk of current demand is indeed on the hands, but the single-machine value of full-body coverage is several times that of fingertips.
Most companies are still in the sample delivery and validation stage, with only a single-digit number truly generating scaled revenue. The jump in orders from thousands to hundreds of thousands of sets is noteworthy, but it must also be seen that a considerable portion of these are framework agreements, and conversion into revenue depends on the delivery pace. The track is very hot, but between the heat and the revenue, there is the distance of mass production ramp-up.
Conclusion
For humanoid robots to truly enter factories and homes, the final threshold is whether they can safely touch this world. Tactile sensing is the key to this threshold, and electronic skin is the teeth on the key.
2026 is called by many the first year of commercialization for electronic skin. Whether this statement holds true depends not on the financing amount or order announcements, but on how many robots with tactile sensing are actually working at this time next year.
This article is compiled based on publicly available network information, for reference only, and does not constitute investment advice.