Covestro Humanoid Robot Materials Put Europe in a Race With China

Covestro has turned humanoid robots into a new materials battleground, despite China already controlling much of the industry’s manufacturing scale. The Covestro humanoid robot materials push focuses on polyurethane products that protect joints, soften hands, carry sensors, and absorb repeated impacts.

That creates an unusual opening for European chemical groups. Covestro and BASF do not need to build the robot’s artificial intelligence or manufacture its motors. They can instead compete for the materials surrounding those systems, including flexible skins, seals, cushioning structures, and protective joint components.

The opportunity comes with a difficult reversal. Europe’s chemical incumbents possess decades of polymer expertise, yet many of the fastest-growing robot manufacturers and component networks sit in China. Wanhua Chemical and other Chinese suppliers can combine local production with immediate access to robot makers moving toward larger manufacturing runs.

The contest will therefore depend on more than laboratory performance. European suppliers must place application engineers close to Chinese robotics teams, qualify materials quickly, and prove that specialized polymers reduce failures over a robot’s working life.

Covestro Humanoid Robot Materials Move Into the Machine

The important change is that polyurethane has moved from a supporting material to a design constraint for humanoid robots.

Covestro is positioning thermoplastic polyurethane, or TPU, across several parts of a humanoid body. TPU is a melt-processable polymer that combines flexibility, abrasion resistance, and elastic recovery.

The company’s proposed applications include fingertip covers, flexible exterior panels, artificial skin, seals, and vibration-damping elements. Cast polyurethane elastomers can also protect feet and joints against impact and repeated mechanical stress.

These applications address problems hidden by most demonstration videos. A robot can execute a short routine while its seals, covers, and damping components remain relatively fresh. Commercial machines must repeat movements for months while exposed to dust, lubricant, vibration, and collisions.

Covestro says its Desmopan TPU can resist oils, greases, and lubricants commonly found around robot joint systems. Selected grades can also be molded directly onto polyamide or polycarbonate housings, reducing the need for adhesives.

That manufacturing detail matters. Every separate fastening or bonding step adds labor, equipment, and another potential failure point. Overmolding a flexible seal onto a rigid housing can simplify assembly while keeping the component compact.

The company is also promoting Platilon TPU films for tactile sensors. These films provide flexible surfaces for printed conductive circuits that must continue working as a robot’s fingers bend.

A tactile sensor converts physical contact into an electrical signal. It helps a hand detect pressure, distribute force, and adjust its grip before an object slips or breaks.

Covestro’s tactile sensing concept places conductive ink on flexible TPU film. The approach is intended to preserve sensor performance through repeated deformation.

The material can also serve as an artificial skin substrate. That role connects two engineering problems that robot makers often treat separately: touch detection and physical protection.

A soft exterior must remain flexible enough for sensing. It must also resist abrasion, oils, cleaning agents, and accidental impacts. Those requirements become harder to balance as manufacturers reduce material thickness and robot weight.

Covestro showcased a broader robotics portfolio during Computex 2026. It included polycarbonate structural materials, rigid films for controls, TPU for electronic skin, and cast elastomers for joint protection.

The company also disclosed work with Carthane on polyurethane components for robotic feet and joint protection. That collaboration points toward application-specific development instead of selling a generic resin.

BASF has taken a similar route through partnerships with Chinese robot developers. In August 2025, it signed an agreement with Shanghai-based Fourier Intelligence covering engineering plastics, polyurethanes, and TPU.

Fourier develops humanoid and rehabilitation robots. It gives BASF a direct environment for testing how materials behave inside machines rather than only under standardized laboratory conditions.

In April 2026, BASF announced another collaboration with Sevnce Robotics. The companies plan to examine materials for lightweight quadruped and humanoid designs while testing robots in chemical-industry inspection work.

These projects turn robotics into a two-way development process. Chemical companies provide materials, while robot builders provide operating data, component geometries, and real failure cases.

That is the immediate event behind the industry interest. European chemical groups are no longer presenting robotics as a distant market forecast. They are publishing robot-specific portfolios and forming development relationships with manufacturers.

Why Polyurethane Matters Inside Robot Joints

Polyurethane earns its place by managing motion, contact, and wear where rigid components alone cannot.

A humanoid joint contains more than a motor. It can include a gearbox, bearings, encoders, wiring, housings, seals, and structures that manage shock or vibration.

These assemblies experience constantly changing loads. A knee carries weight while walking, absorbs force when a foot lands, and changes direction during balance corrections.

A wrist handles smaller loads but may perform more delicate movements. A robot hand needs enough friction to hold an object without applying excessive force.

No single polyurethane formulation can solve all those problems. Suppliers adjust hardness, elasticity, chemical resistance, damping, and processing behavior for each position.

In feet, cast polyurethane elastomers can absorb impact while protecting rigid parts from concentrated loads. In joints, flexible components can damp vibration and shield actuators from contaminants.

Damping is a material’s ability to absorb mechanical energy and reduce oscillation. It can limit noise, protect components, and help stabilize sensor readings.

Seals present another demanding case. They must flex during movement while maintaining contact pressure around rotating or sliding structures.

Compression set describes how much a material remains deformed after a sustained load. A seal with poor recovery can lose contact and allow dust or fluids into a joint.

Chemical resistance also matters because actuators may contain grease, gearbox lubricants, or hydraulic fluids. A material that swells or cracks after exposure can compromise the entire assembly.

TPU offers a useful production advantage because manufacturers can injection-mold it. Robot makers already use injection molding for many covers, housings, and cable-management components.

A flexible layer can sometimes be molded directly onto a rigid substrate. That approach can reduce assembly steps and remove separate adhesive interfaces.

Polyurethane can also contribute to weight reduction. Foamed or structured TPU cushions may distribute force with less mass than solid parts.

Weight affects nearly every humanoid design decision. A heavier limb needs more torque to accelerate, placing additional demand on motors, gearboxes, batteries, and upstream joints.

Weight near a hand or foot carries an additional penalty because the mass sits farther from the body. Moving it can require more energy than moving equivalent mass near the torso.

Materials that reduce distal weight can therefore improve more than battery life. They can change actuator requirements and help a robot respond faster.

Artificial skin introduces another layer of complexity. The surface must tolerate repeated touching, stretching, and cleaning without blocking the sensors beneath it.

Covestro says some Desmopan grades are produced without plasticizers. Plasticizers are additives used to increase flexibility, but they can migrate from some materials over time.

The company also offers grades assessed within the ISO 10993 framework for biological evaluation. That qualification can matter when robots touch people, food-related items, or medical equipment.

These claims remain product-specific. Robot manufacturers still need to test the chosen grade in the finished component and intended operating environment.

A polymer that performs well in a material data sheet can behave differently after molding. Geometry, wall thickness, temperature, additives, and contact with other materials all affect durability.

That qualification work creates a potential advantage for established chemical suppliers. They can combine formulation knowledge with processing support and regional testing resources.

However, technical support must keep pace with robot development cycles. A material that requires lengthy qualification can lose to an adequate alternative already available near the assembly line.

European Chemistry Meets China’s Robotics Scale

Europe owns valuable polymer expertise, but China controls the faster feedback loop between component suppliers, factories, and robot makers.

Covestro and BASF enter the contest with broad materials portfolios and long experience in automotive, electronics, footwear, machinery, and industrial sealing. Many robotics requirements resemble problems those industries have already addressed.

Automotive suppliers understand impact resistance, soft-touch surfaces, noise control, chemical exposure, and large-scale molding. Footwear materials offer experience with cushioning, grip, fatigue, and repeated deformation.

That existing knowledge can shorten development. It also gives European groups established test methods and manufacturing relationships across several regions.

China’s advantage comes from the density of its robotics supply chain. Motors, reducers, sensors, batteries, electronics, machined parts, and final assembly can often be sourced within connected industrial clusters.

The country also has a large domestic testing ground. Robot developers can place early machines inside factories, warehouses, public facilities, laboratories, and government-supported projects.

According to an industry deployment report, China had more than 140 humanoid manufacturers and more than 330 models during 2025. The figures came from China’s industry ministry.

The same report cited Omdia estimates showing that more than 13,000 humanoids shipped globally during 2025. AgiBot and Unitree each accounted for more than 5,000 units.

Those estimates require caution because companies and analysts can define a humanoid shipment differently. Research platforms, demonstration units, and commercially deployed workers do not represent equal demand.

Still, the direction is clear. Chinese companies have moved from isolated prototypes toward repeatable production faster than most Western competitors.

That volume generates information about weak components. Cracked covers, worn seals, noisy joints, damaged cables, and sensor drift become material-selection data.

Chinese chemical producers can respond from inside the same industrial system. Wanhua Chemical has broad capabilities in polyurethane feedstocks, TPU, polycarbonate, and other performance materials.

Wanhua has not disclosed a robotics strategy as detailed as Covestro’s public humanoid portfolio. Its scale and domestic customer access nevertheless make it an important competitive reference.

This is not simply Covestro against one Chinese company. It is a contest between two supplier environments.

European incumbents bring specialized formulations, international qualification experience, and close relationships with global manufacturers. Chinese suppliers bring speed, local capacity, and access to rapidly scaling robot programs.

BASF’s partnership with Fourier shows how European companies are responding. Rather than serving China only through exports, BASF is working directly with a Chinese humanoid developer.

The Fourier agreement covers technical, commercial, marketing, and business-model cooperation. Its material scope includes engineering plastics, polyurethane systems, and TPU.

That structure gives BASF exposure to design choices before a component reaches final procurement. Early involvement can make a supplier’s formulation harder to replace later.

Covestro follows a comparable logic by offering connected materials across housings, hands, surfaces, seals, and feet. A broader portfolio can reduce the number of suppliers a robot maker must qualify.

The risk is that robot companies may avoid dependence on premium or proprietary materials. As volumes rise, they will pressure suppliers to reduce cost and qualify second sources.

Chinese competitors can exploit that pressure. They can offer locally manufactured alternatives, faster samples, and engineering teams within driving distance of customers.

European firms must therefore prove measurable value. Better tactile performance or a softer surface will not be enough if the robot fails elsewhere before the material’s durability matters.

The winning argument will involve total operating cost. A more expensive component can make economic sense if it reduces downtime, maintenance, warranty claims, or assembly complexity.

That calculation remains difficult because most humanoid deployments have short operating histories. Buyers lack years of field data for comparing material choices.

The Real Prize Is Reliability, Not Robot Skin

The largest opportunity lies in extending useful operating life, not making humanoids look more human.

Soft skin and realistic hands attract attention because people immediately understand them. Industrial customers care more about whether the robot completes a shift without intervention.

Repeated motion exposes weaknesses that controlled demonstrations hide. A joint may cycle thousands of times while experiencing vibration, heat, contamination, and changing loads.

Flexible materials age under those conditions. They can harden, soften, tear, swell, delaminate, or develop permanent deformation.

Each failure can create consequences beyond the part itself. A damaged seal can contaminate a gearbox, while a worn foot pad can alter traction and balance.

A failed cable guide can interrupt power or data. A cracked sensor substrate can reduce tactile accuracy and cause the controller to apply the wrong grip force.

This makes polyurethane part of a system-level reliability problem. Material performance affects mechanical components, electrical signals, software behavior, and maintenance schedules.

Covestro’s robotics portfolio spans hands, exterior panels, feet, wheels, and joint-protection components. The company emphasizes wear resistance, fatigue performance, cushioning, and chemical resistance.

Those properties are relevant, but public product pages do not establish field reliability inside a specific humanoid. Independent endurance results remain scarce.

Robot makers must test complete assemblies under representative loads. That means combining motion cycles, temperature changes, chemical exposure, dust, cleaning, and unexpected contact.

The testing period can conflict with commercial urgency. Startups need visible deployment progress, while buyers want evidence that machines will remain productive.

Materials suppliers can help resolve that tension with accelerated testing. They can compare formulations under controlled stress and identify likely degradation modes before deployment.

However, accelerated tests cannot perfectly reproduce field conditions. Real factories introduce irregular impacts, maintenance errors, debris, and tasks that change over time.

Reliability also affects safety. A surface that tears may expose sharp structures or wiring, while a degraded foot component can reduce stability near workers.

Collaborative environments create higher expectations than fenced industrial robot cells. A humanoid may move through spaces designed for people and operate close to untrained staff.

Material selection must therefore support predictable failure behavior. A component should not only last; engineers should understand how it degrades and when it needs replacement.

Serviceability becomes part of the design. A durable joint cover brings limited value if technicians must disassemble an entire limb to replace it.

Suppliers that understand molding and assembly can influence these choices. They can recommend snap-fit structures, replaceable overmolded parts, or geometries that reduce stress concentration.

The same expertise can support modular maintenance. Standardized protective elements could be replaced independently from expensive actuators.

This is where European chemical firms can draw on automotive experience. Vehicle parts undergo established durability, environmental, and chemical-resistance testing before large production runs.

Yet humanoid robots introduce different movement patterns and load paths. Their designs are changing too quickly for a single qualification template to cover every machine.

The market needs common test language. Robot builders, component suppliers, and material producers must agree on cycle counts, loads, contamination, and acceptable degradation.

Without comparable data, buyers will struggle to distinguish genuine durability from marketing language. Suppliers will also find it harder to charge for higher-performing formulations.

The opportunity is therefore tied to standardization. A recognized test regime could turn material performance into a procurement metric rather than a design preference.

China’s Lead Does Not Guarantee Polymer Dominance

China’s manufacturing advantage is formidable, but robot volume alone does not determine which materials become industry standards.

A large domestic market helps suppliers improve quickly. It does not automatically produce the best formulation for every safety, durability, or sensing requirement.

Specialized polymers depend on accumulated chemistry knowledge, careful processing, and consistent quality. Small changes in formulation can affect adhesion, fatigue, friction, and chemical resistance.

Global robot makers may also require materials that meet regulations across several markets. Documentation, traceability, and stable production can become as important as nominal performance.

Covestro and BASF already operate global manufacturing and application-development networks. That reach can help a robot company use related materials across factories in Asia, Europe, and North America.

European suppliers also have opportunities outside the humanoid manufacturers themselves. Actuator makers, sensor developers, contract molders, and industrial customers can influence material specifications.

A robot startup may choose the platform architecture, but an automotive customer can impose cleaning, safety, or maintenance requirements. Those requirements can favor qualified materials.

The European industry retains a substantial installed base of industrial automation. The International Federation of Robotics reported 67,819 robot installations in the European Union during 2024.

That figure covers conventional industrial robots rather than humanoids. It still represents manufacturers familiar with automation procurement and lifecycle economics.

China operates at a different scale. The federation reported that Chinese factories installed 295,000 industrial robots during 2024, representing more than half of global installations.

Domestic suppliers captured 57 percent of China’s industrial robot installations that year. That share illustrates how quickly local manufacturers have moved into their home market.

The later factory robot data showed Chinese manufacturers increasing unit installations again during 2025. Local scale can spill into humanoid component sourcing.

Still, humanoid robots remain a less mature category than fixed industrial arms. Their commercial use cases have not settled, and many deployments remain trials.

This uncertainty gives materials suppliers room to shape the architecture. A standardized industrial robot already has established component patterns. A humanoid still invites experiments in skin, feet, hands, seals, and cushioning.

European firms can benefit if premium materials enable a task that cheaper designs cannot perform reliably. They can lose if the market standardizes around simpler machines with minimal soft surfaces.

Wheeled humanoids illustrate that uncertainty. Removing legs can reduce the number of high-load joints and foot components, narrowing some polyurethane opportunities.

Task-specific machines create another challenge. A fixed robot arm may perform factory work more cheaply than a human-shaped system, especially in a structured environment.

The Associated Press cited analysts who questioned whether real demand matches China’s production capacity. Limited use cases and difficult economics remain central concerns.

That skepticism matters to chemical companies. A supplier can win a large share of a small market and still generate modest revenue.

Material content per robot may also decline. Engineers routinely remove parts, reduce thickness, and simplify assemblies as a product approaches mass production.

Competition can intensify before demand becomes large. Several chemical companies may pursue the same early programs, giving robot manufacturers leverage during qualification.

For European incumbents, the strategic case depends on optionality. Technologies developed for humanoids can also serve cobots, logistics machines, prosthetics, medical devices, and conventional automation.

That wider market reduces dependence on a single forecast. It also allows companies to refine materials in applications already producing meaningful volumes.

What the Humanoid Materials Race Must Prove Next

Three signals will show whether European chemistry can convert robot interest into a defensible business.

The first signal is disclosed production qualification. Partnerships and product demonstrations show intent, but they do not confirm repeat orders.

Investors and industrial buyers should look for a material selected for a named production robot. The strongest evidence would include the component, manufacturing location, and expected production stage.

A qualification inside hands, joint seals, feet, or tactile sensors would strengthen the Covestro humanoid robot materials thesis. Repeated orders would matter more than a single development batch.

A lack of named production programs would weaken the case. It would suggest that robot-specific material portfolios remain early marketing efforts.

The second signal is comparative endurance data. Suppliers need evidence showing how their materials behave after repeated motion, chemical exposure, and real cleaning cycles.

Useful disclosures would measure abrasion, compression recovery, adhesion, signal stability, and component replacement intervals. They should test finished assemblies rather than isolated samples.

Independent testing would carry greater weight than supplier claims. Shared protocols could also help robot makers compare materials without repeating every evaluation from the beginning.

If standardized testing emerges, European companies could benefit from their qualification experience. If buyers prioritize immediate cost instead, Chinese suppliers could gain faster.

The third signal is the location of application development. European firms need engineers near the manufacturers producing the largest number of machines.

BASF’s work with Fourier and Sevnce shows one approach. Covestro’s collaborations and Asian product demonstrations point in the same direction.

A supplier that receives early geometry, processing, and failure data can tune a material before competitors see the requirement. That position creates more defensibility than shipping resin after the design is frozen.

Watch whether Wanhua and other Chinese chemical companies announce deeper partnerships with humanoid makers. A coordinated local response would increase pressure on Covestro and BASF.

Also watch whether robot manufacturers publish second-source strategies. Dual sourcing would limit any supplier’s pricing power, even when its material wins the first design.

The broader demand signal remains commercial deployment. China’s shipment lead is meaningful, but sustained industrial use matters more than units built for demonstrations or research.

Factory customers will ask about uptime, task completion, maintenance labor, and safety incidents. Those measurements will reveal whether advanced materials create value at system level.

Procurement teams will need to compare technical sheets, test reports, deployment records, and supplier claims. A searchable knowledge base guide can help teams preserve that evidence across long qualification cycles.

The central question is no longer whether polyurethane belongs inside humanoid robots. Existing applications already show why flexible, durable polymers are useful around moving and human-facing components.

The question is who captures the learning created by real deployment. European chemical companies start with deep materials knowledge. Chinese rivals operate beside the world’s fastest-growing robot production network.

Covestro can win meaningful positions if its materials reduce failures, simplify assembly, or improve safe contact. It must prove those gains inside machines that work beyond controlled demonstrations.

For developers, buyers, and investors, the next step is straightforward. Ignore the soft-skin spectacle and follow production qualifications, endurance results, and repeat material orders. Those signals will reveal whether European polyurethane expertise can hold its ground as China scales the humanoid supply chain.