50,000 robots roll off the production line, and less than 10 grams of adhesive in their batteries are rewriting the rules of a hundred-billion-dollar supply chain

In 2026, humanoid robots crossed the threshold into mass production. CINNO Research expects global shipments to exceed 50,000 units, up 700% year on year. Unitree went public, and a robot sprinter broke the 9-second barrier in the 100-meter race at the Robot Games—from laboratories to factories, the industry finally felt, for the first time, like it had entered a true “first year of mass production.”

Everyone is watching the robot itself. I’m watching the battery inside it—or more precisely, the few grams of adhesive inside that battery.

A humanoid robot’s battery typically holds less than 2 kWh. Unitree H1 has a 0.864kWh battery, with static operating time of less than 4 hours; Tesla Optimus Gen2 packs a 2.3kWh high-nickel ternary battery, but can only run dynamically for about two hours. Compared with the 60 to 100kWh battery packs commonly found in EVs, that is barely a fraction. Yet it is precisely this “small” size that exposes the temper of the entire battery supply chain.

Why are batteries so difficult? Four dimensions are tightening simultaneously

Space：The humanoid robot’s torso must accommodate the battery, motors, reducers, control boards, and sensors, leaving the battery with a narrow space of awkward geometry. It is not a neat rectangle, but an irregular cavity following the contours of the frame, forcing the cells to be arranged wherever there is room.

Weight：Robots are sensitive to every gram. Add one gram to the battery, and the joints and motors have to carry one extra gram; both runtime and movement quality suffer. The 2.3kWh battery pack in Tesla Optimus Gen2 has been reduced to around 10 kilograms, demanding an energy density far higher than that of automotive batteries.

Power：The instantaneous power peaks during startup, climbing, and emergency stops are very high. Taking Unitree H1 as an example, the peak power of its joint motors can reach several kilowatts. The battery must support a high instantaneous discharge rate, yet its capacity is only 0.864kWh. This means an extremely steep discharge curve, putting both the cells and interconnect materials to the test.

Heat dissipation：As energy density continues to rise, heat becomes more concentrated. Dissipating heat in a space the size of a palm is far more difficult than in an EV. EV battery packs have liquid-cooling plates and vehicle-wide air channels, while robot batteries often rely only on the housing and structural components for auxiliary heat dissipation, leaving an extremely narrow thermal-conduction window for the adhesive.

These four challenges are precisely what is driving the next generation of battery technology—and the fundamental reason solid-state batteries are being brought to the forefront.

Solid-state batteries are the key, but the adhesive must evolve with them

Solid-state batteries replace liquid electrolytes with solid electrolytes, offering higher energy density, improved safety, and a smaller footprint. XPeng IRON uses all-solid-state batteries, with official figures claiming a 30% weight reduction and a 30% increase in capacity; GAC GoMate achieves 6 hours of runtime with all-solid-state batteries; Farasis Energy’s first-generation sulfide all-solid-state cells have an energy density of 400Wh/kg and are said to enable robots to run continuously for 8 to 12 hours; EVE Energy’s “Longquan No. 2” achieves a volumetric energy density of 700Wh/L and is specifically designed for humanoid robots and low-altitude aircraft. CALB’s self-developed all-solid-state cells exceed 430Wh/kg. CATL plans to produce all-solid-state batteries in small batches in 2027. Overseas players are also moving ahead: Samsung SDI showcased pouch-type all-solid-state batteries at InterBattery 2026, focusing on lightweighting; LG Energy Solution has directly planned an “anode-free” solid-state battery for robots, aiming to maximize volumetric energy density.

The technical routes have not yet converged. Oxide, sulfide, and polymer technologies are each developing independently, while pouch, prismatic, and anode-free formats are all being tested. But the direction is consistent: smaller, lighter, and denser.

And every step forward for solid-state batteries requires the adhesive inside the battery to evolve as well.

That is where the phrase “miniaturization adaptation” comes from. It is not simply about using less adhesive. Every performance metric of the adhesive must be put through a new round of screening within a smaller space, with more concentrated heat generation and more demanding weight constraints.

An expert’s view: Four hurdles in miniaturization adaptation

First hurdle: space. Automotive battery-pack gaps are measured in millimeters, but in the small batteries used in humanoid robots, the gaps between cells and between cells and the housing have already reached the sub-millimeter level. The adhesive must have low viscosity—around 1,000 to 2,000 centipoise—so that capillary action can carry it into these narrow gaps without trapping air bubbles. It must also support precise dispensing, with both the adhesive volume and dispensing path accurately controlled so that it does not flow where it should not. The old approach of high-volume potting fails completely here.

More importantly, the irregular structure of robot batteries creates irregular dispensing paths. Conventional automotive battery packs are regular cuboids, allowing dispensing robots to move in straight lines and across flat surfaces. Robot battery cavities have curves, steps, and narrow gaps, requiring more flexible dispensing equipment and more precise visual positioning. This places higher demands on the adhesive’s rheological properties: it must flow in, but also stop in place.

Second hurdle: heat dissipation. With smaller batteries, energy density is actually higher, and heat generation becomes more concentrated during high-rate discharge. The thermal conductivity of thermally conductive potting compounds must be pushed upward—from the conventional level of around 1.0 to a range of 1.5 to 3.0 or even higher. The adhesive layer must also be thin: the thinner it is, the lower the thermal resistance. But adding more thermal-conductive filler increases viscosity, creating a contradiction that requires careful formulation to find the right balance.

A common approach in the industry is to use mixed filler systems based on aluminum oxide, boron nitride, aluminum nitride, and other materials. Aluminum oxide is inexpensive and provides good insulation, but its improvement in thermal conductivity is limited; the platelet structure of boron nitride helps form thermal-conduction pathways, but it is expensive and difficult to disperse. How to maintain a thermal conductivity of 1.5 W/m·K while keeping viscosity within a range suitable for dispensing, and at the same time ensuring that the cured material does not crack or delaminate, is an equation formulation engineers solve every day.

The potting process must also generate little heat and exhibit low shrinkage, so that the heat released during curing does not damage the delicate cells. Although solid-state batteries have better thermal stability, the aluminum-plastic film used in pouch packaging has limited heat resistance, and curing temperatures above 80°C may cause the sealed edges to fail.

Third hurdle: light weight and safety. Robots are sensitive to every gram. The adhesive layer must be thin and the amount used is measured in grams, but bonding strength cannot be compromised; a thin adhesive layer still has to withstand vibration and impact. Requirements for insulation, flame retardancy, and sealing are also more stringent. Although solid-state batteries are safer, pouch-cell packaging, insulation protection, and moisture-resistant sealing remain essential. Resistance to temperature cycling and vibration are non-negotiable requirements.

Take insulation as an example. The busbars and sampling harnesses in robot batteries are densely packed, and the breakdown voltage of the adhesive layer is typically required to be ≥10 kV/mm. For flame retardancy, UL94 V-0 is the baseline, while some customers even require V-0 together with low smoke and halogen-free properties. With all these metrics stacked together, the formulation window is quite narrow.

Fourth hurdle: process. As batteries become smaller, production lines become more automated, and the adhesive’s curing method must be compatible: either UV curing for positioning within seconds, or low-temperature rapid curing to fit the takt time. Batch consistency must also be much tighter than before—a robot battery pack may contain less than 10 grams of adhesive in total, and a small batch variation can scrap the entire pack.

This also explains why robot battery production lines place far higher demands on dispensing equipment than automotive production lines. A single dispensing point on an automotive battery pack may use dozens of grams of adhesive, whereas a robot battery may use only a fraction of a gram. If the adhesive-volume error exceeds 5%, it may affect thermal conduction or bonding. High-precision metering, in-line inspection, and closed-loop control are becoming standard equipment on robot battery production lines.

The types of adhesive are changing too

Adhesives for humanoid robot batteries are not a single product category, but a combination of materials. Common types include:

• Thermally conductive potting compound：Used to fill the gaps between cells and the housing while providing thermal conduction, insulation, and cushioning. The mainstream systems are two-component polyurethane and silicone gel: polyurethane offers good toughness and low cost, while silicone gel has a wide temperature range and low stress.

• Structural bonding adhesive：Used to secure cells and bond module frames, requiring high strength and vibration resistance. Epoxy and acrylic structural adhesives are common choices.

• Thermally conductive gap filler：Used between the bottom of the cells and the cooling structure. It offers controllable thickness and high thermal conductivity, with common forms including thermal pads and dispensable thermal gels.

• Conformal coating：Used to protect PCBs and sampling circuits against moisture while providing insulation. It must be thin and uniform without affecting heat dissipation.

Every type of adhesive must be reselected in response to “miniaturization.” Automotive grades cannot simply be transferred directly; formulations and process windows must be readjusted for sub-millimeter gaps, gram-level usage, and irregular cavities.

Market opportunity: foreign players have a first-mover advantage, while domestic players have a scenario advantage

Who is competing for this market? Foreign players have a first-mover advantage. Companies such as 3M, Henkel, and Dow have built up deep expertise in electronic adhesives and thermal-conductive materials. Henkel’s thermally conductive potting compounds lead the automotive market in share; 3M’s thermal gap fillers have accumulated long-term experience in consumer electronics and automotive electronics; and Dow’s silicone gel systems have earned a strong reputation in high-reliability applications.

But domestic manufacturers have opportunities as well. Chinese companies have spent years cultivating applications such as battery thermal potting and magnet bonding, matching performance while offering faster delivery and response times. Huitian New Materials has entered the supply chains of mainstream battery manufacturers in structural and thermal-conductive adhesives for power batteries; Debon Technology has technical reserves in electronic packaging and thermal-conductive materials; and Kafuter has a cost advantage in silicone sealants. Adhesives for humanoid robot batteries are becoming another new battlefield for domestic substitution.

More importantly, robot battery adhesives are highly customized. Battery structures, thermal-management solutions, and production-line takt times vary from one robot manufacturer to another, making it difficult for standard products to provide full coverage. This gives domestic adhesive companies that stay close to customers and respond quickly an opportunity to enter the market.

CINNO Research forecasts that demand for solid-state batteries will rise from 0.05GWh in 2025 to 74.2GWh in 2035—a more than thousandfold increase. Battery adhesives will follow, creating a substantial incremental market. Assuming battery adhesive consumption of approximately 80 to 120 tons per GWh, the corresponding demand for battery adhesives in 2035 could reach tens of thousands of tons, making the market impossible to overlook.

Conclusion: 50,000 units are only the beginning—the competition for adhesives is just getting started

The 50,000-unit mark for humanoid robots is only the beginning. Batteries are getting smaller, and adhesives are becoming finer. From millimeter-scale to sub-millimeter-scale, from dozens of grams to just a few grams, from conventional thermal conduction to thin layers with high thermal conductivity, and from manual dispensing to high-precision automation—the competition in this field is only just beginning.

Whoever can be the first to overcome the four hurdles of “miniaturized adaptation” will secure a position in the humanoid robot battery supply chain over the next few years. For adhesive companies, this is not only a new market, but also an opportunity to shift from an “automotive mindset” to a “robotics mindset.”

Materiomart Technical TeamBy keeping up with the technological evolution of adhesives for humanoid robot batteries, we can provide customers with material selection and sample testing support for thermal-conductive potting, structural bonding, thermally conductive gap fillers, and more. Please feel free to contact us for further information.