Humanoid robotics stocks.

Advantech explicitly offers Jetson Thor-based controllers for humanoids. It integrates systems rather than fabricating AI chips or DRAM.

Allegro explicitly targets humanoid joint motor drivers and magnetic position feedback. Position sensing is not direct fingertip contact sensing.

Allient explicitly markets humanoid motors and integrated motion. The source does not establish a specific miniature finger-motor design.

Amphenol explicitly maps humanoid interconnects. It supplies electrical connections, not complete sensors, camera processors, or battery cells.

Amprius markets high-energy cells for humanoid applications. The page does not establish commercial humanoid order volumes.

The humanoid portfolio covers precision feedback, motor control, and tactile signal chains. These are electronics, not sensing flexures.

Arm explicitly maps licensed processor architecture to humanoids. IP licensing is distinct from manufacturing the robot’s processor chips.

CATL identifies its batteries in the Galbot S1 at a CATL factory. A named deployment does not establish broad humanoid battery market share.

EVE explicitly discusses battery development for humanoids. Development claims do not demonstrate large-scale shipments.

The integrated-actuator white paper explicitly discusses humanoid joint requirements; strain-wave expertise does not establish a planetary-gear supply role.

The humanoid application portfolio covers power, motor control, and sensing. Hand relevance is control electronics rather than mechanical finger manufacture.

Intel names humanoid demonstrations using Core Ultra. This compute exposure does not represent ownership of the spun-out RealSense camera business.

Keli reports its six-axis force sensors at the wrists of Botler humanoids in Zeekr factory testing. Wrist feedback is not the same as fingertip tactile skin.

FMH frameless motors explicitly target humanoid joints. This motor source does not establish production of complete reducers or miniature finger motors.

Leadshine markets humanoid motors, joint modules, and complete dexterous hands. Documented hand designs use geared or screw transmissions, not the model’s tendon.

The robot-battery portfolio explicitly includes humanoids and cell/module/pack options. Humanoid revenue is not separately quantified.

The Boston Dynamics agreement specifically concerns Atlas vision modules. It does not establish inertial or wrist force-sensing supply.

Melexis documents joint-position ICs and Tactaxis 3D tactile force sensing for humanoids. These are distinct sensing technologies.

Microchip explicitly markets humanoid processing, motion, and sensing electronics. It is not a mechanical force-transducer manufacturer.

Micron explicitly discusses humanoid memory requirements and provides DRAM/storage technology. Robotics revenue is not separately quantified.

MinebeaMitsumi explicitly maps compact motors and bearings to humanoid motion. Connectors alone do not establish a force-sensing business.

MPS documents humanoid SoC power stages and catalogs BMS/motor-control ICs. Supporting power circuitry is not memory manufacturing.

MOONS’ explicitly discusses slotless and coreless motor development for humanoids. Motor capability does not imply tendon or finger-link production.

Nidec explicitly markets motors and FLEXWAVE reducers for humanoids. The reducer architecture is strain-wave.

Tuopu’s March 2026 exchange filing describes humanoid actuation R&D, prototype validation, and small-scale deliveries. It does not identify customers or establish volume leadership.

Novanta’s humanoid portfolio includes motors, encoders, and ATI force/torque sensing. Force sensors are distinguished from fingertip skin.

Jetson Thor and the Isaac/GR00T stack explicitly target humanoids, with named robot developers. NVIDIA is not a DRAM or standalone heat-sink manufacturer.

NXP describes distributed robot-body processing and deterministic communications for humanoids, distinct from the main AI accelerator.

OMNIVISION announced a global-shutter sensor and ISP solution for humanoid multi-camera sensing on NVIDIA platforms in October 2024.

The Dragonwing IQ10 reference design explicitly supports humanoids. Reference-design availability is not a statement of market share.

Kollmorgen publishes a humanoid motor-design white paper. The cited frameless joint motors do not establish a miniature finger-motor product.

Renesas humanoid reference designs cover system compute, servo control, BMS, and dexterous-hand electronics.

The PAL Robotics case study documents humanoid joint encoder use. RLS is an associate of Renishaw, not a wholly owned business.

DYNAMIXEL Q targets humanoids and ROBOTIS offers complete dexterous hands. Hand integration does not prove supply of the illustrative tendon cable.

Sanhua’s 2025 annual report discloses bionic-robot electromechanical actuator R&D, trial production, and sampling. Specific OEM names and scale are not assumed.

The Humanoid partnership includes actuator development and supply. This is evidence for joint systems, not a confirmed finger or tendon supplier.

ST explicitly describes humanoid inertial, imaging, and depth sensing. Its processing role is embedded control, not high-end AI GPUs.

SUNON explicitly markets cooling for humanoid heads, torsos, limbs, and packs. Its compute link is thermal hardware, not processor design.

TDK’s robotics page explicitly covers IMUs and pressure-based tactile sensing for humanoids. An IMU measures motion, not contact force.

Humanoid reference designs cover motor drives, BMS, sensing, controllers, and multi-axis hand control. TI supplies chips, not complete mechanical hands.

Timken explicitly maps Cone Drive, SPINEA, CGI, and bearings to humanoid motion. Different subsidiaries provide different gear architectures.

Vishay’s September 2026 announcement explicitly describes humanoid motor-control demonstrations. Data-center power exhibits do not prove robot BMS or memory supply.

VPG explicitly maps strain gauges and force/torque transducers to humanoid actuator feedback. VPG and Vishay Intertechnology are separate companies.

The April 2026 investor presentation discloses humanoid-actuator development following the Tokyo Robotics acquisition. This remains development evidence.

The BOXER platform supports robot cameras, sensors, and actuators. It is an embedded system, not a standalone semiconductor or heat-sink supplier.

AMD documents processors, adaptive compute, and Kria robotics modules. Embedded processing is the relevant role, not memory-chip or heat-sink manufacturing.

Magnetic angle/position products list robots among applications. Their previous fingertip touch link has been removed.

Dunkermotoren documents motor/gear/drive combinations for mobile robots and automation under AMETEK. No humanoid design win is asserted.

Basler stereo cameras explicitly support vision-guided robotics. Camera integration is established; humanoid share is not.

CKD documents compact electric grippers and actuators. The source does not establish structural robot links or humanoid hand supply.

Industrial machine-vision systems guide and inspect automated processes. Camera-system integration is distinct from manufacturing CMOS chips or lens glass.

Coherent provides lasers, detectors, and illumination optics for robotic 3D sensing. Illumination optics are not the illustrated camera’s imaging lens stack.

Exlar actuators drive industrial robot tooling and additional axes. They are not verified lightweight structural links or rotary motor stators.

The NSK collaboration concerns robotics actuator development. It does not quantify humanoid sales or establish standalone reducer supply.

Ennoconn lists robotics and digital-twin systems integration. This is broader automation software/hardware exposure, not verified onboard humanoid silicon.

The listed company documents industrial servo motors and drives. Separate companies using the Estun brand are not automatically consolidated into this exposure.

FANUC documents industrial servo motors and amplifiers. Packaging and power requirements differ from mobile humanoid joints.

The verified product page documents harmonic reducers for six-joint robots. Other motor, hand, and sensor links are withheld pending a specific primary source.

JT128 supplies 3D ranging for robots. Lidar belongs with perception, not a wrist force/torque sensor or a CMOS image-sensor chip.

The corporate product catalogue includes precision bearings, ball screws, and DATORKER strain-wave gears. No humanoid customer is identified here.

Honeywell documents compact force sensors for robot grippers. HON is Honeywell Technologies after the June 2026 aerospace spin-off; HONA is separate.

IEI documents AMR controllers and embedded computers. AMR compatibility does not establish humanoid use or standalone cooling-component manufacture.

The cited product is a compact motor for robot vacuum cleaners. A dexterous-hand role is not established and has been removed.

JTEKT’s technical publication specifically addresses bearings for precision robot reducers; the cited evidence is industrial robotics.

KEYENCE documents industrial inspection and robot guidance. These vision products are not force/torque sensors or an onboard humanoid camera contract.

Leaderdrive documents strain-wave reducers, integrated rotary actuators, and frameless motors; these do not imply planetary-gear products.

LUSTER documents industrial imaging and vision inspection. This does not prove an onboard humanoid camera-board design.

Genio platforms explicitly support intelligent robotics. Platform suitability is not proof of a particular humanoid production win.

MELSERVO covers industrial rotary, linear, and direct-drive motors. Humanoid customer adoption is not established here.

Moog documents brushless motors and servo technology for robotic motion. Aerospace scale is not a measure of humanoid exposure.

The approximately 60% share is Nabtesco’s estimate for medium-to-large industrial robot precision reducers, not humanoid market share.

Nachi’s company profile documents bearings and industrial robot manufacturing. Humanoid-specific bearing sales are not established.

Nikon’s encoder brochure documents position feedback for robot joints and servo motors. This entry is not about consumer cameras.

NSK documents robot bearings, including wave bearings for strain-wave reducers. Bearings support the gearbox; they are not its toothed gears.

NTN’s robot catalogue documents rolling bearings for robot mechanisms. This supports bearing races and rolling elements, not toothed gears.

NJ5 coordinates factory robots and servo axes. It is not an onboard humanoid AI compute module.

The robotics application article documents Hyperlux sensing. It does not identify humanoid production contracts.

Orbbec documents stereo, structured-light, and time-of-flight camera systems for robotics. It is a camera supplier, not a CMOS wafer manufacturer.

The cell catalogue explicitly lists robotics applications. Cell production does not establish standalone enclosure or busbar manufacture.

The thin-section bearing guide lists robotics and harmonic drives as applications. This is bearing exposure, not a reducer manufacturer claim.

The Hyundai/Kia agreement is for service-robot battery development. The release does not specifically establish supply to Atlas or another humanoid.

Sensata’s encoder products list robotics applications. ST is Sensata; STM is STMicroelectronics.

Inovance documents servo drives, motion control, and industrial automation; a humanoid production contract is not established by this source.

Huandong develops precision robot transmissions within Shuanghuan. The cited disclosure is not evidence of humanoid market-share leadership.

Kaydon documents thin-section and slewing bearings for robot mechanisms, including waist positions. No humanoid-specific share is stated.

SMC sells electric and pneumatic robot grippers, including elastic fingers. These are industrial end effectors, not five-finger humanoid hands.

Industrial imaging and depth-sensing products support robotic perception; this listing represents the diversified Sony parent.

Sunny lists robot camera modules and optics. Module integration and optical manufacturing are supported, but named humanoid shipments are not.

TE documents force, torque, and position sensors for robots and cobots. Finger-force relevance is supported; a specific humanoid customer is not.

Teledyne documents industrial imaging systems and cameras. Complete camera products do not establish supply of the illustrated optical lens stack.

Ball screw/spline units support linear and rotary motion. They are not planetary or strain-wave gear reducers.

TKH’s vision portfolio addresses machine vision and automated processes. Broader industrial imaging is distinct from verified humanoid deployment.

The official catalogue lists planetary reducers and geared motors. This is a suitable gearbox-architecture match, without an asserted humanoid customer.

The motor-control catalogue supports an enabling-electronics link. It does not establish robot battery management or humanoid contracts.

The source establishes a six-axis IMU for precision positioning. Balance relevance is inferred; it is not a six-axis force/torque transducer.

K-series frameless motors are a relevant actuation technology. The cited page does not identify humanoid customers or robot-specific shipments.

ROHM documents industrial motor and power-conversion electronics. The link is component capability, without an identified humanoid customer.

Samsung manufactures LPDDR memory. This component entry does not represent Samsung SDI batteries or the separate Rainbow Robotics business.

LPDDR products support low-power on-device AI. Their use in a particular humanoid is inferred, not established by the cited mobile-memory release.

The source verifies global-shutter CMOS products. Humanoid adoption is an application inference, not a confirmed customer relationship.