The components and engineering that turn commands into useful physical work.
AS OF 2026-09-13DATED · NON-EXHAUSTIVE
Supply reviewed 2026-09-13. “Supply unclear” means the reviewed evidence does not settle node-wide availability; “Mixed supply” means local tightening coexists with expansion or headroom. Neither means a universal shortage. Expansion and pricing summaries are our assessments, explained and sourced in each node below; an unknown expansion time is not assigned a standard number of years.
Motion in Physical AI
Intelligence still needs an actuator
A robot can understand a task and still fail to perform it. Something must turn electrical or hydraulic power into a movement, carry the load, measure the result and repeat it without overheating or wearing out. Motion is that physical execution chain. It includes the components, the engineering that makes them work together, and the production work that turns a successful design into dependable assemblies.
That makes Motion a cross-family enabling market, but not a single interchangeable product. An industrial wrist, a humanoid knee and a warehouse wheel can draw on related motors, feedback devices, bearings and control electronics. Their loads, duty cycles and packaging differ. An agricultural steering circuit or an aircraft propeller preserves a more distinct route. The useful commonality is often a supplier's engineering and manufacturing capability, not a part number that every machine must buy.
There is concrete evidence of this overlap. maxon's robotics applications range from industrial and mobile robots to legged machines, including customized drives for ANYmal. Novanta's motion stack brings together motors, drives and feedback. These demonstrate participation across applications and adjacent component roles. They do not establish a universal supplier, a shortage, or the economics of any particular robot programme.
Our read: a supplier can carry accumulated engineering and manufacturing know-how into several machine markets. The opportunity depends on that knowledge remaining difficult to reproduce as designs and production methods become cheaper.
What changes as machines become useful
There are several ways demand can grow, and they should not be collapsed into one robot-unit forecast. A new application can add machines; a more dexterous machine can add controlled axes; a harder duty cycle can require more capable components; and a builder can outsource an actuator it previously assembled internally. These create different revenue opportunities. The last changes the supplier's scope even if the machine count is unchanged.
The reverse also matters. Making an existing industrial arm more adaptable does not automatically add another motor to it. Adding autonomy to a truck need not replace its entire drivetrain. A humanoid designer can trade an expensive high-ratio joint for a larger motor and a different transmission. Improvements that make a robot affordable may lower the value of components in each machine even while enlarging the market. The investment question is which supplier sells more useful work after those changes, not which catalogue contains the most robotics vocabulary.
The upper component categories recur across controlled electric axes. The lower transmissions depend on the chosen architecture. The three assemblies are different purchases, not stages in one production line.
The shared connector abbreviates repeated component requirements; it is not a kit, supplier or common factory. Drive electronics may be embedded or external. Dashed paths are design-dependent: direct drive can omit reduction gears; a linear motor or belt can replace a screw. Matching, assembly and testing occur within each supplier or machine builder's delivered scope.
The electric-motion components
These are economic supply nodes, not a mandatory shopping list. Some are purchased separately, some arrive inside an actuator, and some are made by the machine builder. A supplier can occupy several nodes. Adding component sales to the full actuator's selling price would double-count the same hardware.
Electric motors
Supply unclear
Role
Feeds electric axes
Expansion
Winding, magnets and thermal validation
Pricing power
Application-specific; premium unproven
Why it exists
The motor converts electricity into torque. A packaged motor includes mechanical structure that helps locate and support it; a frameless motor supplies the rotor and stator for integration into the customer's mechanism. The latter saves the designer from accepting a finished motor's geometry, but leaves important mechanical and thermal work with the buyer. Kollmorgen's frameless-motor explanation makes this distinction explicit.
maxon (private): compact motors and drive-system integration across several robotics applications.
Yaskawa: Sigma servomotors, including rotary and direct-drive designs, alongside the amplifiers that control them. Its industrial automation range is a different purchase from a complete robot joint.
Worth understanding
The relevant performance is what the motor can sustain in the actual joint, not just its short-duration peak torque. A compact actuator with a poor heat path can require a larger motor, reduced duty or additional cooling. Conversely, better packaging can improve useful performance without a new electromagnetic principle. That is why the ability to adapt a proven motor family and support customer integration may be more valuable than a striking prototype specification.
The shared demand is for controllable torque in constrained space. It does not imply that an aircraft propulsion motor, a precision wrist motor and a wheel motor use the same production process or achieve the same margins.
maxon's 24 June 2026 review describes magnet-related disruption in 2025 that had since eased, alongside renewed orders and capacity investment. This identifies an input risk and a supply response, without settling current availability for a particular motor. Our expansion assessment is that more winding and assembly capacity must be matched with reliable magnet supply and thermal validation. The potential premium is for meeting the customer's duty in a constrained space; the reviewed disclosures do not isolate its size.
A drive meters electrical power into the motor using feedback and control loops. It sits between a command and physical torque: an instruction to move is not itself the current needed to produce that movement. Its interfaces must agree with the motor, feedback device and machine controller. onsemi's robotics architecture identifies the motor's inverter, power switches, gate drivers and position sensors; these are physical control inputs rather than another language model.
Relevant businesses
Novanta: servo drives alongside motors and feedback components; the breadth can reduce compatibility work.
maxon (private): integrated compact drives, illustrating the alternative of buying motor and electronics together.
Yaskawa: SERVOPACK servo amplifiers, including dual-axis EtherCAT drives for rotary, linear and direct-drive motors.
Worth understanding
Physical AI can make a compact, well-integrated drive more useful where electronics must fit inside a moving joint. But this is not the same market as the computer running a perception model. Much of the value concerns predictable response, noise, heat and integration, rather than the ability to execute a larger neural network. At this boundary, “more AI compute” and “more motion electronics” have different demand mechanisms.
Our inference is that a design win may persist when replacing the drive entails retuning and verification. That is a potential switching cost within a design, not proof that the next generation must retain the same supplier. Standard interfaces can simultaneously make adoption easier and reduce dependence.
Yaskawa's July 2026 results, page 8 report stronger AC servo and controller sales across regions, supported by semiconductor, electronic-component and machine-tool demand. That establishes a recovering market, not a saturated servo-drive line. Expansion requires electronics supply, production testing and firmware support for the chosen motor and interfaces. Our pricing assessment rests on the retuning and validation cost described above; neither node-wide lead times nor a servo-drive-specific price premium is established by the segment results.
Motion feedback tells the controller what the mechanism actually did. An encoder measures position or rotation; force and torque sensing can reveal loading and interaction. They solve different problems and are not both mandatory in every axis. Internal joint feedback also differs from a camera observing the outside world, even though both belong to the broader sensing market.
Relevant businesses
Renishaw: optical position measurement, including the readhead, scale and signal-processing relationship.
Novanta: encoders and force/torque sensing within a broader robotics component portfolio.
HEIDENHAIN (private): motor and joint-position encoders, including dual-feedback designs and secondary encoders measuring downstream of the gear.
Worth understanding
Renishaw distinguishes accuracy, resolution and repeatability: reporting smaller steps does not establish that the reported position is closer to reality. Nor does accurate motor-angle measurement eliminate downstream gearbox compliance, structural deflection or calibration error. A claim about encoder resolution therefore cannot be carried straight through to the robot's task accuracy.
There is a potential content opportunity if machines need additional measurements to interact reliably with varying loads. But direct sensing competes with estimation. Novanta's 2019 virtual-torque paper describes using calibrated motor current and gear compliance to estimate joint torque without adding a torque sensor at each joint. This is a design method, not proof it replaces direct measurement in every task. The economic implication is that better interaction can increase the value of calibration and integration while avoiding some discrete sensor purchases.
Renishaw's FY2026 interim presentation, page 7 shows stronger optical and magnetic encoder demand, especially from semiconductor and electronics equipment, while product mix weighed on Position Measurement's margin. This is evidence of differentiated products with different economics. It does not settle robot-feedback capacity. Our inference is that precision manufacture, calibration and reliable integration make expansion more involved than adding generic sensor assembly; the required time and a node-wide pricing premium remain unresolved.
A transmission changes motion; a bearing supports the mechanism and its loads. Combining the functions in one assembly does not make them the same function. THK's cross-roller rings, for example, support radial, axial and moment loads within one bearing arrangement. Its application list extends beyond robot joints to precision tables, medical equipment and semiconductor machinery.
Relevant businesses
THK: cross-roller bearings that support radial, axial and moment loads in robot joints and other precision equipment.
NSK: ball and roller bearings for robot manufacturers, with bearing selection varying by joint load, speed and packaging.
Schaeffler: XZU double-row angular-contact needle roller bearings developed for lightweight robots and cobots.
Worth understanding
Brakes are a distinct, conditional input discussed here for context, not part of the bearing node. Some axes must hold a load when power is removed; others do not need a separate holding brake. A holding brake is also not automatically suitable for repeated dynamic stopping. mayr's brake-system guidance distinguishes those operating requirements. A design can buy its brake separately or inside an actuator; the map does not imply a compulsory brake purchase.
For the related holding function, mayr (private) supplies safety brakes. That is a different supplier role from bearing manufacture.
The economic significance can exceed a component's price when poor load support or holding performance compromises the whole machine. That can justify careful qualification. It does not, by itself, show concentrated supply or permit the vendor to charge whatever the avoided failure would cost.
The current financial evidence is broader than this node. THK's August 2026 results cover industrial linear-motion products; their announced Dalian expansion specifically targets ball screws, so it cannot be counted as additional robot-bearing capacity. For bearings, our expansion assessment concerns precise finishing, consistent preload and qualification under the intended loads. Those requirements can support a qualified supplier's position, but the reviewed material does not establish current capacity utilization, an expansion clock or pricing power for robot bearings.
A reducer lets a faster motor drive a slower, higher-torque output. Choosing one involves more than reduction ratio: the design must tolerate the load, fit the available space and deliver the required motion over its working life. Strain-wave, cycloidal and planetary designs are alternatives with different implementation trade-offs, not successive generations that every robot traverses. Schaeffler's December 2025 actuator presentation includes all three in its rotary-joint design space.
Nabtesco: established precision reducers, with its February 2026 presentation distinguishing its medium/heavy industrial-robot base from expansion toward smaller applications.
Schaeffler: an August 2026 announcement of a formed strain-wave gear, with mass manufacture planned for 2027, not established at this report's date.
Leaderdrive: strain-wave gear families spanning cup, hat, hollow-shaft and compact configurations. Its separately offered actuators include more of the working axis.
Worth understanding
This is a promising area for cross-family reuse of precision manufacturing. Harmonic Drive Systems describes thin-metal processing, small-module gear processing, polishing and measurement among its core capabilities. The inference is that repeatable production depends on an accumulated process system, not merely possession of the gear geometry. Establishing the size of that advantage would still require comparative cost, yield, life and customer evidence.
Two limits matter. Direct-drive architectures can remove the reducer altogether. Within geared architectures, more competing designs and cheaper production can expand unit demand while reducing price per joint. A valuable process position and an unlimited scarcity story are very different claims.
Supply is mixed. Nabtesco's Q2 2026 presentation, page 7 shows Tsu utilization at 85% and Changzhou at 100%, with Hamamatsu expanding output, on a reference basis of 20 days a month and three shifts. These are Q2 plant observations, not September availability by specification. Harmonic Drive Systems' August briefing, printed page 15, describes full North American production alongside unused Ariake floor space. Our inference is that local constraints coexist with room to expand; empty space still needs equipment, staff and qualification. Management also expects robot affordability to put pressure on gearbox costs. Specialized manufacture can be valuable while customers demand cheaper joints.
A screw mechanism converts rotary input into linear travel. It can produce a straight-line actuator or move a linkage that rotates a joint. A linear mechanism is therefore an alternative route to some of the same machine functions, not simply another component to add to every rotary joint's bill of materials.
Relevant businesses
THK: ball screws and related linear-motion products.
Moog: planetary roller screws, including application-specific designs.
NSK: precision-ground ball screws in standard and application-specific designs, supplying the transmission rather than necessarily the complete actuator.
Worth understanding
Load, travel, speed, stiffness, life and installation determine the suitable design. Ball screws and roller screws should not be ranked by name alone. Moog's technical comparison explicitly shows why application and configuration matter; THK's ball-screw/spline combinations illustrate established uses in SCARA and assembly robots, not a market invented by humanoids.
If a robot generation adopts more linear actuation, the opportunity migrates toward these mechanisms and their integration. It does not necessarily increase the addressable market for the reducer used in the displaced rotary design. Nor does prototype interest establish that the chosen screw is available at the price, yield and life needed for volume machines.
Expansion differs between stocked and bespoke products. NSK's catalogue distinguishes standard stocked ball screws from custom designs; this is an offered supply model, not a current inventory check. THK's August 2026 plan, printed page 17 schedules completion of a Dalian ball-screw plant extension for December 2027. That is a specific construction milestone, not a universal screw lead time or confirmed qualified output. Our inference is that specialized designs can retain value through load, life and fit requirements, while standard products face easier substitution. Current availability across ball and roller screws remains unresolved.
A supplied actuator is more than its bill of materials. Its supplier takes responsibility for a defined combination of mechanical fit, motion performance and repeatable manufacture. The customer still owns the surrounding machine. The engineering and production work below explains that boundary; it is not a common factory or a separately established services market through which every actuator must pass.
Engineering the assembly
This work translates a task's load and motion requirements into a compatible mechanical, electrical and thermal design. It decides what the customer can buy from catalogues, what needs adaptation, and what must be developed specifically. The result is a design that can be built and verified, not yet a stream of qualified units.
A joint must supply the required torque at the required speed for long enough to complete the task. Gearing exchanges speed for torque, but introduces its own losses, compliance and mechanical limits. Removing the gearbox shifts more of the job to the motor, drive and thermal design. Moving weight toward the end of an arm increases what upstream joints must move. Optimizing one component in isolation can therefore make the complete machine worse.
Kollmorgen's humanoid design guide treats joint geometry, gearing and heat dissipation as coupled choices. Motor heat must be considered alongside adjacent feedback devices, gearing and other joint materials. Our inference is that an apparently superior standalone part can lose its advantage once the complete assembly's constraints are included. Compatibility work has economic value because customers purchase useful motion, not a sum of isolated specifications.
Who owns this work changes who captures that value. A machine builder with the scale and expertise to design its own joint can buy components directly and keep the integration knowledge. A smaller builder may prefer an actuator vendor that accepts more of the interface and performance responsibility. A broad component supplier can support both routes, but its addressable revenue differs.
Moog's electromechanical offering makes this concrete: it manufactures building blocks internally and offers packages combining a linear actuator, servodrive and software. Engineering sits within the delivered product's scope. It should not be assigned a separate revenue stream merely because the work is important. Repeat use of a design platform and the responsibility the customer actually buys determine whether that knowledge scales economically.
Producing consistent units
Making one working actuator and making many consistent actuators are different achievements. The production route must control alignment, tolerances, wiring, assembly sequence and the measurements used to accept the finished unit. Kollmorgen's frameless-motor manufacturing guidance discusses assembly fixtures and the magnetic forces encountered while installing rotor and stator. These are practical production obligations left by a component-level purchase.
For investors, nominal capacity is only a starting point. A line may be able to process more pieces without being ready to supply a customer's qualified mix. Conversely, a difficult product can have ample available capacity after prior investment. To argue that production is binding requires evidence about the actual constrained operation, accepted output, lead times and the time needed to add or qualify capacity.
Potential advantages include stable processes, test know-how and the ability to ramp a customer without a large rise in failures or rework. Those advantages can reside with a specialist, an actuator assembler or the machine producer. The right question is where yield and acceptance responsibility sit, rather than whether the company calls itself a robotics business.
Evidence of a prototype or an announced factory is not evidence of repeatable shipped volume. Equally, the absence of public yield data is an unknown, not permission to assert a production bottleneck.
Rotary actuators
Supply unclear
Role
Supplies an angular-motion assembly
Expansion
Assembly, testing and customer validation
Pricing power
Integration value; make-versus-buy risk
Why it exists
A rotary actuator packages the components needed to produce controlled angular movement. Depending on the product, it may include a motor, transmission, supporting bearings, feedback, brake and drive electronics. The package boundary is commercial as well as physical: a component kit and a tested, controlled joint transfer different amounts of work and responsibility to the buyer.
Relevant businesses
Harmonic Drive LLC, part of Harmonic Drive Systems: rotary actuators and direct-drive products alongside separately offered gears and component sets. The catalogue establishes different purchase boundaries, not identical content in every actuator.
Leaderdrive: KAH hollow-shaft rotary actuators with specified reduction ratios, encoders and brake options. The catalogue distinguishes the assembled product from a standalone reducer.
Schaeffler: rotary-actuator development and supply for Hexagon Robotics' AEON shoulders and elbows, announced in April 2026. The agreement establishes a supplier role, not proof of scaled shipments.
Worth understanding
Harmonic Drive LLC's product range separates component sets, gear units, actuators and direct-drive products. The US company is a Harmonic Drive Systems subsidiary, not a separately listed issuer. This range is a useful antidote to treating every “joint” as the same revenue unit. A supplier selling an actuator can gain content that a component supplier does not, while also taking on more purchased inputs, assembly cost and integration risk.
Arms and humanoids can reuse aspects of rotary-actuator engineering, but there is no single ideal joint. A design that prioritizes static stiffness and precise positioning may differ from one that must tolerate contact, remain backdrivable or move a leg efficiently. Our inference is that replacing an actuator can entail mechanical redesign, controls work and revalidation in addition to its purchase price; the extent depends on how closely the replacement matches the existing interfaces.
Our inference is that qualified modules can be valuable where they save scarce customer engineering time. The countervailing force is make-versus-buy: once volumes justify internal design, the builder may retain that work and buy components instead. A module's broader scope is not automatically a more durable business model.
Harmonic Drive Systems' August 2026 briefing, printed page 17 provides a concrete boundary: management says its mechatronics products earn more profit per unit than gearboxes alone, but a slightly lower profit margin. Broader supplied scope can therefore increase profit dollars without raising the margin percentage. This company-level comparison does not establish the economics or available capacity of humanoid joints. Our expansion assessment includes matched assembly, end-of-line testing and customer validation; its duration remains programme-specific.
An electric linear actuator integrates the drive with a mechanism that produces controlled extension or travel. Its useful specification concerns the complete force-and-motion duty, including how the mechanism is supported and connected to the load. The screw alone is not the finished actuator, and a linkage's geometry changes the relationship between actuator force and joint torque.
Relevant businesses
Moog: electric linear servoactuators and matched drive/software packages; its Construction business supplies integrated cylinders for off-highway equipment.
Thomson, a Regal Rexnord business: Electrak HD ball-screw actuators with onboard electronics and configurable control interfaces. The offered configurations establish product scope, not suitability for every robotic duty cycle.
Ewellix, part of Schaeffler: electromechanical linear actuators, including the EMA-100 with ball- or roller-screw configurations for construction machinery.
Worth understanding
This is an established product boundary, not just a humanoid concept. Moog Construction's electric cylinders combine motor, gearbox and ball screw for off-highway machinery. Their application engineering and manufacture sit with the supplier, while the machine builder integrates the load and controls. That is a different purchase from a bare screw. Schaeffler's December 2025 presentation adds a humanoid roller-screw concept; it does not establish comparable delivery or adoption.
Convergence here is with existing precision linear-motion capabilities. Schaeffler's April 2025 Ewellix EMA-100 announcement describes an actuator designed to replace hydraulic cylinders on small construction machines, with ball-screw and roller-screw configurations. This is a concrete route by which value can migrate between the electric and hydraulic chains. It establishes a supplied design, not displacement at scale. Geometry, loads, endurance and packaging still prevent an existing industrial actuator from becoming a drop-in humanoid joint.
The economic test is whether the supplier earns a repeatable position in the chosen architecture. If the customer changes linkage or moves to a rotary design, the unit-content thesis changes. If the supplier offers both routes, it may retain the customer but sell a different mix of products. Neither outcome is captured by simply multiplying a robot forecast by a fixed screw count.
The public segment data do not isolate this product. Moog's July 2026 earnings presentation, pages 6–7 attributes Industrial growth to data-centre cooling pumps and much of its margin improvement to tariff refunds. Neither is evidence of scarce electric linear actuators or their pricing power. Our assessment is that expansion must match screws, motors and load-bearing mechanics, then demonstrate endurance at the customer's duty. Application fit can justify paying for an integrated unit, but a durable premium and the time to add qualified supply remain unknown.
A mobile robot's drive can combine motor, transmission, controller and wheel into a supplied assembly. maxon's Wheel Drive data sheet describes such a package with network-interface options. The same commercial boundary issue applies: a wheel motor, a drive module and a complete mobile robot are different products.
The task is sustained traction and controllable movement under the vehicle's loads, rather than reproducing an arm joint's motion. Wheel and floor interaction, installation and vehicle dynamics remain part of the complete machine's design. Buying a module does not deliver navigation, obstacle avoidance, fleet software or a working warehouse material flow.
Relevant businesses
maxon (private): Wheel Drive modules combining motor, transmission, electronics and wheel, with specified network interfaces. It supplies the traction assembly, not the robot's navigation or operating service.
Siemens: Simotics E ArgoDrive driving-and-steering units for automated guided vehicles. The range joined Siemens through its July 2025 acquisition of ebm-papst's industrial-drive business.
Worth understanding
Managed-site and public-facing mobile robots may draw on related compact-drive expertise. Their operating environments and service obligations can still differ substantially. More applications can broaden a supplier's engineering platform while leaving individual configurations fragmented.
The opportunity is strongest if reuse reduces the engineering cost of each new customer without turning the product into an easily substituted commodity. The evidence would be repeat programmes, common qualified platforms and a profitable mix of standard versus custom work. A long list of possible robot applications establishes none of those economics by itself.
maxon's June 2026 review describes a move toward integrated mechatronic systems and capacity additions, but does not disclose Wheel Drive utilization or margins. Our inference is that expansion depends on repeatable module assembly and validation in the customer's vehicle. Common interfaces and reusable designs can reduce engineering cost per programme. Whether suppliers keep that saving as profit, pass it to customers, or lose the module purchase to in-house integration is not resolved by the current public evidence.
Choose a supplied electrohydraulic actuator, or integrate hydraulic components directly into the machine. Propulsion, steering and braking retain their own product boundaries. Diamonds mark machine functions, not additional supply nodes.
Dashed hydraulic paths show the packaged-actuator option; solid outer paths show component integration by the machine builder. These are functional relationships, not verified sales between the named companies. Pumps and fluid connections remain necessary. Propulsion parts and vehicle integration are explained in the report without catch-all component boxes.
Electric precision joints do not stand in for all Physical AI. The following products expose different supply relationships: hydraulic components can be bought separately or within a supplied actuator; an aircraft builder can buy a matched propulsion unit; steering and braking are distinct automotive systems. These are selected commercial boundaries, not exhaustive bills of materials. Unmapped inputs remain necessary.
Hydraulic cylinders
Supply unclear
Role
Applies hydraulic force
Expansion
Machining, sealing and endurance testing
Pricing power
Duty-specific; premium unproven
Why it exists
A cylinder converts fluid pressure into force and travel. Its construction must carry the load, contain pressure and maintain sealing through the required duty. It is a supplied working component, not the hydraulic pump, valve or complete machine. Parker's mobile-cylinder range distinguishes welded-rod and telescopic products for equipment including material handling and mining.
Relevant businesses
Parker Hannifin: mobile hydraulic cylinders, including welded-rod and telescopic designs. This evidence establishes cylinder supply; it is not a proxy for all hydraulic hardware.
KYB: hydraulic cylinders for construction machinery, with the load-carrying cylinder supplied separately from pumps, valves and machine integration.
Worth understanding
The important implication is continuity. An off-road machine can gain sensing and autonomy while retaining its cylinders. The incremental opportunity may sit in controllability and interfaces rather than replacement of every working component. Conversely, additional machine output or a changed load cycle can increase cylinder demand or alter the required design. These are different demand mechanisms.
Electromechanical cylinders are an alternative in some applications, not a mandatory next stage. Changing to a screw-driven design moves value toward motors, screws and controls, while retaining a hydraulic architecture preserves demand for sealing, fluid handling and cylinder manufacture. The comparison concerns the complete duty and installation, not whether one route is inherently “more AI.”
KYB's August 2026 results, pages 2 and 4 report stronger construction-machinery sales within a wider hydraulic-components business. They do not isolate cylinder prices, available capacity or margins. Our expansion assessment concerns machining and finishing the cylinder, maintaining sealing quality and testing its working life. A proven fit to demanding duty can help retain a supplier, while alternative cylinder makers and the customer's electric-actuation option constrain the value it can capture. A node-wide supply rating and expansion time remain unresolved.
This is the interface that converts a command into regulated hydraulic action. A valve controls fluid flow; an electronically controlled valve or valve/controller package connects that work to the machine's control system. It is distinct from the cylinder applying the force. Danfoss's EHi steering product combines a valve and controller with an automatic-guidance interface. It does not contain the whole pump-and-cylinder system.
Relevant businesses
Danfoss (private): PVG proportional-valve products; its separate EHi valve/controller serves electronic steering and guidance interfaces. Neither example establishes that it supplies the downstream actuator's entire scope.
Parker Hannifin: mobile control valves, including load-sensing and pressure-compensated designs for controlling several machine functions.
This distinction identifies a plausible incremental market: a machine can buy more capable control while retaining existing working hardware. But a steering valve and an implement-control valve are application-specific implementations, not interchangeable purchases. The supplier must match flow, pressure, interfaces and fault-response requirements to the intended system.
The equipment builder may integrate the final function, or an actuator supplier may incorporate the controls into its package. Selling the interface can therefore give a vendor content in an autonomy upgrade without making it the supplier of all the hydraulics. A retrofit and a new machine should not receive the same assumed component content.
Danfoss's August 2026 half-year release reports stronger construction and mining markets alongside soft demand for large agricultural equipment. Valve suppliers therefore face several demand cycles, not a single autonomy-led surge. The release does not establish valve-specific utilization or pricing. Our inference is that precise valve manufacture and validated electronic control create work to repeat when capacity or product variants expand. Integration can help retain a supplier; the size of that advantage over another qualified valve remains unproven.
This node is a supplied actuator, not the abstract milestone “the hydraulics work.” A vendor packages hydraulic motion with a defined degree of integration. Bosch Rexroth's industrial self-contained actuators combine an electric motor, pump and hydraulic cylinder, with shut-off and control functions. Their product comparison matters: controls are included in CytroForce but excluded in CytroMotion. Even within one vendor, buying an actuator does not imply the same delivered scope.
Relevant businesses
Bosch Rexroth, part of the private Bosch group: self-contained electrohydraulic actuators, with control scope varying by product. This establishes an actual module-supply role, not autonomous-machine assurance or a shared factory with automotive Bosch.
Moog: electrohydrostatic actuation systems combining an electric motor, pump, manifold and cylinder. Its modular offering also permits a narrower delivered scope, so the actual package must be checked.
Worth understanding
Such a package can replace work the machine builder would otherwise do with separate components and a central hydraulic supply. The vendor earns a broader order and assumes more interface responsibility; the customer still selects and integrates the assembly for its load and operating environment. The map shows functional requirements, not a claim that Parker cylinders or Danfoss steering valves are installed in these Bosch products.
There is also a useful cross-family boundary here. Electrically powered motion need not mean a screw-driven mechanism: an electric motor can instead drive a local hydraulic pump. The architectures compete over the complete duty, packaging and service requirements. Capability can transfer across applications without industrial actuators becoming drop-in robot joints.
Bosch Rexroth's April 2026 review says prior investment left production capacity for a recovery, even as cost and competitive pressure remained high. That company-wide headroom cannot establish availability of a particular self-contained actuator. It does make a blanket hydraulics-shortage claim difficult to support. Our inference is that expansion can use existing manufacturing infrastructure while still requiring package-specific testing and customer qualification. Pricing depends on the engineering and installation work the package saves relative to separate components or electric alternatives; the reviewed evidence does not quantify a durable premium.
In an electric multirotor, a motor, electronic speed controller and propeller work together to produce thrust. T-MOTOR's June 2026 propulsion guide treats their matching as a system problem: a component's operating point depends on the other components and the intended load. The aircraft builder can select parts and own that work, or buy a matched unit. The mapped node is the latter purchase; “motor, ESC and propeller” is its component explanation, not another undifferentiated market box.
Hobbywing: X8 G2 integrated propulsion units with motor, controller and propeller specifications. The documented scope is propulsion, not flight autonomy.
Worth understanding
There is overlap in the general ability to manufacture motors and control their power. That is too abstract to establish the same precision-motion market. A robot wrist's low-speed load handling and an aircraft's thrust production need not share the same motor design, transmission, qualification evidence or supply base. The map therefore does not route aircraft through a mandatory precision reducer.
A supplied unit moves matching and some integration work to the propulsion vendor. That can spread development and testing across several aircraft designs, but the assembly's revenue must not be added to the value of the same motor and controller as if it were extra hardware. Separate component markets still exist; investigating their individual supply economics would require separating the real products, not treating the entire input set as one node.
The customer still integrates the unit into an aircraft with its own structure, energy system, flight controls and operating requirements. Qualified performance at that level matters more than an isolated component's claimed maximum thrust. A common propulsion platform could spread development work across several aircraft designs; conversely, an aircraft maker can choose in-house matching or another package.
Sanrui Intelligence's H1 2026 report, pages 11 and 21 defines integrated propulsion as matched motor, controller and propeller units and reports RMB162.4 million of revenue at a 59.72% gross margin for that product line. This establishes an operating business beyond its catalogue. The same report describes flexible production for varied, often customized orders. Our inference is that maintaining system performance and production consistency across that mix matters to expansion. One supplier's gross margin is not evidence of industry scarcity or durable pricing power; node-wide utilization and an expansion time remain unresolved.
A road vehicle already has mechanisms that steer and brake before it gains autonomy. The change is in how commands are generated, transmitted and executed, and how the vehicle responds to faults. Treating autonomy as an entirely new purchase of all underlying vehicle hardware would overstate incremental demand.
A platform that already has electronically controllable actuation differs from a retrofit requiring substantial hardware changes. By-wire systems also serve driver-operated vehicles; removing a mechanical driver connection is not unique to autonomy. The relevant comparison is what the target platform adds, replaces or reorganizes. Steering and braking are kept separate below because they control different functions and involve different product and supplier choices, not successive stages in one system.
Steer-by-wire systems
Supply unclear
Role
Executes vehicle steering
Expansion
Vehicle-programme launch and validation
Pricing power
Programme-specific; OEM bargaining
Why it exists
Steer-by-wire replaces the mechanical connection between a driver's steering input and road-wheel actuation with electronic commands. The supplied system must produce steering force and handle specified faults; designs with a driver interface also reproduce steering feedback. This is automotive systems engineering and qualification, not a compact industrial servo sold into a new enclosure.
Relevant businesses
Nexteer Automotive: steer-by-wire systems, with its first two production launches reported in its August 2026 results.
Bosch (private): steer-by-wire within its act-by-wire portfolio, describing redundant logic, power and communication. Its offering is separate from its brake-by-wire system.
ZF (private): steer-by-wire systems combining a handwheel actuator with a front-axle steering actuator and electronic control.
Worth understanding
Nexteer's 12 August 2026 results report its first two steer-by-wire production launches, in China and Mexico. This is evidence of actual programmes, without establishing broad autonomous-fleet adoption or the size of the resulting market.
The potential advantage is a qualified position in a demanding vehicle programme. Replacing it can require changes to actuation, electronics, software and validation, rather than merely selecting another motor. The counterweight is that vehicle makers control architecture and purchasing; a programme win is not permanent control of the next platform. Evidence about qualified alternatives and retained economics must carry that argument.
Our expansion assessment centres on vehicle-programme validation and production launch. Nexteer's August 2026 update includes expanded testing capability, but its new Thailand production line serves column electric power steering, not steer-by-wire; it must not be counted as capacity for this node. Programme qualification can make switching costly, while new-platform bidding leaves the vehicle maker bargaining power. The reviewed disclosures do not isolate steer-by-wire capacity, pricing or margins.
Brake-by-wire executes a braking request transmitted electrically. The wheel-end mechanism need not become fully electric. Bosch's architecture uses two independent hydraulic brake actuators and retains hydraulic pressure at the wheel brakes. Removing the pedal's mechanical command connection therefore does not mean removing all hydraulic components.
Relevant businesses
Bosch (private): a brake-by-wire actuator working with an ESP unit on separate redundant channels, each capable of building wheel-brake pressure in the described fault case. The cited offering establishes system scope, not safety of every vehicle installation.
ZF (private): brake-by-wire offerings spanning integrated electrohydraulic brake control and electromechanical wheel-brake designs. The architectures differ in their use of hydraulics.
That distinction changes the content thesis. Additional electronic control or fault tolerance can create supplier value while reorganizing existing hardware; it is not automatically a wholly new braking bill of materials. The finished vehicle must still meet its operating and fault-response requirements. Neither an important safety function nor redundant parts alone establishes a bottleneck.
The supplier's economically distinct work is delivering and supporting the specified braking system into a vehicle programme. Its performance, interfaces and acceptance differ from steering even when one group offers both. Common ownership can create coordination advantages without making the systems substitutes or requiring the vehicle maker to buy both from the same vendor.
AUMOVIO's 10 September 2026 announcement distinguishes its established one-box brakes from a new distributed system with a first series order and production planned for the end of 2027. This is a vehicle-programme milestone, not a general time-to-capacity estimate. The company also describes reusing one electronic pedal across vehicle architectures to reduce variants and costs. Our inference is that qualification and integration can protect a programme position, while vehicle makers seek those savings. Product generations have different launch requirements; node-wide supply and retained pricing power remain unclear.
The most defensible starting point is a distinction between physical necessity, difficult execution and scarce supply. A motor is necessary in an electric axis. Meeting a difficult duty cycle can require specialized design. Neither establishes that qualified supply cannot expand. An investment thesis has to make the last step explicitly, or explain another way the supplier earns attractive returns without a shortage.
The node evidence points to different economic mechanisms. Reducer suppliers show local tightening and additional capacity. A ball-screw factory extension has a dated construction plan. Integrated propulsion has disclosed product revenue and gross margin. By-wire systems advance through vehicle programmes. None of these observations supports applying one supply rating or expansion clock to the whole branch.
Supply can also improve through process change. Schaeffler's August 2026 formed-gear announcement proposes a cheaper manufacturing route, with series production planned for 2027. That remains a supplier claim and forward plan. If the route works, cheaper gearing could enable more robots while changing what existing suppliers earn per joint. More content per machine raises both the revenue opportunity and the customer's incentive to reduce its cost.
Our interpretation is that three positions merit particular attention. First, a difficult component whose performance can be transferred across several applications without being redesigned from scratch. Second, an integration platform that lowers the customer's engineering burden while retaining enough differentiation to avoid price-only competition. Third, production know-how that turns demanding designs into consistent accepted output. These are hypotheses about how value can be created and retained, not a ranking of the named companies.
The tests differ. For a component specialist, look for actual adoption across customer platforms and why cheaper alternatives fail the relevant requirement. For an actuator supplier, examine what work it takes over, what remains custom and whether customers keep buying the module at volume. For a production advantage, distinguish qualified capacity and yields from factory floor space. Evidence of repeat purchases, stable economics and reproducible performance is more informative than a list of prototypes.
The central risk is not only that Physical AI grows slowly. It is that it grows successfully through architectures, sourcing choices and cost reductions that distribute value differently from today's expectations. A strong Motion thesis must survive that possibility. The search is for businesses that enable cheaper, more capable machines and still have a reason to be paid well for their contribution.
A map of the work, not a shortage claim
This report was researched as of 13 September 2026. Dated disclosures retain their own observation periods; undated technical and product pages were consulted for the capabilities they describe. Supplier materials establish designs, offerings and stated plans. They are not independent proof of comparative performance, customer adoption, capacity tightness or investment value.
The maps stop at motion assemblies and their integration work. Energy supply, scene sensing, model development, complete-machine controls, deployment and service remain necessary but are covered elsewhere in the Physical AI enabling markets and the linked family reports. This is a selected cross-family view, not an exhaustive vendor census or a claim that all machines converge into one product.