Planetary Reducers

Planetary reducers are the rare joint component that scales ahead of the build ramp, there is no global deficit. Unlike roller screws or harmonic drives, a planetary gear is made by standard spur and helical gear cutting on the broadest machine base in the whole reducer family², with no flexspline-grinding bottleneck to gate it. The broad precision-planetary market already turns out on the order of fifteen million reducers a year across robotics, automation and industry², a deep, elastic reservoir the humanoid build draws on, even though the robot-grade high-precision tier proper is smaller (about US$466 million, on the order of a million units a year)³, and because the gear-cutting base is broad and elastic, output can ramp on ordinary brownfield line adds rather than waiting on scarce precision grinders. The figures here track the whole global humanoid fleet, not one country’s, and at roughly nine planetary reducers per robot¹ (revised up from five, as China’s rotary quasi-direct-drive legs and waist default to planetary stages) that fleet needs about four hundred and twenty-five million reducers by 2040¹⁰, yet cumulative global production over the same window delivers well over a billion, leaving production about three times ahead of need. The durable crossover, the year cumulative production drops below cumulative need, never arrives globally, and that global no-deficit result is the robust finding. Building from China alone is a softer, modeled margin that now narrows: at nine reducers per robot China-only output covers the fleet only into the late 2040s, crossing around 2048⁶, so we read it as lower-confidence, not a co-equal finding. The thin United States niche⁷ falls behind sooner, around 2032, meeting well under half the 2040 need. Treat planetary as a pressure-release valve for the strain-wave and roller-screw chokepoints, not a constraint in its own right.

A planetary gear reducer multiplies a motor’s torque through planet gears orbiting a central sun gear inside a fixed ring gear, a compact, coaxial, moderate-ratio stage. In a humanoid it sits in the body quasi-direct-drive joints, the heavy-load lower-limb and waist joints⁹, and in some hand and linear-actuator gearheads, while the high-ratio rotary joints typically use harmonic drives instead. A budget design like Unitree’s G1 is planetary-heavy, running a two-stage planetary reduction across all twenty-three of its joint modules¹; mainstream humanoids mix harmonic and planetary, so we model about nine planetary reducers per robot as a design-mix anchor, revised up from five as the rotary quasi-direct-drive designs that dominate the shipped fleet default their leg and waist stages to planetary gearing (the earlier “twelve” conflated Optimus hand worm and lead-screw drives that are not planetary at all). The grade that matters here is the robot- and automation-grade precision planetary reducer, servo-class, with controlled backlash, not the ultra-high-precision sub-one-arcminute niche of only about nineteen thousand units a year⁵, and not the full commodity industrial-planetary base of tens of millions of units behind it⁴.

Because a planetary reducer is built into a robot for life, the reducers the fleet needs are a cumulative stock, the total number of robots ever built worldwide multiplied by about nine¹, not a per-year flow. Integrating the consensus shipment ramp¹⁰, the global installed fleet reaches about 585,000 robots by 2030, 10 million by 2035 and 47 million by 2040, requiring roughly 5.3 million, 91 million and 424 million precision planetary reducers respectively. In the chart below, the black line is that worldwide cumulative need, the grey band is cumulative global production, and there is no red wedge: production stays above need for the whole horizon.

Production is the only thing that differs between the Global, China and US views. The worldwide humanoid fleet and the reducers it needs stay identical in all three; only the supply line changes, asking whether one bloc’s factories alone could equip the whole global humanoid fleet. Real output follows an S-curve rather than a smooth ramp. Global precision-planetary output, the broad reservoir across robotics, automation and industry, not the robot-grade tier alone, runs from about fifteen million units a year today² to roughly forty-five million by 2030 as humanoid and collaborative-robot demand pulls in ordinary gear-cutting lines, then about one hundred and five million by 2035 and one hundred and seventy million by 2040. The precision-planetary market is set to pass five billion dollars by 2027 on its own², and the broad industrial-planetary base behind it, about four-and-a-half billion dollars and tens of millions of units a year⁴, is the elastic reservoir those lines draw on. The post-2030 figures are a modeled extrapolation: a humanoid-driven step-change above the cited five-to-eight-and-a-half-percent organic growth³, gated only by brownfield gear-line adds, not by any scarce-grinder ceiling. China climbs from about three million units a year, roughly a fifth of global precision-planetary production today³, to a modeled sixteen million by 2030 and ninety-five million by 2040, as makers like Ningbo ZhongDa Leader, Newstart, Zhaowei and KOFON build humanoid-reducer lines ahead of binding orders⁶⁸. The United States holds far flatter, from about one million units a year to three-and-a-half million, a niche served by Cone Drive in Michigan, GAM and Onvio⁷, with the servo-planetary leaders sitting in Germany, Taiwan and Japan and no announced humanoid-grade reshoring.

For most components this section names the crossover year; for planetary reducers there isn’t one to name globally. Cumulative global production runs about three times ahead of the fleet’s need at 2040, roughly one-and-a-third billion reducers produced against a four-hundred-and-twenty-five-million requirement¹⁰, so the durable crossover never arrives, the robust finding. Building from China alone is softer and modeled, and the higher nine-per-robot count narrows it: China-only output now covers the global fleet only into the late 2040s, crossing around 2048, and leans on a fully modeled China ramp⁶, so we treat it as low-confidence rather than a co-equal result. The one bloc that clearly falls behind early is the United States: a flat domestic niche of a few million units a year⁷ crosses below the worldwide fleet’s need around 2032 and meets well under half the 2040 requirement, a reshoring gap, not a global supply gap. The contrast with the grinder-bound parts is the whole point: where roller screws and cross-roller bearings cross in the mid-2030s and meet well under half the 2040 need, planetary reducers simply track the ramp the entire way.

Humanoids are not the only draw on the planetary pool, and quadrupeds are the form factor that leans on it hardest. A four-legged robot is a near-pure planetary machine: each of its twelve quasi-direct-drive joints pairs a large-diameter frameless motor with a single low-ratio planetary stage, the MIT Mini-Cheetah / Unitree architecture, so a quadruped carries about eleven planetary reducers, more per unit than a humanoid’s five (a humanoid splits its joints between planetary and harmonic, while a QDD quadruped is planetary almost throughout). By 2040 the quadruped fleet reaches about five million units, adding roughly fifty-seven million precision planetary reducers to the humanoids’ two-hundred-and-thirty-five-million need, about a quarter again on top, while cobots and industrial arms add none (cobots are all-harmonic, industrial arms all-RV). The broad, elastic gear-cutting reservoir still absorbs it: a combined need near three hundred million by 2040 sits far under the billion-plus cumulative output, so quadrupeds raise the draw without opening a global crossover. The one honest tightening is on the narrow robot-grade QDD tier specifically, quadrupeds concentrate their demand on exactly the servo-class, low-backlash planetary slice that humanoids also want, so they are the swing consumer of that sub-tier, the same coupling the elasticity caveat below flags. Drones, by contrast, the largest fleet of all, draw nothing here: a direct-drive propeller couples its motor straight to the prop and needs no reducer of any kind.

What makes harmonic drives tight does not apply here. A harmonic flexspline demands continuous-generating gear-grinding on a thin, mostly European and Japanese machine base, a slow, capital-heavy, qualification-gated step that pins its expansion. A planetary gear is standard spur and helical cutting: ordinary CNC hobbing and grinding that runs on a vast worldwide pool of machines and trades, the same heat-treat and gear-cutting shops but at a far looser precision bar than strain-wave flexsplines or cycloidal cam-disks². The supplier base is broad, Wittenstein leads at about a quarter of the precision-planetary market and the top five hold roughly fifty-six percent², with Chinese entrants a fast-growing long tail³, and the line-add barrier is the lowest of any reducer, so a budget design can drop a planetary stage in precisely where a roller-screw or harmonic drive would otherwise be the chokepoint. One honest caveat tempers the “no barrier” verdict. Precision planetary gear cutting and grinding does draw on the same broad precision-gear capacity, the same gear-grinder OEMs, skilled gear-cutters and metrology, that harmonic flexsplines and cross-roller raceways also use. Those three siblings are operationalized in this model as a shared precision-grinding pool that throttles their effective output; planetary is deliberately not a member of that pool and is modeled standalone here, because its precision bar is looser and its machine base far broader. But that is a known coupling, not a clean separation: if humanoid demand strains the whole precision-gear toolchain at once, planetary’s real shared slice could be tighter than this standalone curve implies. The verdict, elastic, no global crossover, is robust across a wide range of assumptions, but it rests on planetary keeping its broad, lower-barrier machine base largely to itself.