Support Bearings

Support bearings are the commodity half of the bearing family, and on supply they do not bind anywhere, the cleanest yes in this report. The part here is the ordinary deep-groove ball, needle and standard roller support bearing that steadies a shaft or carries a structural load, explicitly not the precision crossed-roller joint bearing, which is its own component (the cross-roller link) and which does bind in the mid-2030s. That split is the whole story, and it lands on opposite sides. A humanoid carries roughly forty-four commodity support bearings¹², so even a worldwide fleet of forty-seven million robots by 2040 needs only about two billion seventy million of them in total⁹. Set that against an industry that already ships tens of billions of bearings every single year, WorldMetrics counts roughly eighty-seven billion bearing units in 2023 across more than eleven hundred manufacturers, with ball bearings about sixty-two percent of volume⁴, and the arithmetic is decisive: a single year of commodity output is about twenty-four times the entire cumulative 2040 fleet need, and cumulative production by 2040 runs something like three hundred and seventy times that need. Run the four-series gap and the global, China and US crossover years are all null⁹: standard bearings are made on a vast, multi-sourced, low-barrier base, the big-eight majors (SKF, Schaeffler, NSK, NTN, JTEKT, Timken, MinebeaMitsumi) hold only about twenty to thirty percent between them³, the rest spread across thousands of makers, with no precision-grinding gate. This is the “same family, opposite verdict” finding: the precision crossed-roller bearing binds because it shares one finite raceway-grinding pool with roller screws and harmonic drives⁸, while the commodity support bearing, drawing on no such pool, never does.

A support bearing is the ordinary, unglamorous rolling-element bearing, a deep-groove ball bearing, needle bearing or standard cylindrical roller bearing, that simply holds a rotating shaft true or carries a structural load, anywhere a part spins or pivots without needing micron-class precision. Every actuator in a humanoid uses several beyond its one precision joint bearing: motor shafts, gear stages, idlers, pulleys and structural pivots all ride on commodity support bearings. Counting roughly two to three sets across each of the twenty to forty rotary actuators¹, plus structural bearings, gives a per-robot figure on the order of forty to a hundred and twenty bearings; the model uses about forty-four commodity support bearings per robot (one-to-two beyond each of Optimus’s twenty-eight actuators²), a deliberately mid count. The crucial distinction is grade: these are commodity-tier bearings, made to ordinary tolerances on automated high-volume lines, not the precision raceway-ground, P4/P2-class crossed-roller rings that go into the actuator output joints. Those precision joint bearings are a separate, supply-tight component (the cross-roller link) and are not counted here; this line is only the easy, elastic commodity half.

Because a support bearing is built into a robot and stays there for its service life, the bearings the fleet needs are a cumulative stock, the total number of robots ever built worldwide multiplied by about forty-four each¹, not a per-year flow. Integrating the consensus shipment ramp⁹, the worldwide installed fleet reaches about five hundred and eighty-five thousand robots by 2030, ten million by 2035 and forty-seven million by 2040, requiring only roughly twenty-six million, four hundred and forty-four million, and two billion seventy million commodity support bearings respectively. Hold those figures against an industry that ships tens of billions of bearings every year⁴: the entire cumulative 2040 fleet need is barely a few weeks of current world output. In the chart below the black line is that worldwide cumulative need and the grey band is cumulative global production, and the band sits so far above the line that no red shortfall wedge ever opens. That absence of a gap is the finding.

Production is the only thing that differs between the Global, China and US views. The worldwide humanoid fleet and the bearings it needs stay identical in all three; only the supply line changes, asking whether one bloc’s factories alone could equip the whole global fleet. Here every bloc wins that contest by orders of magnitude. Global commodity ball-and-roller support-bearing output runs on the order of forty billion units a year today, derived from the roughly eighty-seven-billion-unit total bearing base in 2023, of which ball bearings are about sixty-two percent⁴, with the bearings market itself about fifty-nine billion dollars in 2024 growing around nine percent a year³, rising modestly to perhaps fifty-six billion units by 2040, since unit volumes grow far slower than revenue as the mix shifts toward higher-value tiers. A single maker illustrates the scale: MinebeaMitsumi alone targets about a hundred and eighty million ball bearings a month, roughly two billion a year⁵. China is the world’s largest commodity-bearing producer by volume, organized around five industrial bases, Wafangdian (ZWZ), Luoyang (LYC), Liaocheng, Jiangsu and Zhejiang, where Renben/C&U alone makes more than one and a quarter billion sets a year and Harbin (HRB) about eighty million, atop thousands of smaller makers⁶; we put China near eighteen billion commodity units a year today, rising to about thirty billion by 2040. The US is a thin niche, about four percent of global ball-bearing tonnage, roughly a hundred and sixty thousand tons a year (Timken, MinebeaMitsumi-US/NHBB, RBC, Kaydon-SKF)⁷, which still works out to roughly two billion commodity units a year and rises slowly. The post-2030 figures are a modeled extrapolation of this mature commodity base at about three percent a year, not a reported unit forecast.

For every view, the answer is never. Cumulative global production stays hundreds of times above the fleet’s cumulative need at every year through the 2040 horizon, so there is no global crossover⁹; by 2040 the world has produced something on the order of seven hundred and sixty billion commodity support bearings against a cumulative fleet need of about two billion seventy million⁴, coverage of roughly three hundred and seventy times, with a single year’s output alone about twenty-four times the entire cumulative need. China-only never falls behind either, running on its own about a hundred and eighty-five times the cumulative 2040 need, and, as with stepper motors and harness sets, even the thin US-only niche never crosses: a roughly two-billion-a-year output banks something like seventeen times the cumulative 2040 fleet need on its own⁷. So all three crossover years are null. This is the sharpest contrast in the whole report with its precision sibling. The crossed-roller joint bearing, same broad family, raceway-ground to micron tolerance, binds globally around the middle of the 2030s, with China-only crossing earlier and the US slice earlier still, because it competes for a finite precision-grinding pool⁸. The commodity support bearing, sitting outside that pool, never crosses in any view. Same family, opposite verdict.

Most links in this report bind because some deep node, precision raceway-grinding, suspended-coil winding, heavy rare earths, cannot scale fast enough. Commodity support bearings have no such node, and the cleanest way to see it is the split inside the bearing family itself. The precision crossed-roller joint bearing is supply-tight precisely because its ground-and-superfinished V-races draw on the same scarce precision raceway-grinding lines, metrology and skilled-grinder bench that roller screws and harmonic flexsplines also need⁸, one finite pool, three hungry parts, so it binds early. The commodity support bearing draws on none of that. A deep-groove ball or standard roller bearing is turned, hardened, lap-finished and assembled on fully automated high-volume lines that need no micron-class raceway grinding, so it sits outside the shared precision-grinding pool that throttles its precision sibling, its effective output equals its standalone output, with no pool drag. The manufacturing base is also the most fragmented and multi-sourced in the entire bill of materials: more than eleven hundred manufacturers and sixty-four hundred distributors worldwide⁴, with the big-eight majors holding only about twenty to thirty percent of total unit volume between them³ and China’s five industrial bases adding billions of units more⁶. A new commodity line builds in about a year, the per-unit cost is cents to a few dollars, and the global base already turns out tens of billions of units a year against a cumulative fleet need that never exceeds a few weeks of it. There is, in short, no binding input to map. If anything ever shifted the picture it would be a demand-side surprise, a future design multiplying the support-bearing count many-fold, but even a large multiple of forty-four per robot still vanishes against tens of billions of units a year. Support bearings earn their place for completeness and for the honest record that the commodity base is China-led by volume⁶; on the question this report exists to answer, can the world build enough, they are the clearest yes, and the precise foil to the crossed-roller bearing that shares their family but not their fate.

Cobots carry roughly twelve support bearings per arm (about two per joint) and industrial arms about six, large per-unit counts, but support bearings are a broad commodity class with deep, elastic supply, so both add volume without moving the (non-binding) global picture. Quadrupeds carry roughly twelve commodity support bearings apiece, standard deep-groove and angular-contact units around their quasi-direct-drive joints, not the crossed-roller rings tracked separately, so a quadruped fleet of about 5.15 million by 2040 adds only on the order of sixty million more to a base of tens of billions a year, again without moving the picture. Drones, by contrast, draw nothing here at all: a multirotor’s propeller motors ride on their own tiny commodity ball bearings, not the robot-scale support bearings this link counts, so even the largest fleet in the report by units contributes zero to the tally.