Schaeffler's Formed Gearboxes Could Cut Humanoid Cost

Why a Gearbox Is the Bottleneck in Humanoid Cost

A strain wave gearbox (also called a harmonic drive) is the component that lets a humanoid robot's joint deliver high torque in a compact, backlash-free package, which is why nearly every humanoid arm and leg joint uses one. Because every joint needs one and each has to hold tight tolerances, strain wave gearboxes account for roughly half the total manufacturing cost of a humanoid robot, according to Schaeffler.

That cost concentration is the real reason humanoid robots have stayed expensive even as battery packs, motors, and compute have gotten cheaper on their own separate cost curves. A robot maker can source a commodity motor or battery from several suppliers competing on price; a precision strain wave gearbox has historically come from a small number of specialist manufacturers using slow, machining-heavy processes.

Conventional strain wave gearbox manufacturing relies on precision machining, cutting a component from solid stock in a process measured in minutes per part. That approach scales to research-lab and low-volume humanoid production, but it does not scale to the tens of thousands of units per year the industry is targeting for actual commercial deployment.

What Schaeffler's Forming Process Changes

Schaeffler's alternative is a forming process: shaping the gearbox's key components under high pressing force rather than machining them from solid stock. The company says forming completes in seconds what conventional machining takes minutes to do, while achieving comparable torque and efficiency to the machined part it replaces, and improving dimensional accuracy and process stability in the process.

The stated results are a manufacturing-cost reduction of more than 25% and a material-consumption reduction of more than 75% compared with the machined equivalent. Material reduction of that scale usually comes from forming displacing metal into shape rather than cutting most of a solid billet away as scrap, which is consistent with how forming processes behave in other high-volume metal-parts industries.

Schaeffler is not new to forming at scale: the company says it has already supplied more than 2 million formed strain wave gearboxes to the automotive sector over the past decade, a track record it is now applying to a second application with a very different cost target.

Timeline and Where Production Starts

Schaeffler has completed validation testing on the formed gearboxes and set 2027 as the start date for mass manufacturing. Production begins in Germany, then rolls out to other regions, a staged approach consistent with getting a new manufacturing line running at one site before replicating it elsewhere.

David Kehr, president of Schaeffler's humanoid robotics division, said the company deploys humanoids along its own global value chain and has first-hand knowledge of the technology requirements involved, framing the gearbox program as informed by Schaeffler's own use of humanoids in its factories rather than a purely external supplier bet.

A 2027 start for mass manufacturing means the near-term supply of humanoid robots through 2026 and into 2027 will still rely on the conventional machined gearboxes it is meant to replace. The cost impact of this shift will show up in humanoid pricing on a lag, not immediately.

Why Forming Beats Machining at High Volume

Machining removes material to reach a final shape, which is why it produces so much scrap on a part this intricate: a strain wave gearbox's flex spline and wave generator need precise, thin-walled geometry, and cutting that shape from a solid billet means discarding most of the starting material as chips. Forming instead pushes material into the target shape under pressure, so more of the starting stock ends up in the finished part rather than on the shop floor as waste.

That difference compounds at volume. A machining line adds cycle time roughly linearly with part count, since each part still needs its own multi-minute cutting pass. A forming press, once tooled for a specific gearbox geometry, can complete the same shaping step in seconds because it is pressing the whole shape at once rather than removing material pass by pass, which is the core reason Schaeffler can credibly target tens of thousands of units a year instead of the low-volume runs machining tops out at.

The tradeoff is upfront tooling cost and qualification time: a forming die has to be engineered and validated for each specific gearbox geometry before it can run at volume, which is exactly the multi-year validation-to-production timeline Schaeffler has been running since it demonstrated the process and set 2027 as the mass-manufacturing start date.

What This Means for Humanoid Robot Buyers and Watchers

A component-level cost cut this large, if it holds at commercial volume, is one of the more concrete paths toward humanoid robots reaching price points closer to industrial cobots than to today's six-figure research platforms. Gearbox cost alone will not get there; battery, sensor, and integration costs still have their own curves to travel.

The more durable signal is which humanoid makers actually adopt Schaeffler's formed gearboxes once mass production starts in 2027, since a cost advantage at the component level only shows up in a finished robot's price if a manufacturer switches suppliers or renegotiates on the strength of the new option. Track that adoption, not the announcement itself, before crediting any specific price drop to this change.

• Ask any humanoid vendor citing lower prices in 2027 or later whether the reduction traces to a specific component change like this one, or to volume pricing across the whole bill of materials.
• Treat 'mass manufacturing starts in 2027' as a supply-chain milestone, not a shipping date for robots built with the new part.
• Compare cost claims against Schaeffler's own stated baseline (conventional machining), not against an unstated one.