At BMW’s sprawling plant in Spartanburg, South Carolina, a humanoid robot called Figure 02 has been clocking 10-hour shifts on the assembly line, picking up sheet-metal parts and placing them into fixtures alongside human workers. Over an 11-month pilot that wrapped earlier this year, the robot contributed to the production of roughly 30,000 BMW X3 SUVs, according to Figure AI, the California startup that built it. Now BMW is scaling up, signing a formal commercial agreement with Figure AI and separately deploying a different humanoid robot at a factory in Germany, as the automaker pushes to trim the higher assembly costs that come with building electric vehicles.
From pilot to production contract
The commercial deal, announced through a joint press release, is structured around milestones rather than a single large rollout. BMW and Figure AI will identify specific tasks suited to a general-purpose humanoid, then expand deployment in stages at Spartanburg. Figure CEO Brett Adcock described the robots as a lever to “increase productivity and reduce costs,” though neither company has disclosed dollar figures or savings targets tied to the agreement.
BMW board member Milan Nedeljkovic confirmed that the Spartanburg pilot gave engineers operational data on cycle times, placement accuracy, and how often human workers had to intervene. Those internal benchmarks have not been published, but the fact that BMW moved from trial to commercial contract suggests the numbers cleared whatever threshold the company set. For an automaker that builds more than 400,000 vehicles a year at Spartanburg alone, even modest per-unit savings on repetitive tasks could compound quickly.
A second front in Germany
Separately, BMW announced it will bring the AEON humanoid robot to Plant Leipzig in Germany, marking the first time the company has used humanoid robots in production on its home turf. Developed with Hexagon Robotics, AEON is slated for multifunctional work including high-voltage battery assembly and component manufacturing, according to BMW Group’s official communications.
Battery assembly is one of the most labor-intensive and ergonomically demanding steps unique to EV production. Packs are heavy, connections require precision, and the high-voltage environment adds safety protocols that slow human workers. Deploying a humanoid robot there signals that BMW sees the technology not as a blanket replacement for people but as a targeted tool for the new tasks that electrification has introduced.
The Leipzig and Spartanburg programs use robots from different makers with different designs, so direct performance comparisons are not yet possible. BMW has not said whether the two deployments will eventually converge on a single platform or continue as parallel experiments.
Where the gaps are
The 30,000-vehicle figure is the most concrete data point available, but it comes with caveats. It tells us Figure 02 was active on a live line building a high-volume model; it does not tell us how much labor cost was offset, how many human shifts were replaced, or whether the robot’s error rate met BMW’s internal targets. Both companies reference internal KPI frameworks covering cycle time and accuracy, yet neither has opened those numbers to outside review.
The scope of the commercial agreement is only partially defined. Which vehicle models beyond the X3 will involve humanoid robots, how many units BMW plans to deploy, and what timeline governs each milestone all remain undisclosed. Whether the deal extends beyond Spartanburg to other BMW plants in the U.S. or elsewhere has not been confirmed.
Labor impact is another open question. Spartanburg is a non-union plant in a right-to-work state, which gives BMW more flexibility to restructure workflows than it would have at a unionized facility. Neither BMW nor any independent research body has published data on how humanoid adoption at these plants affects headcount, job categories, or retraining programs. Company statements emphasize that the robots handle ergonomically challenging tasks, but without workforce data, it is impossible to say whether the net effect is substitution, augmentation, or a mix of both.
The competitive landscape
BMW is not the only automaker experimenting with humanoid robots. Tesla has been testing its Optimus robot in its own factories, and Hyundai’s ownership of Boston Dynamics gives it access to some of the most advanced bipedal hardware in the world. Mercedes-Benz has run trials with Apollo robots from Apptronik at its facilities. What distinguishes BMW’s effort as of April 2026 is the combination of a completed, months-long pilot with published vehicle-count data and a follow-on commercial contract. Most rivals are still in earlier proof-of-concept stages or have shared fewer operational details.
That said, the cost-reduction promise in BMW’s announcements rests on strategic positioning rather than audited financials. The logic of automation economics supports the idea that robots performing repetitive physical tasks can lower per-unit labor costs, but the magnitude of those savings in a real automotive environment, with all its complexity, remains unproven in any public data from any manufacturer.
What to watch next
The next meaningful signals will come when BMW either expands the Spartanburg deployment to additional vehicle lines or releases performance data from Leipzig’s battery-assembly trials. Investors and industry analysts will also be watching for any mention of humanoid-robot economics in BMW’s quarterly earnings calls; so far, the company has kept the topic in its manufacturing communications rather than its financial disclosures.
For workers on the plant floor, the more immediate question is whether these robots create new technical roles, such as robot supervisors and maintenance specialists, fast enough to offset any positions they absorb. BMW has signaled that retraining is part of the plan, but details are thin.
What is clear is that BMW has moved past the demonstration stage. The automaker has run a humanoid robot through nearly a year of real production shifts, signed a contract to do more, and opened a second deployment focused squarely on EV-specific assembly. Whether that translates into the cost savings both BMW and Figure AI are banking on will depend on data that, for now, only the companies themselves possess.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
