BMW Manufacturing is promoting humanoid robots from trial hardware to assigned work at Plant Spartanburg, South Carolina, after an eleven-month Figure 02 deployment in body shop operations and a fresh commitment to Figure 03 for logistics sequencing. Figure AI chief executive Brett Adcock said the Figure 02 run showed humanoids are “no longer lab experiments” and can support a “flexible, reliable manufacturing workforce,” framing the shift as operational rather than theatrical [E1]. Ulrich Wieland, vice president for production control and logistics, called Spartanburg the “birthplace of humanoid robotics” in BMW’s day-to-day manufacturing and said the group is “looking forward to deploying Figure 03 for a sequencing use case in logistics” [E2].
Nvidia’s Halos for Robotics, described in a 22 June product narrative, targets the certification and assurance layer that humanoids need if they are to share aisles with people instead of staying behind fences. The stack is billed as a full functional-safety system for physical AI, with the vendor arguing that “traditional safety which was built for structured environments can not work anymore as the spaces become more unstructured and robots move out of cages” and that “AI-driven safety is the key” [E3]. That pitch implies the competitive race is splitting: locomotion and manipulation still matter, but gating value may sit in auditable safety logic and integration paperwork.
Factory buyers appear to be weighting cost per productive hour, changeover tolerance, and audit trails more heavily than demo reels. BMW’s sequencing assignment is narrow compared with general-purpose hype, yet it is exactly the kind of repeatable material-handling loop plants already budget for with conventional automation [E2]. If Figure 03 performs on shift schedules and exception handling, humanoids start to resemble line items in workforce planning rather than innovation trophies [E1].
Safety certification and site integration look like the unglamorous bottlenecks that can stall payroll status even when hardware walks convincingly. Halos is positioned so that “safety and productivity are no longer in tension,” a claim that only holds once regulators and internal EHS teams accept AI-mediated stop logic in unstructured layouts [E3]. Without that acceptance, pilots can renew indefinitely while production planners keep humans on the sequencing desk [E2].
Spartanburg’s arc from body shop pilot to logistics sequencing suggests BMW is standardizing a playbook: prove durability in a controlled cell, then migrate to flows where timing errors ripple downstream. Wieland’s language ties humanoid history to this plant specifically, which may help suppliers benchmark reliability claims against a named reference site rather than trade-show footage [E2][E1]. Other automakers and contract manufacturers will likely read the Figure 03 scope as a template for where they might—or might not—authorize capital [E1].
Near-term inference: humanoid vendors that ship safety stacks, documented integration paths, and hourly economics will outrun those selling vision-only autonomy. Nvidia’s Halos entry pressures the market to treat functional safety as productized infrastructure, not a post hoc consulting line [E3]. BMW’s payroll metaphor is still early, but the combination of an eleven-month Figure 02 track record and a named Figure 03 logistics job moves the category from “possible someday” toward “priced shift work,” with certification and factory fit deciding who actually gets hours on the floor [E1][E2][E3].