Brett Adcock's Figure AI has the kind of customer proof humanoid robotics companies have been trying to manufacture for years: BMW Group said Figure 02 supported production of more than 30,000 BMW X3 vehicles over a ten-month period at BMW Group Plant Spartanburg, and Figure 03 is moving next into a logistics sequencing project at the same site.
That is not the same as the viral framing in an X post that Figure AI's humanoid robot "built a car." BMW's narrower claim is more useful because it is measurable: the work happened inside a real production process, not on a trade-show stage.
Figure AI just built a car with a humanoid robot - the physical AI thesis is no longer theoretical
For Adcock, who founded Figure AI in 2022 after building companies in recruiting and electric aviation, the Spartanburg disclosure lands at the exact point Figure AI has been trying to force the market to evaluate: not whether humanoids can look capable in edited videos, but whether they can repeat useful work in places where downtime, safety and cycle time matter. Adcock has been explicit about that ambition from the beginning. In Figure AI's 2022 Master Plan, he wrote that his prior companies were Archer, which he described as a "$2.7B IPO," and Vettery, which he described as a "$100M exit." He framed Figure AI as a 30-year company aimed at general-purpose humanoids for work humans do not want to do. The BMW deployment is the first public example that gives that thesis a production number attached to it.
The claim is operational, not magical
BMW's release says Figure 02 supported X3 production over ten months, and ties that experience to a new Figure 03 project in Spartanburg logistics.
The next task is less photogenic than a robot appearing to assemble a car, but more revealing about the commercial wedge. BMW says the Figure 03 project will focus on parts sequencing in logistics at Spartanburg.
That workflow matters because automotive plants are dense systems of small timing dependencies. A humanoid that can slot into parts sequencing is not replacing an entire assembly line. It is taking a constrained, repetitive, ergonomically awkward piece of work and trying to make it machine-repeatable without forcing BMW to rebuild the plant around a fixed automation cell.
That is the economic case for humanoids in factories: use a human-shaped machine where the environment was already designed for humans. Figure AI's published materials present Figure 03 as a human-scale platform intended for factory and logistics work.
The missing numbers still matter
The BMW release confirms volume of vehicle production supported: more than 30,000 X3s. It does not disclose the number of Figure robots deployed, the uptime those robots achieved, the pace of the insertion task, the economics of the BMW relationship, or whether BMW is paying commercial rates beyond pilot terms.
Those omissions are not small details in humanoid robotics. A task can work in production and still fail commercially if it requires too much human supervision, too many maintenance interventions, or too many robots to justify the labor savings. The difference between one robot doing a sliver of a workflow and a fleet operating across shifts is the difference between a proof point and a business.
That is why the BMW disclosure cuts both ways. It gives Adcock a stronger answer to skeptics than another lab video would. It also raises the standard for the next set of disclosures. RuntimeWire reported in May that Adcock was staking Figure AI's near-term narrative on deployments and on the argument that intelligence, not sheer robot scale, would decide the humanoid race. BMW has now given him a customer-sanctioned production credential. The market will still want fleet size, utilization and contract economics.
The skepticism around those specifics was already visible before this release. TechCrunch reported in June 2025 that Adcock, appearing at a Bloomberg conference, did not provide specifics on whether the BMW relationship was a pilot or had commercial value, while saying Figure AI got value from running robots on the factory floor every day and tracking their performance.
BMW's new statement does not answer the commercial-value question. It does, however, answer a different one: whether BMW is willing to put its name on the claim that Figure AI's robot worked in production conditions for a sustained period.
The capital behind the bet
Figure AI has raised the stakes around every production proof point because it has raised capital as if the category is going to consolidate quickly. In September 2025, Figure AI said it had exceeded $1 billion in Series C committed capital at a $39 billion post-money valuation. The company said the round was led by Parkway Venture Capital, with investment from Brookfield Asset Management, NVIDIA, Macquarie Capital, Intel Capital, Align Ventures, Tamarack Global, LG Technology Ventures, Salesforce, T-Mobile Ventures and Qualcomm Ventures.
That financing was not described as money for a research lab. Figure AI said the capital would go toward bringing humanoids into real-world environments at scale, expanding its BotQ manufacturing, scaling Helix, building GPU infrastructure, and collecting real-world data to train embodied intelligence systems.
BMW's Spartanburg deployment is therefore not just a customer milestone. It is a data asset, a fundraising proof point and a manufacturing rehearsal. Every successful production cycle gives Figure AI more information about failure modes, safety, hands, perception and workflow design. Every missing commercial metric keeps the valuation debate alive.
Humanoid robotics is splitting into wedges
Figure AI is not alone in trying to turn physical AI into a category investors can underwrite. Apptronik said in February 2026 that it closed a $520 million Series A-X extension, bringing its total Series A to more than $935 million and total capital raised to nearly $1 billion, with partners including Mercedes-Benz, GXO and Jabil. Agility Robotics said in June 2026 that it agreed to go public through a merger with Churchill Capital Corp XI at a $2.5 billion pre-money equity value, and said Digit was commercially deployed with Schaeffler, GXO, Toyota Motor Manufacturing Canada and Mercado Libre.
The category is also splitting by use case. Figure AI and Apptronik are pushing hard into factories, logistics and eventually home use. Agility has been more warehouse and distribution oriented. 1X is taking the home robot pitch directly to consumers, with NEO pre-orders and first shipments planned for 2026. Unitree has moved the conversation on price: Notebookcheck reported that Unitree R1 launched at $5,900, though the report also noted that its usefulness in everyday life remained uncertain.
That backdrop makes BMW's role in Figure AI's story more important. Consumer humanoids can sell a dream before they prove utility. Industrial humanoids do not get that luxury for long. A plant manager needs throughput, reliability and safety. BMW's disclosure gives Figure AI a real answer on the first layer of that test: the robot was used in a production workflow tied to more than 30,000 vehicles.
The next layer is harder. Figure AI now has to show that a humanoid can move from one qualified factory task to a repeatable deployment model across customers, sites and shifts. That is where Adcock's founder thesis either compounds or stalls. BMW has supplied the first credible production sentence. Figure AI still has to turn it into a paragraph of operating metrics.