Figure Retires the F.02: ''Battle-Scarred'' Robots, 30,000 BMWs, and Hard Lessons Learned

There is something jarring about seeing a banged up humanoid robot. In an industry dominated by gleaming renders and pristine laboratory demos, the latest update from Figure AI offers a different kind of visual: a Figure 02 unit covered in scratches, scuffs, and industrial grime.

Figure CEO Brett Adcock announced today that the company is officially retiring its Figure 02 (F.02) fleet. The retirement follows an 11-month deployment at BMW Manufacturing’s Spartanburg, South Carolina facility—a partnership that has been the subject of intense scrutiny regarding the gap between PR hype and operational reality.

To mark the end of the F.02’s lifecycle, Figure released a breakdown of the data gathered from the factory floor. The numbers are significant: the company claims the robots contributed to the production of over 30,000 BMW X3 vehicles, loading more than 90,000 sheet metal parts over the course of the year.

The "Grime" as a Badge of Honor

The most compelling part of today’s announcement isn't just the metrics, but the visual evidence accompanying them. Video footage shared by Adcock and the official Figure account shows the robots in a state of clear wear-and-tear—evidence of what Adcock describes as "real world deployment".

This aesthetic of endurance serves a strategic purpose. It directly addresses skepticism that arose earlier this year when reports suggested Figure's presence at BMW was merely a "small-scale feasibility study" rather than a functional integration. By showing the "scars" of the assembly line, Figure is validating its previous claims of operating for months on end.

The Data: What Did It Actually Do?

According to the post-mortem report released by Figure, the deployment ramped up rapidly. Within six months of the F.02's initial "bring up," units were delivered to Spartanburg. By the 10-month mark, the robots were running full deployments on an active assembly line.

The primary use case was sheet-metal loading. This is a classic "pick-and-place" task: the humanoid picks metal parts from a bin and places them onto a welding fixture within a 5-millimeter tolerance. Once placed, traditional six-axis industrial arms take over for the welding.

Figure shared the following KPIs for the deployment:

• Runtime: 1,250+ hours accumulated.
• Volume: 90,000+ parts loaded.
• Distance: An estimated 1.2 million steps, or roughly 200 miles walked.
• Schedule: The robots eventually ran 10-hour shifts, Monday through Friday.

The performance targets were demanding. The robot had to maintain a total cycle time of 84 seconds (with 37 seconds allocated for the load time) and achieve a placement accuracy of greater than 99%.

The "Bitter Lesson" of Hardware

Perhaps the most valuable section of Figure’s report is the admission of failure points. Transparency regarding hardware failure is rare in the humanoid sector, where companies often curate footage to show only flawless execution. This candor provides a concrete example of what Boston Dynamics executives recently described as the "Phase One" reality of the industry—the grinding necessity of validating hardware reliability before mass deployment.

Figure admitted that the forearm was the "top hardware failure point" during the BMW residency. The complexity of packing three degrees of freedom, thermal constraints, and dynamic cabling into a human-sized limb proved difficult. The constant motion stressed the microcontroller-based PCB and the cabling that distributed communications to the wrist.

This failure analysis directly informed the design of the recently unveiled Figure 03. In the new model, Figure has "completely re-architected" the wrist electronics to eliminate the distribution board and dynamic cabling, moving to a system where motor controllers communicate directly with the main computer.

From Pilot to Product

This "retirement" marks a transition point for Figure. The F.02 served as the bridge—or perhaps the sacrificial lamb—between the lab and the line.

The distinction between "pilot" and "deployment" has been a sore spot for the company, leading to threats of legal action against media outlets that questioned the scale of the BMW deal. However, today's retrospective frames the F.02 specifically as a learning tool. "Figure 02 taught us early lessons on what it takes to ship," the company stated.

The retirement of the fleet clears the deck for the Figure 03, which Figure claims is "ready for the world at scale". While the F.02 proved that a humanoid can work on a line for 1,250 hours, the F.03 will now have to prove it can do so economically and reliably for thousands more.