| Date | 30 Jun 2026 |
|---|---|
| Region | U.S. & Allies |
Figure is presenting a first logistics demonstration of its third-generation humanoid, not a second production tally, and the 30,000-car figure belongs to Figure 02.
IN BRIEF: On June 30, 2026, Figure said its Figure 03 robot had arrived in Hall 52 at BMW Group Plant Spartanburg for a first logistics-sequencing demonstration, after Figure 02 contributed to the assembly of 30,000 cars at BMW last year, according to Figure’s post.
RECORD:
Date: June 30, 2026
Beat: Humanoid robotics / automotive logistics
Jurisdiction: N/A (BMW Group Plant Spartanburg named; country not stated on the source)
Entity: Figure; BMW Group
Instrument / cite: N/A
Source type: Company news post
Confidence: reported (company claim of a demonstration; no independent production tally for Figure 03)
Record ID: N/A
Source: Figure
Figure told readers on June 30, 2026 that its partnership with BMW Group has moved to a new task, and that the move is a demonstration rather than a counted production run. The post, titled “F.03 Arrives at BMW,” says the latest-generation robot, Figure 03, arrived in Hall 52, one of the assembly and logistics halls at BMW Group Plant Spartanburg. The same paragraph places that arrival after “the successful deployment of Figure 02 on the assembly line in 2025.” No BMW executive is named. No contract, purchase order, or safety instrument is cited.
The only production figure in the post belongs to the prior machine. Figure writes: “Figure 02 contributed to the assembly of 30,000 cars at BMW last year.” The post does not say Figure 03 has contributed to any vehicle count, shift hours, or part total. It says Figure “has transitioned from classic pick-and-place sheet metal loading to an entirely new tier of complexity: the sequencing use case.” A 30,000-car contribution claim sits next to a first demonstration. They are not the same class of evidence. “Last year,” on a post dated June 30, 2026, is Figure’s phrasing; the post also dates the Figure 02 assembly-line deployment to 2025 and does not break the 30,000 cars by model or month.
Figure’s highlights label the Spartanburg appearance as “the first demonstration of Figure 03 performing a logistics workflow at BMW Group Plant Spartanburg.” The control stack named is Helix 02, “Figure’s proprietary pixels-to-actions VLA,” which the post says coordinates the robot’s hands, arms, torso, and feet. The behavior described is loco-manipulation: the robot manipulates parts while stepping and repositioning its body to pull a heavy cart on caster wheels. A separate highlight says Figure 03 “fluidly switches between precise picking and placement of thin-walled, individual parts and forceful manipulation requiring whole-body coordination – pulling a large metal cart down the line.” Figure argues this class of dynamic material manipulation “is structurally infeasible to solve with traditional, fixed automation or six-axis robotic arm.” That is Figure’s claim about the task boundary. The post does not present it as a BMW engineering finding, and it supplies no cycle time against a fixed cell or a six-axis arm.
The sequencing section is the operational content. Figure writes that sequencing is integral to selecting parts for the assembly line, and that it is “an intractable sorting environment: parts do not arrive in mathematically perfect orientations.” The following section says the task “cannot be solved reliably with a fixed series of hard-coded motions,” because carts, bins, and parts do not arrive in exactly the same position every time. Parts “may have shifted, rotated, partially occluded, or presented differently within a container.” Each interaction, Figure says, requires the robot to perceive the scene and make small corrections on the fly. When manipulation and locomotion run together, Figure 03 “must grasp parts with both hands while adjusting its foot placement, shifting its body to maintain reach and balance, and precisely placing each part into the correct slot.” Helix 02 “powers this loco-manipulation behavior through high-frequency visual-motor control, continuously adapting Figure 03’s movements to spatial variation and correcting small errors as the task unfolds.” No error rate, grasp success rate, or parts-per-hour figure accompanies that description.
What the post does not supply is as material as what it does. There is no statement that Figure 03 is running a production shift. There is no count of parts sequenced, no hours of runtime, and no safety standard, works notice, or BMW confirmation inside the text. The closing sentence is company voice, not a named official: “With Figure 03 enabled by Helix 02 at BMW Group Plant Spartanburg, we are demonstrating that general-purpose physical AI can master the cognitive and dexterous tasks that have bottlenecked manufacturing logistics for generations.” “Demonstrating” and “can master” are the verbs Figure chose. They describe a capability claim attached to a first logistics demonstration, anchored to a plant and a hall, and set beside a prior-generation contribution of 30,000 cars.
For founder reputation, the post extends a named OEM relationship from sheet-metal loading into sequencing, which Figure itself calls a higher tier of complexity. For the category narrative, the argument is that a humanoid running a pixels-to-actions model can take a sorting job that fixed automation and a six-axis arm are said not to solve. For investor scrutiny, the bankable number on this page remains the Figure 02 contribution. Figure 03 is still in demonstration language. For an enterprise buyer, Hall 52 is a site reference, not a deployment certificate. Confidence stays at reported until BMW or a production tally says otherwise.
WHY IT MATTERS: Figure is using a named BMW Group plant to argue that general-purpose physical AI can take sequencing work the company says fixed automation cannot reliably do. The post’s own wording keeps Figure 03 at a first demonstration, so the 30,000-car line should not be read as evidence that the new robot is already on a counted production contribution.