## Does Figure 03 Replace Figure 02 at BMW's Spartanburg Plant?

**Figure 02 helped build more than 30,000 BMW X3 SUVs over a ten-month deployment at BMW's Spartanburg, South Carolina plant.** That production milestone — reported this week at BMW's "Home of X" event marking the launch of the 2027 X5 series — now serves as the validation data point BMW needed to greenlight [Figure AI](https://humanoidintel.ai/companies/figure-ai)'s next-generation platform. Figure 03 is now active at Spartanburg, shifting from body-shop welding support to a logistics-focused role: picking unsorted vehicle components, organizing them into sequencing carts, and delivering parts to assembly workers in the correct order.

The hardware upgrade is substantive. Per the source, Figure 03 includes tactile sensors in its hands, palm-mounted cameras, wireless charging, soft components designed to improve worker co-location safety, and speech-to-speech communication capabilities — a meaningful step up from the sheet-metal insertion tasks Figure 02 handled in the welding area. The deployment is not a controlled demo; the robot is working on a live production floor alongside BMW employees.

For the broader humanoid-in-automotive thesis, this is the clearest production-scale data point yet: a named OEM, a specific plant, a verifiable vehicle count, and a second-generation robot already in rotation.

---

## From Body Shop to Logistics: What Figure 03 Is Actually Doing

The task transition between Figure 02 and Figure 03 at Spartanburg is strategically significant. Body-shop work — inserting sheet-metal parts into a welding process — is highly constrained, repetitive, and spatially bounded. It is, frankly, one of the easier environments in which to prove out a humanoid.

Logistics sequencing is harder in a different way. The robot must handle a wider variety of part geometries, manage variability in bin presentation, and correctly sequence components for downstream assembly accuracy. Errors here don't just halt a single welding cycle — they can disrupt the entire assembly line sequence. That BMW is comfortable moving Figure 03 into this role suggests the ten-month Figure 02 deployment produced more than a vehicle count; it likely generated the operational trust and safety data required to expand scope.

[Dexterous manipulation](https://humanoidintel.ai/glossary/dexterous-manipulation) of unsorted components — parts that arrive in arbitrary orientations without structured fixturing — is precisely the capability gap that has kept humanoids out of mainstream logistics. Palm-mounted cameras and tactile sensors in the hands directly address this: they give the robot the proprioceptive and visual feedback needed to grasp irregularly presented parts without pre-positioning. Whether Figure 03's grasp success rates in unstructured pick tasks are production-competitive with dedicated fixed automation is a question the source does not answer — and one BMW's engineering team will be monitoring closely.

---

## Physical AI at Spartanburg: Robots Are One Layer of a Larger Stack

BMW's framing at the "Home of X" event is telling. The company explicitly situates humanoid robots within a broader [Physical AI](https://humanoidintel.ai/glossary/physical-ai) strategy rather than treating them as standalone novelties. The Spartanburg plant already runs virtual 3D factory simulations for production planning, AI-powered camera-and-sensor quality inspection, and digital ergonomics tools that model worker strain before a line goes live.

This matters for how analysts should interpret the deployment. BMW is not betting on humanoids as a single-point solution. The robots enter an environment already instrumented with AI perception, digital twins, and simulation tooling — infrastructure that reduces the sim-to-real gap and provides contextual data that can improve robot policy performance over time. The implication: BMW's manufacturing environment may be more favorable to humanoid deployment than a typical brownfield factory precisely because the surrounding AI stack is already mature.

The "we are not replacing workers" messaging is standard OEM positioning, but it also reflects a real operational constraint: BMW's Spartanburg workforce and union dynamics require robots to be positioned as augmentation tools targeting physically demanding or safety-critical tasks. Whether that framing holds as the robots' capabilities expand into higher-value tasks is a question every automotive OEM deploying humanoids will eventually face.

---

## What the 30,000-Vehicle Number Actually Means

A deployment scale of more than 30,000 vehicles over ten months is the most concrete production-floor validation number any humanoid company has published to date. To be precise about what it measures: Figure 02 *supported* production of those vehicles — it performed sheet-metal insertion tasks as part of the welding process, not full vehicle assembly. The robot was one node in a larger manufacturing workflow.

That distinction matters when extrapolating. It does not mean a humanoid built 30,000 cars. It means a humanoid executed a specific, bounded task reliably enough — at sufficient uptime, cycle time, and defect rates — that BMW kept it running across the production of 30,000 units. That is a meaningful engineering proof point, not a deployment exaggeration, but it should be read with precision.

For Figure AI, the number is a commercial asset. It converts a proof-of-concept narrative into a trackable production metric that prospective OEM customers can benchmark against. Competitors including [Agility Robotics](https://humanoidintel.ai/companies/agility-robotics), [Apptronik](https://humanoidintel.ai/companies/apptronik), and others pursuing automotive partnerships will need equivalent production-floor data to compete for the next wave of OEM contracts.

---

## Industry Trajectory: The Automotive Beachhead Is Real

The Figure AI / BMW Spartanburg deployment is now the most documented humanoid-in-automotive case study in the industry. It establishes several reference points that will shape how the next 18 months of OEM deals get structured:

- **Task selection matters more than raw capability.** Figure 02's success in a constrained welding-assist role — not general-purpose manipulation — is what earned Figure 03 its expanded logistics scope. OEMs will continue to deploy humanoids into bounded, high-repetition tasks before expanding their operational envelope.
- **Second-generation hardware cycles are fast.** The move from Figure 02 to Figure 03 within a single OEM deployment signals that hardware iteration timelines in this space remain aggressive. Customers will need to think about platform upgrades as part of their deployment contracts.
- **The logistics sequencing task is the real test.** Body-shop welding assist was a controlled entry point. Parts sequencing from unsorted bins is a harder, more economically valuable problem. Figure 03's performance on that task — data BMW has not yet released — will be the number worth watching.

---

## Key Takeaways

- **Figure 02 supported production of more than 30,000 BMW X3 SUVs** over a ten-month deployment at BMW's Spartanburg, South Carolina plant — the most documented production-floor humanoid deployment on record.
- **Figure 03 is now active at Spartanburg** in a logistics role, organizing unsorted vehicle components into sequencing carts for assembly workers — a more complex manipulation task than its predecessor handled.
- **Figure 03 hardware upgrades** include tactile-sensor-equipped hands, palm-mounted cameras, wireless charging, softer safety components, and speech-to-speech communication.
- **BMW's deployment is embedded in a broader Physical AI stack** that includes digital twin simulation, AI quality inspection, and ergonomics tooling — a favorable environment for humanoid integration.
- **The body-shop-to-logistics task expansion** is the real signal: BMW's operational confidence in humanoids has grown enough to move them into a higher-variability, higher-consequence workflow.

---

## Frequently Asked Questions

**How many vehicles did Figure 02 help build at BMW?**
Figure 02 supported production of more than 30,000 BMW X3 SUVs during a ten-month deployment at BMW's Spartanburg, South Carolina plant, performing sheet-metal insertion tasks in the body shop's welding process.

**What is Figure 03 doing at BMW Spartanburg?**
Figure 03 is handling parts logistics: picking up unsorted vehicle components, organizing them into sequencing carts, and delivering parts to assembly workers in the correct order. This is a different, more variable task than Figure 02's welding-assist role.

**How does Figure 03 differ from Figure 02?**
Per BMW's "Home of X" event reporting, Figure 03 includes tactile sensors in its hands, cameras mounted in its palms, wireless charging, softer components for improved co-worker safety, and speech-to-speech communication capabilities.

**Is BMW replacing workers with humanoid robots?**
BMW states the robots are intended to handle repetitive, physically demanding, and safety-critical tasks, freeing workers for higher-judgment work. The company has not announced workforce reductions tied to the humanoid deployment.

**What is Physical AI and why does BMW use the term?**
[Physical AI](https://humanoidintel.ai/glossary/physical-ai) refers to artificial intelligence systems that control machines operating in the physical world. BMW uses the term to describe its broader factory intelligence strategy, which includes humanoid robots, digital twin simulations, AI quality inspection, and ergonomics modeling — not robots in isolation.

**Which other humanoid companies are pursuing automotive deployments?**
Agility Robotics, Apptronik, and several others are in discussions or early trials with automotive OEMs, but none have published a production-floor vehicle count comparable to the Figure AI / BMW data point as of this reporting.