The Future Is Here: BMW Puts Humanoid Robots to Work

At BMW’s massive South Carolina plant, next‑generation humanoid robots are quietly taking over some of the toughest factory jobs — raising fresh questions about who really wins as “physical AI” moves from demo to daily work.

Story Snapshot

BMW is now using Figure 03 humanoid robots on the factory floor at Plant Spartanburg to sort and sequence car parts.
Earlier Figure 02 robots helped assemble more than 30,000 BMW X3 vehicles by loading sheet‑metal panels for welding.
BMW says these robots extend automation and improve safety, not replace workers, but hard proof on jobs and injuries is still thin.
This deployment is one of the first paid, large‑scale humanoid robot programs in auto manufacturing, signaling a new phase of factory automation.

BMW’s humanoid robots move from experiment to real factory work

BMW’s Spartanburg plant in South Carolina, which builds key sport utility models for the global market, has become one of the first car factories where humanoid robots are part of regular production work, not just flashy demos. After a pilot with the Figure 02 robot in the body shop, BMW and Figure AI began using the newer Figure 03 units under a commercial agreement that shifts robots into daily logistics and assembly support, marking a move from research testing to paid industrial deployment.

During the earlier trial, Figure 02 robots loaded sheet‑metal panels onto welding fixtures, helping build more than 30,000 BMW X3 vehicles over about 10–11 months. They worked long, repeated shifts, lifting parts from bins and placing them with millimeter‑level accuracy so fixed welding machines could do the actual joining. Industry analysts say this kind of narrow, high‑repetition task is exactly where humanoid robots first make sense, because these jobs are hard on human bodies yet stubbornly expensive to automate with traditional fixed machines.

What the new Figure 03 robots actually do on the line

The new Figure 03 robots are now being used for “just‑in‑sequence” logistics at Spartanburg, where parts arrive in large, unsorted containers and must be picked, sorted, and loaded into trolleys in the right build order for the assembly line. BMW says the robot’s job is to handle this sorting flow so the right part reaches the right worker at the right time, helping keep production on schedule and reducing delays, though the company has not yet shared hard before‑and‑after numbers on time saved or bottlenecks removed.

Figure 03 includes several hardware upgrades designed for working closer to people, including a softer exterior, wireless charging, and improved hands with tactile sensors and cameras in the palms. Company materials describe speech‑to‑speech audio tools so workers can talk to the robots, and a control system that pairs visual analysis with whole‑body motion planning, aiming to let the robot move through busy aisles and handle different parts with greater awareness. Case‑study data from industry sites claims around 99 percent placement accuracy and a roughly $25 per robot‑hour cost at Spartanburg, suggesting BMW sees a clear business case, though these figures come from partners and have not been independently audited.

Jobs, safety, and the growing fear of “robots taking over”

BMW publicly frames these humanoid robots as an extension of its existing automation, saying they are meant to take on hazardous, repetitive, or physically demanding tasks and improve ergonomics and safety for human associates, not replace them. The company points to tasks like heavy part lifting and awkward loading, which can strain backs and joints over time, as early targets for robots, yet it has not released detailed data showing lower injury rates or better long‑term health outcomes since the deployments began.

Outside BMW’s press releases, public debate looks very different: social media clips and commentary often claim these robots are “taking over” factories and putting workers at risk of being replaced. That fear draws on a long history of automation in auto plants, where past waves of robots did lead to fewer human jobs in some areas and fueled fights between workers, unions, and management. Experts who track the humanoid market say real deployments today are still small and focused on narrow jobs like tote moving, part loading, and simple inspection, but they also warn that plans from carmakers and robot firms point toward much larger rollouts in the 2030s.

What we know, what we do not, and why it matters beyond BMW

For now, many of the most eye‑catching numbers around Figure 02 and Figure 03 — from success rates near 99 percent to fleets of dozens of units — come from BMW, Figure AI, or partner case studies, not independent audits, so the public must take company claims largely on trust. There is also still no open data on long‑term reliability, repair costs, or energy use for these robots in real factory life, even though those details will decide whether “physical AI” truly saves money or simply shifts costs from workers to machines and software.

Figure 02 ran 11 months inside BMW Spartanburg and helped build 30,000 X3 vehicles.

Nobody announced it.

Now Figure 03 with Helix 02 — a pixels-to-actions model — is handling logistics sequencing at the same plant.

The press covers the demo. The robot keeps working.

— Vivek Kotecha (@vbkotecha) July 5, 2026

At the same time, BMW’s program fits a wider pattern: auto manufacturing and warehouse logistics are becoming the first major test beds for humanoid robots worldwide, as firms like BMW, Tesla, and Mercedes push robots into jobs built around human bodies but often ignored by traditional automation. For Americans who already feel the system favors big companies and high‑tech elites over regular workers, this story is another warning sign that key decisions about work, safety, and dignity on the job are being made by corporate boards and investors long before voters or local communities get a real say.