BMW’s New Humanoid Takes Over Tasks

BMW has put a next-generation humanoid robot on factory work in South Carolina, and the move shows how fast physical artificial intelligence is moving from lab demos to real plant floors.

Quick Take

BMW says Figure 03 is now being deployed at Plant Spartanburg for logistics sequencing.
The earlier Figure 02 pilot ran nearly a year in the body shop and helped build more than 30,000 BMW X3 vehicles.
BMW says the new robot adds softer parts, wireless charging, speech-to-speech audio, and better hands.
The company says the robot is meant to support workers, not replace them.

BMW Moves From Pilot to Production Use

BMW Group says Figure 03 is moving into logistics at Plant Spartanburg after a Figure 02 pilot proved the concept in the body shop. The company says the new job is sequencing parts. That means the robot will take unsorted parts from large containers, sort them, and place them into trolleys for assembly lines.

BMW’s earlier test lasted about 11 months and focused on sheet-metal work in the body shop. BMW Blog said the Figure 02 pilot helped build more than 30,000 BMW X3 vehicles by inserting sheet-metal parts for welding. The reporting points to a narrow but real use case: repetitive work inside a controlled plant, not a science-fiction-style factory takeover.

What Changed in Figure 03

BMW says Figure 03 brings several hardware upgrades meant for work around people. Those changes include soft exterior parts, wireless charging, speech-to-speech audio, tactile sensors in the hands, and palm cameras. BMW also says the robot has improved hand design and better movement control for line-side tasks.

Secondary reporting adds that Figure 03 is built for finer manipulation and better uptime than its predecessor. One case study says the new model has improved hand dexterity, more payload capacity, and longer battery runtime. That matters because factory work is not won by hype. It is won by whether a machine can repeat the same job, hour after hour, without constant human rescue.

Why BMW Calls It Physical AI

BMW ties the Spartanburg rollout to its iFACTORY program, which uses more automation, digital planning, and smarter production flow. The company says Figure 03 is part of a broader push to use physical artificial intelligence in manufacturing. In plain terms, BMW is using a humanoid robot where a human-shaped machine can move through human-built spaces without a full plant rebuild.

BMW also says it sees humanoid robots as an extension of its current automation, not a replacement for employees. That is the right public line for any company introducing robots on a factory floor. The real test is whether workers gain help with hard, repetitive tasks while keeping their jobs and skills intact. BMW has not released public safety data, injury-rate data, or long-term cost figures for the new deployment.

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

The wider industry picture helps explain why this story matters. Automotive plants are among the first places where humanoid robots are finding practical use, mostly in narrow logistics and material-handling jobs. Even supportive analysts say broad factory-scale adoption is still limited, and the strongest cases today are tightly controlled tasks rather than full general labor replacement.

What Remains Unproven

BMW’s announcement is strong on promise, but thin on outside verification. Public reporting does not yet include independent audits of uptime, maintenance cost, or energy use for Figure 03. It also does not show quantified before-and-after data for delay reduction or worker safety improvements. Those gaps matter because big automation claims should be measured, not just marketed.

Social media and some headlines are already framing the robot as if it “took over” the plant. That language is more dramatic than the evidence. The available reporting shows a targeted deployment in one South Carolina plant, built around sequencing and part handling. For readers who care about jobs, manufacturing strength, and honest reporting, the key question is not whether robots can help. It is whether the numbers will eventually match the sales pitch.