Humanoid Robots Are Actually Working in Factories Now. Here's What's Really Deployed.
For years, humanoid robotics was mostly a genre of demo video: a robot doing a backflip, folding laundry once on stage, walking smoothly across a controlled set. In 2026, industry reporting suggests that's shifted, at least partially, into actual factory deployment, though the exact scale is murkier than headlines suggest.
Tesla's Optimus Gen 3 units are reportedly working inside Tesla's own facilities, primarily Gigafactory Texas and the Fremont plant, handling tasks like battery cell sorting, light assembly, and repetitive pick-and-place operations. Third-party industry reporting puts the task range at roughly 25 distinct manipulation tasks and cites over 1,000 deployed units, though it's worth being direct about the sourcing here: Tesla has never published an official Optimus production or deployment count. The commonly cited figures come from industry reporting, not company disclosure.
Figure AI's Figure 02 offers a more independently documented case: an eleven-month pilot at BMW Group's Plant Spartanburg, with additional reported deployment in Amazon warehouse settings. Eleven months of continuous real-world operation is a substantially different claim than a one-off demo, and it's backed by a named manufacturing partner rather than only the robotics company's own account.
Tesla has also reportedly set a goal of ramping Optimus V3 production toward 1 million units per year by late 2026, according to industry sources, an extraordinarily ambitious target with no real precedent in robotics manufacturing, and one that should be read as a stated ambition rather than a confirmed near-term outcome.
Why It Matters
Humanoid robotics has spent years generating more hype than deployed hardware. What's different in 2026 is that there are now specific, named factory pilots with specific, named tasks, battery cell sorting, light assembly, pick-and-place work, rather than only choreographed demo footage. That's a real shift, even if the scale being reported is smaller and less confirmed than headlines suggest.
The gap between "robots are working in factories" and "robots have solved general-purpose physical labor" is still enormous, and conflating the two leads to bad expectations for how fast this technology actually scales.
Key Details
- Tesla Optimus Gen 3: Reportedly deployed at Gigafactory Texas and the Fremont facility
- Reported Optimus task range: Around 25 distinct manipulation tasks, per third-party reporting
- Reported Optimus deployment count: Over 1,000 units, per third-party reporting; not officially confirmed by Tesla
- Tesla's stated production goal: Ramp Optimus V3 production toward 1 million units/year by late 2026, per third-party reporting
- Figure AI Figure 02: Completed an 11-month pilot at BMW Group Plant Spartanburg
- Figure 02 additional deployment: Reported operation in Amazon warehouse settings
- Verification status: Tesla has not published an official Optimus production count; most figures come from industry reporting, not direct company disclosure
Technical Analysis
The tasks being reported for Optimus Gen 3, battery cell sorting, light assembly, repetitive pick-and-place, are exactly the kind of narrow, structured, repeatable work that robotics has been able to handle for years, just previously with fixed industrial arms rather than a humanoid form factor. The interesting technical claim isn't that a robot can sort battery cells; it's that a general-purpose humanoid body can be reprogrammed across roughly 25 different manipulation tasks without needing task-specific hardware for each one.
Figure AI's eleven-month pilot at BMW is notable specifically for its duration: an eleven-month real-world deployment is a meaningfully different claim than a demo, since it implies the robot kept working reliably across nearly a year of actual production conditions, not just a controlled showcase.
The important caveat, flagged directly by industry reporting: Tesla has never published an official Optimus production count, and the widely cited "over 1,000 units" figure comes from third-party industry sources, not a company disclosure. The gap between highlight-reel demo videos and verified operational reality remains real, and readers should treat specific unit counts as estimates, not confirmed facts, until a company states them directly.
Competitive Comparison
| Product | Better At | Weakness |
|---|---|---|
| Tesla Optimus Gen 3 | Manufacturing scale ambition, vertical integration with Tesla's own factories | No official production/deployment numbers published by Tesla; claims rely on third-party reporting |
| Figure AI Figure 02 | Longest publicly documented pilot duration (11 months at BMW), multi-customer deployment (BMW, Amazon) | Smaller reported deployment scale than Optimus claims |
| Boston Dynamics Atlas | Most technically advanced public demonstrations of dynamic movement and agility | Less publicized evidence of sustained commercial factory deployment compared to Optimus or Figure 02 |
Industry Impact
Manufacturers evaluating humanoid robotics now have real reference points, BMW's eleven-month Figure 02 pilot in particular, rather than only demo footage to judge feasibility against.
Robotics companies face rising pressure to publish verified deployment numbers rather than letting third-party estimates define the public narrative, since the credibility gap between claimed and confirmed figures is now a visible industry-wide issue.
Factory workers and labor groups have a more concrete, if still limited, set of real deployments to point to when discussing automation's near-term impact, rather than purely speculative timelines.
Investors assessing humanoid robotics valuations need to weight company-reported figures differently from independently verified ones, given how much of the current public narrative rests on unconfirmed third-party numbers.
Future Outlook
The following is analysis and prediction, not confirmed fact.
Expect growing pressure on Tesla specifically to publish an official Optimus deployment count, given how widely the unconfirmed "1,000+ units" figure is already being cited as fact across the industry. Whether that pressure produces an actual disclosure or continued silence remains to be seen.
The realistic near-term trajectory is incremental: more narrow, single-task pilots like Figure AI's BMW deployment, rather than a sudden jump to general-purpose humanoid labor. The 1-million-units-per-year production goal reported for Optimus V3, if accurate, would represent an enormous manufacturing scale-up with no precedent in robotics history, worth treating as an ambition rather than a near-term certainty until production data confirms it.
Key Takeaways
- Tesla Optimus Gen 3 and Figure AI's Figure 02 both reportedly have real factory deployments in 2026, not just demo footage.
- Figure AI completed an eleven-month pilot at BMW's Spartanburg plant, a meaningful real-world duration.
- Tesla has never officially confirmed an Optimus production or deployment count; the widely cited "1,000+ units" figure comes from third-party reporting.
- Reported tasks (battery sorting, light assembly, pick-and-place) are structured, repeatable work, not evidence of general-purpose physical labor yet.
- Treat specific unit-count and production-target figures as industry estimates until companies confirm them directly.
Source: Technology.org
