General-purpose robots
Robots built for human spaces could bring automation into many more jobs. Reliable work at a sensible cost will matter more than an impressive demonstration.
What I’m watching: Independent customers and practitioners begin describing humanoids as practical workplace tools rather than impressive prototypes.
What could change the picture?
- What the evidence says
- BMW and its partners describe real pilot work, alongside the safety, training and infrastructure needed to support it. The next useful evidence is repeat customer deployment with a clear account of the costs and benefits.
- Early recognition to watch
- Independent customers and practitioners begin describing humanoids as practical workplace tools rather than impressive prototypes
- Commercial confirmation
- An original end2026 operator/partner update resolving rollout status, then repeated paid task output with intervention, downtime, safety, worker experience, comparator cost and retained supplier economics disclosed.
- What people used to expect
- Our earlier working view: better transfer learning might reduce adaptation work, while dependable customer deployment and integration remain scarce.
- What may be changing
- Operator and partner accounts show bounded production work and supporting factory changes. Integration, safety, sensing and training are candidate capture layers; general autonomy and profitable scale remain unproved.
- Who could benefit
- Hexagon, named AEON technology partner; exposure unassessed: Robot systems, sensing, software and integration are directly connected to the named BMW programme. A deployment statement does not disclose order value, recognised revenue, margin, current listed vehicle or price. NVIDIA, indirect compute candidate only (NVDA): Training and edge compute may benefit from wider deployment. These BMW/Hexagon originals do not establish NVIDIA supply terms or material incremental revenue from this programme.
- Timing
- Weeks for new evidence; quarters or longer for commercial deployment.
- What the market may already expect
- No exchange-verified dated price or valuation is attached. Nasdaq NVDA historical page returned data unavailable on 17 October. Third-party closing-price candidates are not cleared quotes; remaining expectations and instrument attractiveness are unknown.
- The strongest reason to be cautious
- A specialised repetitive task in a highly automated plant may not generalise. Extra barriers, connectivity, training and maintenance can absorb savings. Reported sales are not accepted operating hours; operator and partner statements are interested sources rather than audited economics or independent worker-outcome evidence.
- What would change the view
- Weaken the scaling/capture thesis if wider tasks or sites require persistently high intervention, supporting infrastructure erases customer value, paid repeat orders do not follow, or safety and worker experience deteriorate.
Detailed research notes
BMW’s27 February account reports Figure02 handling more than90,000 components over about1,250 operating hours in a2025 pilot. Its21 September page describes safety barriers and5G changes but still carries April/summer planning text. Hexagon’s15 June update reports production tasks alongside training and targets end2026 deployment. IFR’s30 September release reports about7,000 full-size humanoid sales in 2025 while warning of teleoperation and weak business cases. These are dated baselines interpreted now, not October deployment completions. BMW and Hexagon describe one partner programme, not independent replication.
Follow the evidence
Original post inspected after Grok discovery. The author describes deployment constraints and opportunities in integration and infrastructure. His installation statistics are not independently verified here.
IFR reported about 7,000 full-size humanoid robots sold globally in 2025. Limitation: Sales are not commissioning, reliable task performance, repeat use, worker outcomes or positive deployment economics. IFR says most applications remain specialised and often require teleoperation, while safety, training, maintenance and weak business cases constrain adoption. IFR represents robotics suppliers and the public release does not expose complete sampling or transaction-level evidence. Existing research checked 2026-10-05.
BMW describes enabling infrastructure changes around the humanoid pilot. Limitation: No intervention, error, downtime or matched comparator economics disclosed here. Shared BMW lineage, not independent safety certification or observed labour benefit. Existing research checked 2026-10-12.
Figure reports better transfer from human-behaviour pretraining and less task-specific data. This is issuer-reported technical evidence; profitable, dependable commercial deployment is not established.
Hexagon reports AEON performing production tasks while training. Limitation: Partner statement, not scaled accepted output, autonomous uptime, task cost or worker acceptance. BMW September page retains earlier planning text; full rollout timing remains ambiguous. Existing research checked 2026-10-12.
BMW reports that Figure 02 moved more than 90,000 components in around 1,250 operating hours during a ten-month production pilot. Limitation: A single-company pilot for one repetitive task is not general humanoid capability, a cost per useful task, autonomous uptime, labour substitution, payback or scaled deployment. Intervention, error, downtime and comparator data were not disclosed. BMW and Figure are the operator and technology partner; this result was not independently audited here. Existing research checked 2026-10-05.
low confidence · Assessment 17 Oct 2026. Evidence dates are shown separately.