Industry and products · industry landscape
Industry landscape: where value can sit
Humanoid robotics is not a single OEM ranking. The useful map separates brain, body, integrator and deployment economics.
Updated 2026-06-15. Evidence map only; no trade recommendation.
Value-chain map
Six-layer operating stack
Sold: Robot body, fleet, platform and sometimes software or service bundle.
Missing: Actual units, robot revenue, gross margin, uptime, intervention rate, repeat paid orders and customer ROI/payback.
Sold: Robot foundation models, policies, teleop data loops, simulation, deployment tuning and developer tools.
Missing: ARR, license price, attach rate, deployed robot count, intervention reduction, retention, customer ROI and software gross margin.
Sold: Reducers, actuators, motors, sensors, batteries, compute, electronics, precision parts and test equipment.
Missing: Named humanoid customers, order volume, humanoid-specific revenue, per-robot content value and margin durability.
Sold: Implementation, uptime, maintenance, remote support, safety validation, SLA and outcome-based automation.
Missing: Robot count, contract value, intervention distribution, support burden, service gross margin and repeat expansion.
Sold: Labor substitution, throughput, quality, safety, downtime reduction and process redesign.
Missing: Customer-confirmed payback, utilization, uptime, safety case, repeat fleet orders and economics after service.
Sold: Imperfect public proxies such as TSLA, NVDA, Leaderdrive, UBTECH, automation suppliers and thematic baskets.
Missing: A clean mapping from public security exposure to the robotics value-capture layer, without trade framing.
Scenario lenses
Apple-style vertical robotics stack
Tesla/Figure-style closed stacks may control body, modules, model, deployment, service and data for premium or high-accountability workflows.
Watch: Robot gross margin, software/service attach, repeat paid deployments, lower intervention, field reliability and closed-loop data advantage.
Android-style open embodiment ecosystem
Low-cost bodies plus common model/tooling stacks could expand the installed base and push value toward software, data, tools and integrators.
Watch: Developer activity, common stacks, model attach rate, cross-body model performance, paid tooling and safety/reliability fragmentation.
NVIDIA / Google-style enabling platform
Compute, simulation, middleware, model tooling and inference runtimes can become horizontal control points if many OEMs converge on them.
Watch: Robotics-specific revenue disclosure, OEM attach, production deployment partnerships, developer adoption and fleet runtime using the platform.
Industrial-service hybrid
The hidden platform may be deployment knowledge: safety validation, workflow integration, SLA, maintenance and customer operations.
Watch: SLA revenue, support cost per robot, repeat expansion, maintenance margin, intervention rate and customer-confirmed payback.
Related pages
Robot Teardown · Key metrics · Why now · Robot foundation-model layer
Sources
- Figure BMW deployment post🟢 Figure
- Figure BotQ production ramp post🟢 Figure
- Figure 03 in-house module / supplier language🟢 Figure
- Unitree R1 product page🟢 Unitree
- Unitree G1 product page🟢 Unitree
- Unitree H2 product page🟢 Unitree
- Unitree H2 Plus product page🟢 Unitree
- Unitree G1-D product page🟢 Unitree
- Physical Intelligence π0🟢 PI
- Physical Intelligence openpi🟢 PI
- Physical Intelligence π0.7🟢 PI
- Skild AI Series C🟢 Skild
- Skild AI ABB/UR/NVIDIA partnership🟢 Skild
- Skild AI Zebra robotics arm acquisition🟢 Skild
- Leaderdrive 2025 annual report PDF🟢 filing