Introduction
A factory designed to assemble 10,000 robots per year has not necessarily produced 10,000 robots. Capacity, target capacity, annualized run rate, units built, shipments, deliveries and productive deployments describe different facts. This page keeps those fields separate and attributes company numbers to the publisher. Supplier yield, testing time, rework and design revisions can reduce usable output even when installed equipment has a higher theoretical rate.
Key findings
- Capacity, run rate, units built, shipments, deliveries and productive deployments are different metrics and should not be merged.
- 1X says its current facilities can produce up to 10,000 NEOs per year and targets 100,000+ annual capacity by the end of 2027; these are company capacity claims, not delivered-unit counts.
- AgiBot announced its 20,000th robot off the production line on September 24, 2026; production should remain separate from customer delivery and productive-use counts.
- A World Robot Conference report states that China shipped more than 40,000 humanoids in H1 2026 and accounted for 97% of global shipments under that report's definition.
- No common audited reporting standard currently makes capacity, production and shipment figures directly comparable across companies.
Humanoid Robot Manufacturing Capacity Claims vs Output - evidence comparison
The table records what each source establishes and keeps missing data visible.
| System or method | What the evidence establishes | Evidence class | Main unresolved point |
|---|---|---|---|
| Figure BotQ | Figure publishes factory and ramp updates, including company-reported production and delivery figures. | Company-reported manufacturing evidence | No common audit standard exists for humanoid output. |
| 1X NEO factory | 1X says its Hayward and San Carlos manufacturing footprint can produce up to 10,000 NEOs per year, with a company target of 100,000+ annual capacity by the end of 2027. | Company claim | Private factories disclose limited yield and inventory data. |
| Chinese manufacturers | AgiBot announced its 20,000th robot off the production line on September 24, while Unitree and UBTECH publish different sales, order or factory metrics that require separate labels. | Mixed company evidence | Rapid design revisions can strand installed tooling. |
| Tesla Optimus | Tesla states production ambitions, while current external sales and audited delivery data remain unavailable. | Target and internal program | No common audit standard exists for humanoid output. |
Evidence classification
- Officially documented specifications, standards or project status stated by the responsible organization.
- Real-system evidence demonstrations or deployments performed on physical hardware under described conditions.
- Company claim a numerical or operational statement reported by the company and not independently audited.
- Simulation or research evidence useful for mechanisms, but not proof of field deployment.
- Insufficient public evidence control mode, trial count, version or operating conditions are missing.
Definition and inclusion rules
Manufacturing capacity is the maximum sustainable output a production system can achieve under stated assumptions. Installed capacity refers to equipment already in place. Target capacity is a future goal. Units produced and delivered are observed outputs and should be reported separately.
How the ecosystem is mapped
Capture the exact metric and reporting period. Identify whether capacity is installed, current or targeted. Separate units built, shipments, customer deliveries and productive deployments. Check whether the number covers complete humanoids or a broader robot category, and keep reservations or orders separate from delivered output.
Organizations and evidence
Figure BotQ: Figure publishes factory and ramp updates, but capacity targets and observed deliveries remain separate fields.
1X NEO factory: 1X says current production capacity reaches 10,000 NEOs per year and plans to scale toward 100,000+ per year by the end of 2027. These are company statements about capacity.
China and AgiBot: the World Robot Conference report states more than 40,000 China humanoid shipments in H1 2026 under its definition, while AgiBot separately announced a 20,000-unit production milestone on September 24. Neither figure should be converted into productive deployments without customer-use evidence.
Tesla Optimus: Tesla publishes production ambitions for an internal program, while audited external shipment and delivery data remain unavailable. Those statements are treated as targets rather than observed customer output.
How country comparisons should be made
Cross-company comparison is valid only when the same metric is being reported. A capacity claim should be compared with another capacity claim, while production milestones, shipments, customer deliveries and productive deployments stay in separate columns.
Common classification errors
Common errors include annualizing a short peak run, counting reworked units twice, confusing component capacity with complete robots, treating reservations as paid orders, and ignoring yield, supplier constraints and final testing time.
Practical uses of the map
Useful applications include supply-chain planning, checking whether price targets fit stated production assumptions and separating market forecasts from observed output. The page does not convert capacity into demand or productive deployment without additional evidence.
Limitations and missing information
- No common audit standard exists for humanoid output.
- Private factories disclose limited yield and inventory data.
- Rapid design revisions can strand installed tooling.
- Specifications, prices, repositories and deployment status can change after publication.
- Benchmarks from different robots or environments are not directly comparable.
Conclusion
Humanoid manufacturing claims are useful only when the metric is explicit. Current public records include factory-capacity claims, production milestones and sector shipment estimates, but these do not describe the same thing. Track capacity, units built, shipments, deliveries and productive deployments separately before comparing manufacturers.
Frequently asked questions
How should humanoid robot manufacturing capacity be evaluated?
Record the exact metric and reporting period, identify whether capacity is installed, current, annualized or targeted, and keep prototypes, units built, shipments, customer deliveries and productive deployments separate.
Is the technology ready for practical use?
The page is useful for supply-chain and market analysis when capacity, production, shipments and deployments remain distinct. Productive use still requires separate evidence about customer sites, task hours and support burden.
Sources and methodology
Sources for humanoid robot manufacturing capacity were rechecked on September 26, 2026 using current official records from Figure, 1X, AgiBot and Beijing's World Robot Conference reporting. Company capacity and production figures remain attributed, and sector shipment figures retain the source's definition and scope.
Related TechniaHQRobot guides
Source review notes
Review date recorded July 11, 2026
Supported by the cited sources
- Figure publishes factory and ramp updates, including company-reported production and delivery figures.
- 1X publishes manufacturing plans and capacity claims tied to NEO production.
Not confirmed or incomplete
- No common audit standard exists for humanoid output.
- Private factories disclose limited yield and inventory data.
- Rapid design revisions can strand installed tooling.
Likely to change quickly
- Commercial availability, prices, model versions and software access.
- Deployment counts, company partnerships and repository maintenance status.
Share this article
Share the current TechniaHQRobot article page.
Continue reading
Open the latest robotics reporting, Physical AI analysis and hardware notes.