China Robot Factory Capacity: Build a Delivery-Readiness Ledger

A robot factory's annual capacity figure is a starting point for a delivery discussion. A buyer needs a narrower answer: how many robots of the agreed configuration can pass acceptance for this order, during the weeks when the buyer needs them?

A useful factory visit connects the announced capacity to a dated production allocation, the parts that constrain it and the test that releases each finished unit. The worksheet below helps a procurement or operations team make that connection.

What the Beijing factory report establishes

In its 21 October 2026 report on Lingyi's Beijing embodied-intelligence factory, Beijing Daily describes an automated robot-joint line planned to enter service by year-end. The report gives planned annual capacity of 10,000 robots for 2026 and 20,000 for 2027. Those are planning figures attributed to the factory, rather than an audited count of accepted deliveries.

The same report describes existing assembly and traceability practices. Production head Xia Wei says each robot's serial number connects materials, screw torque, joint calibration and movement records. It also describes final assembly on an overhead conveyor taking 15 minutes. That duration concerns an assembly stage; it does not supply a measured rate for the whole route from incoming components through testing and customer acceptance.

For a buyer, the important gap is the connection between those stages. Joint availability, product variants and testing can change the quantity available for a particular delivery window. China AI Wire's related coverage provides a short account of the factory plans.

Build a ledger for one order

Choose one robot configuration and one delivery period before discussing capacity. Record the hardware revision, firmware baseline, joint bill of materials and the test specification that defines an accepted unit. An aggregate figure spanning several robot models cannot answer the same question without a product-mix assumption.

Use a separate row for each constraint:

Constraint	Evidence to request	Quantity to record
Critical components	Available parts, incoming inspection and confirmed replenishment dates	Usable parts allocated to this order
Joint production	Recent completed batches, first-pass yield and rework records	Good joints per week for the required variants
Assembly	Completed units for the same configuration and staffed working hours	Assembly slots available after other commitments
Calibration and final testing	Test duration, available stations, failures and retest workload	Units that can finish the agreed acceptance process
Release and dispatch	Serial-number release records, unresolved defects and dispatch allocation	Accepted units ready for shipment by date

Ask the operating team to reconcile the rows against the same time period. A component forecast for next month and a testing record from last quarter do not establish next week's available output. Keep planned additions separate until commissioning and production records support their use in the calculation.

The records can be redacted. The useful connection is between a dated batch, its configuration, the quantity entering each stage and the quantity released. A photograph of installed equipment does not establish that connection.

Worked example: find the limiting stage

Illustrative planning exercise. These numbers are assumptions, not Lingyi operating data.

Suppose a robot uses 30 joints. The supplier allocates 1,200 joint-production slots per week to this order, after its other customer commitments. Assume the required joint variants are available in the right proportions and first-pass yield is 90%, with no recovered rework included.

That gives 1,080 good joints, enough for 36 complete robots per week. Counting all 1,200 production attempts as usable joints would overstate what assembly can receive.

Now suppose assembly has 50 weekly slots for the order, while the calibration and final-test operation can release 28 accepted robots per week, including its normal retest workload. Assume all other required parts are available.

The planning bound is the smallest of those stage capacities:

Minimum of 36 joint-supported robots, 50 assembly slots and 28 accepted test outputs = 28 accepted robots per week.

A 100-unit order therefore needs at least four production weeks under these assumptions. That is a capacity calculation, not a delivery promise: commissioning, ramp-up, missing components, changed configurations and transport can extend the calendar. If the testing allocation falls to 20, a faster assembly line does not restore the original bound.

Use the supplier's actual stage records to replace each assumption. If the team can only provide gross factory capacity, leave the order-specific result unresolved.

Take a small batch through the records

During a permitted walkthrough, select a completed batch and trace its serial numbers through material allocation, joint calibration, final testing and release. Check how a failed unit returns for rework and whether a later pass is counted as a new production attempt or as recovery of the same unit.

Then compare the batch with the proposed order. A different joint variant, payload or firmware baseline may need a different test route. Ask which evidence supports transferring the observed rate and which stage needs a new trial. The robotics factory-visit guide covers the wider operating and support questions; this ledger supplies the quantity-and-date calculation.

End the session with a dated sheet showing allocated capacity, the limiting stage, unresolved assumptions and the person responsible for each missing record. Use that sheet to shape the factory-workstream brief for an industry-focused China AI program.