Hikvision made small custom batches behave like scale production

Hikvision's smart-factory problem was not one big product. China Industry News reported that in 2024 the factory made 256 million IoT devices across more than 30,000 hardware models, handled about 10,000 orders a day, averaged only 40 units per order and needed custom configuration for 75% of orders.

The answer was to make the factory search for operational sameness inside commercial variety. Its scheduling agent groups orders with similar process, material and furnace-temperature attributes, while its MPM process system uses BOM knowledge graphs and algorithms to generate routes and update process changes structurally.

That mechanism made frequent switching less destructive. China Industry News reported that intelligent order combining cut average SMT line changeover time by 50% to about 9 minutes, while process design time fell to under 10 minutes and process-design efficiency rose 60%. 36Kr also reported about 1,500 mobile robots raising logistics efficiency by about 40%.

Batch by process similarity, not by customer

The system found common manufacturing work beneath customized sales orders.

Structured process data turned engineering changes from spreadsheet searches into synchronized updates.

Robot delivery made the schedule executable by moving materials to people and stations on time.

Customization does not have to destroy scale economics if the factory batches hidden similarities rather than visible order labels.

Hikvision's Tonglu factory was listed in China's first leading-level smart-factory cultivation batch in 2025. The company also began exporting the same perception, scheduling and factory-automation capabilities to suppliers and outside industries including steel, electronics and machinery.