BYD's push into smart manufacturing is piecing together a critical part of its puzzle. From investing in Paxini to fill the gap in tactile perception, to taking a stake in AgiBot to coordinate complete-robot capabilities, BYD's industrial map around embodied intelligence is becoming increasingly clear. Within the closed loop of "core components—complete robots—application scenarios," spatial perception—the "visual foundation" that allows robots to recognize their environment and move autonomously—remains an unavoidable link. Now, that link may be filled by a company about to list on the Hong Kong Stock Exchange.
According to Zhitong Finance APP, on September 30, CAMSENSE (06802) will be listed on the Main Board of the Hong Kong Stock Exchange. It is reported that Golden Link, a wholly-owned subsidiary of BYD, subscribed to its cornerstone shares for approximately HK$110 million, with industrial capital as the sole anchor investor—clearly pointing to more than just financial returns.
One. The "First Mile" of Physical AI: Why Spatial Perception Is Critical
In the technology stack of embodied intelligence, large models are responsible for "thinking," motion control handles "moving," and spatial perception handles "seeing." A robot must first understand where it is, what surrounds it, and how to reach its target before it can complete any meaningful physical task. This capability is the first gateway for physical AI to move from the virtual to the real world.
CAMSENSE has been deeply engaged in this field for more than a decade. Founded in 2013, by 2025 revenue, CAMSENSE ranks first globally in spatial perception solutions for intelligent robots, first globally in LiDAR shipments, with a global market share exceeding 50%. As of the latest practicable date, cumulative shipments have exceeded 39 million units, and in 2025, total shipments of LiDAR and line laser sensors exceeded 10 million units and 5 million units respectively. The two founders also possess decades of technical experience and R&D backgrounds, with one having a mechanical manufacturing and semiconductor industry background from Xi'an Jiaotong University, and the other having research experience in precision instruments from Tsinghua University and Bell Labs in Chicago. But what truly caught industrial capital's attention is the completeness of its technology chain—the company has established a full self-developed closed loop spanning "ASIC chips—AI spatial positioning algorithms—high-precision optical design—mass production at the tens-of-millions scale."
CAMSENSE is one of the few companies globally that simultaneously possesses self-developed spatial perception AI algorithms and self-developed ASIC chips. Its self-developed sub-millimeter-level AI spatial data processing chip has been deployed in over 25 million terminal devices, making it the first company in China to achieve monocular large-space sub-millimeter positioning accuracy and the world's first to self-develop spatial positioning algorithm chips. Algorithms determine the "intelligence" of perception, chips determine the "efficiency" of perception, and tens-of-millions-scale mass production determines the "reliability" and "cost" of perception. Together, these three elements form the core requirements of spatial perception infrastructure in the era of physical AI.
Two. How Can Tens-of-Millions-Scale Mass Production Capabilities Be Transferred to Automotive Smart Manufacturing?
Consumer-grade robots and industrial production-line robots may appear to operate in vastly different scenarios, but their underlying requirements for spatial perception are cut from the same cloth: high precision, high reliability, and large-scale deliverability. Robots in home environments need real-time mapping, dynamic obstacle avoidance, and autonomous recharging, facing unstructured and complex spaces. Meanwhile, robots on automotive final assembly lines need to complete positioning and grasping coordination during high-speed motion, placing higher demands on sensor consistency, anti-interference capability, and chip-level edge computing power. The sub-millimeter positioning accuracy verified by CAMSENSE in consumer scenarios, the edge processing efficiency of its self-developed ASIC, and the quality control system refined through tens-of-millions-scale production are precisely the capability combination most scarce in industrial scenarios.
BYD's industrial layout provides room for imagination for this transfer: embodied intelligent robots and the new energy vehicle industry chain are highly homologous, and technology, components, and production line capabilities can be transferred and reused. Currently, BYD has deployed more than 2,000 AI visual inspection nodes in core processes such as welding, painting, and final assembly. AI collaborative robots handle precision operations such as door installation and wiring harness connection, compressing cycle time by 15%. The prospectus materials also indicate that BYD's cooperation with CAMSENSE points to the incremental market for industrial robots, and CAMSENSE's LiDAR and depth perception technologies are expected to expand into industrial robot spatial perception scenarios.
Three. From Home to Industry: Higher-Dimensional Reuse of the Same Perception Foundation
The global robot 3D visual perception market is expected to reach RMB 11.7 billion by 2030 and further climb to RMB 152 billion by 2035. The installed base of AI + 3D vision-guided general intelligent robot components is projected to grow from 30,100 units in 2025 to 239,900 units in 2030, with penetration rising from approximately 5.1% to 10.6%. As automotive manufacturing evolves toward flexibility and intelligence, high-precision, highly reliable, and large-scale deliverable spatial perception solutions are expected to become key infrastructure in smart manufacturing systems.
CAMSENSE's technological reach has been quietly extending. According to the prospectus, the company began investing in pre-research of new technologies and new business scenarios starting in 2024, and in 2025 formally engaged with lawn mower robot, pool robot, and certain fleet customers. According to market sources, the company has sent samples and conducted testing with multiple pool cleaning, lawn mowing, and humanoid robot manufacturers, and has received non-binding letter-of-intent order demand from 2 pool cleaning robot brands and 1 lawn mowing robot customer, which is expected to generate sales revenue in the future. The expansion into these new scenarios is essentially the reuse of the same spatial perception capability across different dimensions. From home cleaning to outdoor lawn mowing, from pool cleaning to industrial production lines, the robot's "visual foundation" must cope with progressively increasing environmental complexity, but the underlying technological logic remains consistent.
CAMSENSE's fundraising purposes also explicitly include optimizing its 3D spatial perception AI algorithms and next-generation multi-sensor fusion embodied robot perception solutions. From a financial trajectory perspective, CAMSENSE is at a point where scale effects are beginning to be released. From 2023 to 2025, the company's revenue grew from RMB 332 million to RMB 614 million, with a compound growth rate of 35.9%, achieving annual profitability for the first time in 2025. In the first quarter of 2026, revenue grew 48.7% year-on-year to RMB 196 million, with net profit of RMB 7.757 million, reaching 3.5 times the full-year 2025 figure. As new categories such as dTOF LiDAR and line laser sensors accelerate in volume, the physical AI product structure is moving from单一 to diversified.
BYD's cornerstone investment may be just the prologue to this transfer. In the process of physical AI moving from virtual to real, spatial perception is the first gateway for robots to perceive the physical world. CAMSENSE has spent more than a decade turning a sub-millimeter chip into standard equipment for more than half of the world's intelligent robots, and broader scenarios such as industrial smart manufacturing, embodied robots, and XR are waiting to be empowered by the same technological foundation. With the backing of industrial capital from a leading automaker, CAMSENSE is expected to leverage its technology foundation validated through tens-of-millions-scale mass production to break out of the consumer robot track, open a second growth curve in the automotive smart manufacturing sector, and become a perception infrastructure player not to be ignored in the era of physical AI.