RoboSense

CONTENDER CPS 51

AI-driven robotics technology company providing core components and LiDAR solutions for the robotics and automotive industries.

Shenzhen, Guangdong, China·Founded 2014·~1,500 emp·2498.HK (Hong Kong Stock Exchange) · robosense.cn ↗ ↓ JSON ↓ MD
Researched 2026-02-19 ● Current
RoboSense — robotics.press intelligence card

RoboSense is a credible, operationally mature LiDAR supplier transitioning into a broader AI-driven robotics technology company, with demonstrated manufacturing scale (48,000 m², 95% automation), in-house chip design, and a sharp inflection in non-automotive robotics shipments (303,000 units in 2025, +1,142% YoY). However, the company remains loss-making on an LTM basis (~RMB -383M net loss), many leadership claims lack independent verification, and its strategic expansion into cameras, AI planning models, and dexterous hands introduces execution risk and potential channel conflict that temper the investment case.

Moat NARROW

- In-house chip design across four categories (RISC-V SoC, SPAD-SoC, VCSEL, MEMS) enabling vertically integrated cost and performance optimization - 48,000 m² manufacturing base with 95% automation rate providing scale production advantages - Automotive-grade certifications (ISO 26262, ISO 21434, IATF 16949, ASPICE CL2) creating qualification barriers for OEM/Tier 1 supplier relationships - Breadth of 3,400+ robotics customers across multiple verticals reducing switching incentives at ecosystem level

Management ADEQUATE

Leadership biographies were not available in the provided materials, limiting direct assessment. However, the company's operational execution—building a 48,000 m² automated manufacturing base, achieving automotive-grade certifications, listing on HKEX, and driving 1,142% YoY robotics unit growth—suggests competent operational management. The strategic pivot from pure LiDAR to full-stack robotics is ambitious and coherent but introduces execution risk that has yet to be validated by sustained profitability.

Financials PUBLIC
Bull Case

Robotics segment showed 41.5% gross margin in Q2 2025 on ~34,400 units and RMB 150M revenue, demonstrating that non-automotive robotics is a higher-margin growth vector than automotive ADAS

303,000 robotics LiDAR units sold in 2025 (+1,142% YoY) signals rapid adoption across lawnmowers, delivery robots, cleaning robots, and humanoids

In-house chip portfolio (RISC-V SoC, SPAD-SoC, 2D VCSEL, 2D MEMS) provides structural cost and performance advantages, reducing BOM costs and supply chain dependencies

Mars Intelligent Manufacturing Base (~48,000 m², 95% automation) with automotive-grade certifications (ISO 26262, ISO 21434, IATF 16949, ASPICE CL2) creates a manufacturing moat rare among sensor startups

Broad ecosystem of 3,400+ robotics customers and 350+ automotive OEMs/Tier 1s reduces single-customer dependency and increases probability of breakout platform wins

Bulk-production partnerships with European and North American warehouse-automation companies signal international diversification beyond China

Bear Case

Still loss-making on LTM basis (~RMB -383M net loss as of Sep 2025) with RMB 639M in R&D spend, and path to sustained profitability depends on unproven software monetization

Global No. 1 LiDAR market share claim and 2025 unit leadership figures rely on company statements and press releases without independent third-party verification

LiDAR market faces intense price competition and potential commoditization, particularly in cost-sensitive consumer and indoor mobile robotics segments, threatening the 41.5% robotics gross margin

Strategic expansion into cameras (Active Camera), AI planning models (VTLA-3D), and dexterous hands risks execution stretch and channel conflict with integration partners who may view RoboSense as a competitor

International scaling in EU/NA faces regulatory scrutiny, data localization requirements, and geopolitical risks for a China-headquartered sensor/AI company

Software/AI revenue contribution is not quantified in any available disclosure, raising questions about whether the AI stack is revenue-accretive or remains a cost center

Key Risks

LiDAR ASP compression from market commoditization and alternative sensing modalities (camera-only, radar fusion) could erode robotics segment margins below 35%

R&D intensity (~RMB 639M LTM) may not translate into monetizable software products, prolonging losses beyond 2027

Geopolitical and regulatory risks for a China-HQ company scaling in EU/NA, including potential data handling scrutiny and local content requirements

Customer concentration risk: warehousing and delivery programs may depend on a small number of integrators whose order volumes are undisclosed

Channel conflict as RoboSense moves up the stack into perception software, cameras, and end-effectors, potentially competing with current customers

Embodied intelligence products (dexterous hands, VTLA-3D planning) are early-stage with no disclosed paying customers or unit volumes

Catalysts

2025 audited annual report (expected H1 2026) confirming 303,000 robotics unit volumes, segment margins, and software revenue disclosure

Named multi-year bulk-production contracts with EU/NA warehouse automation customers with disclosed economics

Commercialization milestones for VTLA-3D AI planning model or dexterous hand with quantified revenue or customer deployments

Potential breakeven quarter as robotics mix shift lifts consolidated gross margins toward 30-35% range

Expansion into humanoid robot supply chains with a high-volume platform win from a major OEM

Irreplaceability 4
Market Weight
Tech Differentiation
Operational Deployment
Strategic Momentum
Ecosystem Influence
Coverage Necessity
Fin. Valuation
Fin. Revenue
TypeStandard Research
Published2026-02-19
Length4,873 words · 20 min read
Sources35 sources cited
↓ Download Report ↓ JSON

Generated by automated research. Cross-reference with primary sources before investment decisions.

Multi-DOF dexterous hand Handheld · PROTOTYPE
└─ Multi-degree-of-freedom dexterous hand with hand-eye coordination solutions for upper-body manipulation and lower-body mobility integration in embodied intelligence systems. Commercialization status including number of paying customers, unit volumes, and ASPs not yet publicly documented. Described as early-stage with no volume data disclosed.
VTLA-3D large model Software · PROTOTYPE
└─ Self-developed AI large model for long-horizon planning and autonomous execution of sequential actions in robotics, supported by in-house supercomputing infrastructure and scenario data toolchains. No independent benchmarks or peer-reviewed validations cited in available materials. Supported by in-house supercomputing infrastructure. Revenue contribution not yet disclosed.
Active Camera series (AC2) Sensor · LIMITED · Launched 2025
└─ Active camera system marketed as the 'Real Eye of Robots' for robotics perception applications, announced in 2025. Marketed as the 'Real Eye of Robots'. Featured at IROS and WRC events in 2025. Limited external independent validation available in public sources as of report date.
Digital LiDAR portfolio Sensor · FIELDED
└─ High-performance digital LiDAR lineup optimized for robotics and automotive ADAS applications, featuring in-house chip integration for performance and cost leverage. Company claims No. 1 global rank in robotics LiDAR for 2025. Bulk-production partnerships reported with European and North American warehousing companies. Named 2025 partners include Coco Robotics, Rhino Robotics, Kua Robotics, EasyGo Smart Driving, New Stone Robotics, and LionsBot.
LiDAR perception solutions Software · FIELDED
└─ Integrated hardware and AI perception software solutions that bundle LiDAR with perception algorithms to increase software revenue mix and customer lock-in. Bundles LiDAR hardware with AI perception algorithms targeting non-automotive verticals including cleaning, logistics, industrial operations, public services, and inspection. Software revenue contribution and attach rates not yet quantified in public disclosures.
In-house chip portfolio (RISC-V SoC, SPAD-SoC, 2D VCSEL, 2D MEMS) Sensor · FIELDED
└─ Comprehensive in-house semiconductor development spanning digital computing chips, digital large-area SPAD-SoCs, 2D addressable VCSELs, and 2D MEMS devices for LiDAR and sensing applications. Enables tighter LiDAR integration, BOM cost reduction, and supply chain resilience. Supports 95% automated manufacturing at Mars Intelligent Manufacturing Base (~48,000 m²). Claims are company-stated and have not been independently verified via teardown analyses in available sources.
Lau Wing Kee (Kelvin) Executive Director
Qiu Chunxin Chairman, Chief Scientist, Executive Director, Co-Founder
Qiu Chunchao (Mark Qiu) CEO, Executive Director
RoboSense Media Contact
3D tracking L3 · Radar
Obstacle avoidance L3 · Navigation
Radar L2 · Detection
Multi-robot orchestration L3 · C2 / Fleet Management
LIDAR mapping L3 · Visual Detection
Perimeter Patrol L2 · Patrol & Surveillance
Multi-sensor fusion L3 · Visual Detection
AI / Analytics L2 · Autonomy & Software
Navigation L2 · Autonomy & Software
Autonomous route following L3 · Perimeter Patrol
C2 / Fleet Management L2 · Autonomy & Software
Autonomy & Software L1
Mission planning L3 · C2 / Fleet Management
Computer vision L3 · AI / Analytics
Visual Detection L2 · Detection
Patrol & Surveillance L1
Data fusion L3 · AI / Analytics
SLAM L3 · Navigation
Detection L1

News & Analysis

2
RoboSense: Company ProfileAnalysis
2026-05-02
RoboSense: Company ProfileAnalysis
2026-03-06