Abstract
LiDAR (Light Detection and Ranging) has earned the nickname "the eye of autonomous driving." This is not marketing copy — it accurately describes the sensor's role in the perception chain of self-driving systems. Millimeter-wave radar can measure distance; cameras can detect color; but for precise, real-time three-dimensional point clouds that remain reliable under any lighting condition, LiDAR remains the most irreplaceable option.
The global LiDAR industry crossed a genuine historical inflection point in 2025. That year, global automotive LiDAR shipments reached approximately 2.5 million units, growing over 150% year-on-year, with Chinese manufacturers capturing more than 80% of the global share. Three events converged to drive this milestone: Hesai Technology's ATX series entered mass production across 24 automakers including BYD, Xiaomi, and Li Auto; RoboSense's MX product brought the unit selling price down to approximately RMB 1,500; and the penetration rate of LiDAR in Chinese passenger vehicles jumped from 6% in 2024 to 21% in 2025, with 2.58 million new vehicles equipped.
Meanwhile, overseas LiDAR companies were experiencing a very different fate. Luminar Technologies filed for Chapter 11 bankruptcy in December 2025. Innoviz continued to lose money. Ouster, having merged with Velodyne, saw its market share keep shrinking. Global LiDAR industry dominance has shifted systematically to China — not as a temporary cost-advantage market capture, but as a structural change driven by comprehensive leads in mass-production capability, proprietary chip development, and deep customer integration.
This report uses 2026 as its observation point to systematically review the technology taxonomy, supply chain structure, Chinese market size, key company financials, industrial belt geography, sub-market evolution, technology route competition, risks, and forecasts for 2026–2030 in the automotive LiDAR industry.
Core conclusions:
First, the sub-RMB-2,000 price inflection point became real in 2025. This is not the spoils of a price war but the outcome of chipification technology moving from the laboratory to the production line. RoboSense MX (~RMB 1,500) and Hesai ATX (sub-RMB 2,000 target) mark LiDAR's genuine entry into standard-equipment competition for 100,000–200,000 RMB passenger vehicles.
Second, China's four-vendor bloc (Hesai, RoboSense, Huawei, Seyond) collectively holds approximately 88% of global automotive LiDAR market share — a structure that will not fundamentally change before 2030. The survivable paths for overseas companies are narrowing to niche professional markets.
Third, robotics LiDAR is transitioning from "future story" to "current-quarter revenue." RoboSense's 2025 full-year robot LiDAR shipments reached 303,000 units (global #1, +1,141.8% YoY); in Q1 2026, robot shipments surpassed ADAS shipments for the first time. Hesai's JT series grew +649.1% YoY in H1 2025.
Fourth, the commercialization of L3 autonomous driving (first batch of approvals in December 2025) and Robotaxi scaling (Baidu Apollo's Luobo Kuaipao and others) will generate incremental demand, forming a "dual engine" alongside passenger-vehicle penetration growth.
Fifth, Tesla's pure-vision approach and 4D millimeter-wave radar substitution are the two main pressure sources, but both are more likely to act as cost-reduction catalysts for LiDAR than genuine replacement threats.
Key data highlights:
- China automotive LiDAR market: ~RMB 112 billion (2024), ~RMB 241 billion (2025E), ~RMB 432 billion (2026E)
- China passenger-vehicle LiDAR penetration: 21% (2025), ~27% (2026E), target 60%+ (2030E)
- Hesai Q2 2025 net profit exceeded RMB 40 million; full-year net profit guidance raised to RMB 450 million
- RoboSense FY2025: 912,000 units shipped, revenue ~RMB 1.94 billion; first quarterly profit in Q4 (net profit RMB 104 million)
- RoboSense MX ~RMB 1,500 — first mass-produced sub-RMB-2,000 automotive-grade LiDAR
- Luminar filed Chapter 11 bankruptcy; Chinese vendors absorb global market share
- Global 2025 shipments ~2.5 million units; CAGR 2025–2030E ~40%
Chapter 1: Definitions, Classification, and Full Industry Chain
1. What Is LiDAR?
LiDAR's operating principle can be summarized in one sentence: emit laser pulses toward a target, measure the round-trip flight time (Time of Flight, ToF), multiply by the speed of light and divide by two to obtain the target distance. By scanning the surrounding environment at high frequency and multiple angles, LiDAR can generate a dense real-time three-dimensional point cloud that provides the perception system with precise distance, orientation, and shape information.
Compared to cameras, LiDAR does not rely on ambient lighting — it performs stably in the dark and in strong backlight. Compared to millimeter-wave radar, LiDAR's angular resolution is orders of magnitude higher (millimeter-wave: typically 1°–3°; LiDAR: 0.05°–0.1°), clearly distinguishing pedestrians, cyclists, and road signs rather than returning a blurry echo blob. This dual advantage in range accuracy and spatial resolution is the fundamental reason LiDAR holds a core position in L2+ and higher autonomous driving systems.
Key LiDAR specifications include: line count (channels) — determines vertical resolution; horizontal field of view (HFOV) — typically 100°–120° for automotive forward-facing units; detection range — mainstream automotive: 100–300 m; angular resolution — 0.1°–0.2° for mainstream products; point cloud density (points/s) — 100k–several million points per frame; range accuracy — typically ±2–5 cm.
2. Classification by Measurement Principle: ToF vs. FMCW
Direct Time-of-Flight (dToF) is used by over 90% of current automotive LiDAR. The system emits short laser pulses and records return arrival times using high-precision time-to-digital converters (TDC) or high-speed samplers. Hesai ATX and RoboSense MX both use dToF + 905 nm.
Frequency-Modulated Continuous Wave (FMCW) is the next-generation technology candidate. By transmitting continuously frequency-modulated laser signals, it simultaneously derives target distance and radial velocity (Doppler shift), providing native "4D sensing" (3D position + velocity vector) without multi-frame inference. FMCW also delivers strong anti-interference performance. Representative companies: Aeva (US, AEVA on Nasdaq), Aurora (US, AUR), Seyond (China, partial exploration). However, FMCW demands extreme laser coherence and signal-processing chip sophistication; current mass-production cost is several times that of dToF. No large-scale mass-production automotive deployment exists as of 2025; industry consensus puts broad FMCW mass production no earlier than 2027–2028.
3. Classification by Wavelength: 905 nm vs. 1550 nm
905 nm is the mainstream wavelength for current automotive LiDAR, using GaAs-based VCSEL (vertical-cavity surface-emitting laser) or EEL (edge-emitting laser). Mature semiconductor processes give lower costs; domestic suppliers include Changguang Huaxin (Suzhou, 688048) and Ruibo Photoelectronics (Shenzhen). The key limitation is stricter eye-safety power constraints (IEC 60825-1 limits peak power more tightly at 905 nm), restricting maximum detection range under strong sunlight backgrounds.
1550 nm uses InP-based semiconductor lasers or erbium-doped fiber amplifiers (EDFA). Eye-safety power limits are more relaxed at 1550 nm (allowing ~50× higher peak power for the same safety level), enabling detection ranges beyond 300 m and better fog/rain penetration. Representative vendors: Seyond (Falcon series, China), Luminar (US, now bankrupt). Key limitation: 1550 nm lasers and detectors (InGaAs APD/SPAD) are significantly more expensive, with per-unit costs remaining at hundreds to thousands of RMB as of 2025, restricting mass-market penetration.
4. Classification by Scanning Mechanism
| Scanning Type | Representative Products | Advantages | Limitations |
|---|---|---|---|
| Mechanical spinning | Velodyne HDL-64E (early benchmark) | 360° panoramic, mature | Bulky, expensive, not automotive-grade |
| Prism/mirror (semi-solid) | Hesai AT128, ET25 | Mature mass-production, high accuracy | Limited FoV (within 120°) |
| MEMS micromirror | RoboSense MX, M-Platform | Compact, low-cost, chipified | FoV/scan-rate tradeoff |
| Flash (pure solid-state) | Short-range blind-spot products | No moving parts, high reliability | Short range, limited resolution |
| OPA (optical phased array) | Academic/early-stage | Electronic beam steering | Not yet manufacturable at scale |
The current mainstream automotive mass-production route is semi-solid (MEMS/prism) + 905 nm + dToF, the technical essence behind both Hesai ATX and RoboSense MX.
5. Full Industry Chain
The LiDAR supply chain has four tiers:
Upstream core components: laser emitters (905 nm VCSEL/EEL — Changguang Huaxin leads domestically; 1550 nm still heavily import-dependent); SPAD/APD detectors (SPAD with standard CMOS integration is now mass-producing for 905 nm systems); MEMS micromirrors (chipification is the cost-reduction key; domestic supply chain not fully mature); signal-processing SoC (Hesai self-developed Pandar SoC, RoboSense M-Platform SoC — core cost-control moats); optical elements (mostly domestically sourced).
Midstream integration: Hesai (ATX/AT128/JT), RoboSense (MX/E1/RS-Helios/robot series), Huawei (embedded in ADS system), Seyond (Falcon 1550 nm series), Benewake (robot/short-range specialist), LeiShen Intelligent.
Downstream OEMs and applications: passenger-car OEMs (BYD, Xiaomi, Li Auto, AITO/Huawei, NIO, Chery, Geely, etc.); commercial vehicles (Shaanxi Heavy Truck, DeepWay); Robotaxi (Baidu Apollo/Luobo Kuaipao, WeRide, Pony.ai); robots (sweeping robots, AMR/AGV, humanoid); roadside (Wanjit Technology, Huawei Smart Road).
Chapter 2: Global Competitive Landscape and Technology Route Disputes
1. Global Landscape: China's Four Vendors Dominate; Overseas in Crisis
By 2025, the global automotive LiDAR competitive landscape is clear: Chinese companies dominate; overseas companies are marginalized and collapsing. The four Chinese vendors — Hesai, RoboSense, Huawei, Seyond — collectively hold approximately 88% of the global automotive LiDAR market by shipment volume.
2024 global automotive LiDAR market share (by installation volume):
| Vendor | 2024 units (10k) | Share (%) | HQ |
|---|---|---|---|
| RoboSense | ~4.56 | ~33.5–34.7 | Shenzhen, China |
| Huawei | ~3.51 | ~26–27 | Shenzhen, China |
| Hesai | ~3.21 | ~24–25 | Shanghai, China |
| Seyond | ~1.34 | ~10–13 | Suzhou, China |
| Others (incl. overseas) | <0.5 | <4 | — |
In 2025, the dynamics evolved further: Hesai, with ATX deployed across more automakers, retook the single-quarter shipment lead in Q1 2025 with 195,818 units, and extended its lead in Q2 with 352,095 units. Its full-year shipment target was 1.2–1.5 million units; cumulative shipments exceeded 1 million by September. RoboSense surged in the robotics segment with 303,000 robot LiDAR units for full-year 2025 (global #1) and achieved first-ever quarterly profitability in Q4 (net profit RMB 104 million).
The most significant overseas event of 2025 was Luminar Technologies (LAZR, Nasdaq) filing Chapter 11 bankruptcy in December 2025 — the largest automotive LiDAR company bankruptcy in global industry history. At peak (2021), Luminar's market cap exceeded $9 billion. Its 1550 nm ToF technology was technically well-regarded (detection range >500 m), but commercial timing lagged far behind Chinese competitors: FY2025 revenue was only $67–74 million (RMB 500 million), about 1/6 of Hesai's and 1/4 of RoboSense's; it depended primarily on a single customer (Volvo EX90 program) at far lower-than-expected production volumes; and L3 autonomous driving commercialized far more slowly than projected. Luminar's bankruptcy is the strongest evidence of the Western LiDAR camp's systemic commercial failure.
2. Technology Route Dispute 1: LiDAR vs. Pure Vision
Tesla is the most powerful champion of the pure-vision approach. Elon Musk has repeatedly stated that "anyone who relies on LiDAR is doomed." FSD removed LiDAR entirely in 2022 and relies solely on cameras plus neural networks. In 2025, FSD v13/v14 showed further improvements in some complex scenarios, boosting support for the pure-vision camp.
However, China's mainstream judgment moved in the opposite direction: Xiaomi returned to LiDAR configurations on flagship models in 2025; Li Auto, Xiaomi, and AITO/Huawei hot-selling brands equip their full lineups; L3 approval requirements mandate sensor redundancy; consumer surveys show Chinese buyers have a stronger preference for "LiDAR = safer" compared to US buyers.
Conclusion: Tesla's pure-vision approach poses limited direct threat to China's LiDAR market. Its deeper impact is indirect — forcing LiDAR vendors to continue reducing costs until "the price difference vs. pure-vision is something consumers barely notice." This is positive market pressure, not a substitution threat.
3. Technology Route Dispute 2: 4D Millimeter-Wave Radar vs. LiDAR
4D imaging millimeter-wave radar adds height dimensionality to traditional 3D mmWave, generating limited-resolution 3D point clouds while retaining all-weather (rain/snow/fog) advantages. Mass-production cost is approximately RMB 1,000–1,500 — now comparable to RoboSense MX's selling price, making this the most intense cost competition segment.
In January–May 2025, among vehicles priced below RMB 200,000, 53.94% were equipped with 4D mmWave radar. Huawei's Zunzhan S800 features 3 × 4D mmWave radars plus LiDAR fusion. Core conclusion: In L2+ ADAS in the cost-sensitive segment (sub-200,000 RMB vehicles), 4D mmWave radar and LiDAR genuinely compete. In L3+ safety-redundancy scenarios, industry consensus is multi-sensor fusion (camera + LiDAR + 4D mmWave) — not three-way elimination. 4D mmWave is LiDAR's strongest cost-reduction driver, not its existential threat.
Chapter 3: PEST Analysis — Policy, Economic, Social, Technology Environment
1. Policy: L3 Approval Lands; Smart-Connected Policy Shield
In December 2025, China's Ministry of Industry and Information Technology announced the first batch of L3-level conditional autonomous driving vehicle product approvals: two pure-electric sedans (Deepal Automotive under Changan, and Arcfox under BAIC) were approved for commercial sales and on-road testing on designated highway and urban expressway sections in Beijing and Chongqing — a milestone transition from "testing demonstration" to "licensed commercial sales."
For the LiDAR industry, this policy milestone means: L3 approval requirements explicitly mandate sensor redundancy, and multi-sensor fusion architectures including LiDAR are the standard prerequisite for passing safety tests. As the number of L3-approved vehicle models expands (expected multiple automakers through 2026–2027), LiDAR's institutional status as a "required sensor" will be further reinforced.
Data security regulations (Personal Information Protection Law 2021, Data Security Law 2021, Automotive Data Security Regulations 2021) create soft barriers for overseas LiDAR vendors entering Chinese supply chains. US EAR export controls have not yet added LiDAR devices to restricted lists, but upstream 1550 nm InP epitaxial wafers face increasing scrutiny, adding supply security risk premium for China's 1550 nm route (Seyond primarily).
2. Economic: The Sub-RMB-1,500 Inflection Point and Scale Economics
RoboSense MX at RMB 1,500 ($200) is the first mass-produced automotive-grade LiDAR product to break the sub-RMB-2,000 threshold. Three simultaneous technology milestones enabled this: MEMS micromirror SoC integration (M-Platform), SPAD detector in standard CMOS process manufacturing, and proprietary SoC reducing BOM costs.
A scale-economics positive loop is now operating: Hesai achieved Non-GAAP profitability in 2024 at 501,000 units; with 2025 full-year guidance of 1.2–1.5 million units and net profit guidance of RMB 450 million, the cost-reduction speed is outpacing price declines. RoboSense's gross margin rose from ~8% in 2022 to 17.2% in 2024, confirming scale effects are working.
3. Social: NOA Adoption and User Perception Changes
NOA (Navigate on Autopilot) penetration in the RMB 200,000–250,000 price band rose from 2.1% in January 2024 to 24.7% by October 2024. Projected overall passenger-car NOA penetration reached 20% by end of 2025. User acceptance of "the car drives itself" smart driving is rapidly improving, with positive associations for LiDAR as a safety component.
4. Technology: Chipification and AI Integration
Two main technology progress tracks in 2025: first, chipification integration — RoboSense M-Platform SoC chipifies the full stack (scanning/processing/transmit-receive); Hesai Pandar SoC achieves similarly high integration. Second, AI and LiDAR deep fusion — leading products are deploying lightweight AI models onto LiDAR SoCs for "in-sensor perception," outputting semantic detection results rather than raw point clouds, enabling the transition from "raw sensor" to "intelligent sensing node."
Chapter 4: China Market Size — Shipments, Pricing, Penetration
1. Shipment Volume: From 1.53 Million to 2.5 Million+
China is the world's largest single LiDAR market. 2024 automotive LiDAR installation volume reached 1.53 million units (+178% YoY). 2025 full-year shipments (with RoboSense MX and Hesai ATX ramping) are estimated at 2.4–2.6 million units, central estimate ~2.4 million.
Cross-validation: Hesai target 1.2–1.5 million (cumulative >1 million by September); RoboSense full-year 912,000; Huawei + Seyond ~500,000–600,000; four-vendor total ~2.6–3.0 million (some include robotics/Robotaxi outside of passenger-car front-install).
2. Pricing: Sub-RMB-1,500 Inflection Has Landed
| Product Type | 2022 typical ASP | 2025 typical ASP | 3-Year Decline |
|---|---|---|---|
| Primary forward (mass-production) | RMB 8,000–15,000 | RMB 1,500–3,000 | -70% to -80% |
| Short-range blind-spot solid-state | RMB 3,000–6,000 | RMB 500–1,500 | -75% |
| Premium 1550 nm solid-state | RMB 30,000+ | RMB 8,000–20,000 | -40% to -50% |
| Robotics low-cost single-point | RMB 300–1,500 | RMB 100–500 | -60% |
RoboSense MX at ~RMB 1,500 is an industry milestone price point — the first to genuinely fall below the "sub-RMB 2,000" threshold for automotive-grade front-install LiDAR.
3. Penetration Rate: Jumping from 6% to 21%
China passenger-vehicle LiDAR installation rate: 6% in 2024 (1.36 million vehicles); 21% in 2025 (2.58 million vehicles). 2026E ~27%.
By price segment: above RMB 200,000 — nearly standard (>50% penetration); RMB 150,000–200,000 — rapid penetration begins in 2025 (MX/ATX target range); RMB 100,000–150,000 — main penetration target for 2026–2027.
4. Market Size: From RMB ~112 Billion to RMB ~1 Trillion
| Year | Shipments (10k units) | Avg. ASP (RMB) | Market Size (RMB 100M) |
|---|---|---|---|
| 2023 | ~57 | ~4,500 | ~26 |
| 2024 | ~153 | ~2,800 | ~112 |
| 2025E | ~240 | ~2,200 | ~241 |
| 2026E | ~380 | ~1,900 | ~432 |
| 2028E | ~750 | ~1,600 | ~650 |
| 2030E | ~1,200+ | ~1,400 | ~1,000+ |
Chapter 5: Supply Chain Deep Dive — From Laser to Complete System
1. Laser Emitters: 905 nm Domestication Accelerates; 1550 nm Still Import-Dependent
905 nm VCSEL/EEL domestic supply rate now exceeds 60%. Key domestic vendors: Changguang Huaxin (Suzhou, 688048) — domestic VCSEL leader, monthly capacity of tens of millions of units; Ruibo Photoelectronics (Shenzhen); Lemon Photonics (Shenzhen); Ruijing Laser (Wuhan) — high-peak-power EEL for long-range applications.
1550 nm lasers (InP-based) remain heavily import-dependent from US companies (II-VI/Coherent, nLIGHT), constrained by US EAR — the main supply-security risk for China's 1550 nm route.
2. SPAD/APD Detectors: Domestic Breakthroughs Underway
SPAD (single-photon avalanche diode) working in Geiger mode has become the mainstream choice for front-install mass-production products. The breakthrough: standard CMOS process SPAD chips (28 nm/55 nm node at TSMC/SMIC) are now in mass production at far lower cost — a core enabler of MX and ATX hitting sub-RMB-2,000 pricing. InGaAs SPAD/APD (for 1550 nm) remains costly and in early-stage domestic development.
3. MEMS Micromirrors: Full SoC Chipification Revolution
MEMS micromirror integration into SoC (RoboSense M-Platform) is the key to solving batch consistency — resonant frequency variations between wafer batches are compensated via in-SoC closed-loop control algorithms, enabling factory-calibration-free or simplified-calibration mass production. SCAN-X and other domestic players are active in MEMS chip development, but the supply chain is not yet fully mature.
4. Signal-Processing SoC: Self-Development Is the Core Moat
Hesai Pandar SoC: integrates laser timing control, SPAD array readout, TDC, and DSP into a single chip. RoboSense M-Platform SoC: full-stack integration of scan control, VCSEL driving, SPAD sampling, and raw point cloud processing — enabling MX's sub-RMB-2,000 pricing. Companies without self-developed SoCs rely on TI/Mobileye solutions at RMB 200–500 higher BOM cost per unit, with less algorithm customization.
5. BOM Cost Structure (Primary Semi-Solid Forward Automotive-Grade Reference)
| Cost Item | % of BOM | Domestic Supply Level | Key Risk |
|---|---|---|---|
| Laser emitter module (VCSEL+driver) | 20–30% | 905 nm mature (>60%) | 1550 nm import-dependent |
| Detector module (SPAD+TIA) | 15–25% | 905 nm advancing; 1550 nm early | SPAD domestic consistency |
| Scanning element (MEMS/prism) | 10–20% | MEMS gap; prism mature | MEMS consistency |
| Signal-processing SoC | 15–20% | Leaders self-develop; others import | External procurement cost |
| Optical elements | 10–15% | Mature (>70%) | Low |
| Structure/PCB/assembly | 10–15% | Mature | Low |
Chapter 6: Key Company Deep Analysis
1. Hesai Technology (HSAI, Nasdaq)
Founded 2014 in Shanghai. Listed on Nasdaq in February 2023 (HSAI). The world's first and only listed LiDAR company to achieve full-year GAAP profitability.
Financial trajectory: 2024: revenue RMB 2.08 billion (+10.7%), Non-GAAP net profit ~RMB 14 million, shipments 501,000 units (+126%); 2025 Q1: revenue RMB 530 million (+46.3%), shipments 195,818 units (+231.3%); 2025 Q2: revenue RMB 710 million (+50%+), shipments 352,095 units (+306.9%), net profit >RMB 40 million; 2025 Q3: profitable, full-year net profit guidance raised to RMB 450 million. 2025 guidance: revenue RMB 3.0–3.5 billion, GAAP profit RMB 200–350 million.
Product portfolio: ATX (sub-RMB 2,000 forward mass-production, 24 automakers / 120+ model designations including BYD, Xiaomi, Li Auto, SAIC-Audi); AT128 (128-line prism-mirror, premium forward); ET25 (solid-state blind-spot); JT series (robotics-dedicated, cumulative deliveries >100,000 units, H1 2025 +649.1% YoY).
Competitive advantages: self-developed Pandar SoC (cost moat); broadest global customer coverage (24 automakers); only profitable listed LiDAR company (financing/credibility moat); HK stock dual listing expanding capital access.
2. RoboSense (2498.HK)
Founded 2014 in Shenzhen. Listed on HKEX in January 2024 (2498.HK).
Financial data: 2024: revenue RMB 1.65 billion (+47.2%), shipments 540,000 units (+109.6%), gross margin 17.2% (+8.8 pct), net loss RMB 396 million; 2025 full-year: revenue ~RMB 1.94 billion (+17.7%), total shipments ~912,000 units, robot shipments 303,000 units (+1,141.8%); Q4 2025: first quarterly profit (net income RMB 104 million); Q1 2026: shipments 330,300 (+204.1%), robot shipments surpass ADAS for first time (+1,458.8% YoY). Target: full-year profitability in 2026.
Product portfolio: MX (sub-RMB 1,500 forward mass-production, full-stack SoC, multiple automaker designations); E1 (solid-state ultra-wide blind-spot); RS-Helios (multi-line); robotics series (E1 Robotic/RS-Bpearl covering sweeping robots/AGV/humanoid robots).
Competitive advantages: robotics segment first-mover lead; M-Platform full-stack SoC cost control; first to achieve sub-RMB 1,500 mass-production automotive-grade pricing.
3. Huawei Auto BU (Qiankun Brand, Unlisted)
Huawei's Smart Car Solution BU (Qiankun brand) does not operate as a standalone listed entity. LiDAR is embedded in ADS (Intelligent Driving) sensor suites. 2024 estimated installation: 350,000 units (26–27% share). Primary deployments: AITO M9 (1 × 192-line + 3 × solid-state), Zhijie S7, Xiangjie S9, Avatr series, BAIC Arcfox. ADS 4.0 Ultra supports L3 highway driving. Strategic moat: LiDAR deeply coupled within ADS algorithm+chip+sensor ecosystem — near-impossible to replace individual sensors without switching entire ADS package.
4. Seyond (HKEX De-SPAC listing, December 2025)
Founded 2016 in Suzhou by former Intel RealSense team. Listed on HKEX Main Board via De-SPAC in December 2025; opening-day surge >60%, market cap >HKD 20 billion; total capital raised ~HKD 1.03 billion. Core product: Falcon series 1550 nm solid-state LiDAR (detection range >300 m). 2024 deliveries: >230,000 units; NIO series cumulative deliveries >600,000 units (historical total). Updated prospectus: gross margin turned positive to 12.6%. New designations: 3 state-owned automakers + 1 joint-venture automaker. Key risk: heavy customer concentration on NIO (historically >80% of revenue); 1550 nm cost curve far slower than 905 nm.
5. Benewake (A-Share IPO in Progress)
Benewake (Beijing) Co., Ltd. specializes in short-range and robotics LiDAR. Applications: sweeping robots, AMR/AGV, drones, service robots. Delivered >100,000 single-point LiDAR units to pool-cleaning robot customers. Named in 2025 China LiDAR Industry Top 10 Innovators.
6. Wanjit Technology (300552, Roadside/V2X)
Roadside LiDAR leader; core products are roadside perception LiDAR + RSU (roadside units) + V2X systems for smart highways and city intelligent road networks. Not competing directly with Hesai/RoboSense in automotive front-install.
7. Luminar (LAZR) — Bankruptcy in December 2025
Filed Chapter 11 in December 2025. FY2025 revenue ~$67–74 million. Core lesson: in a manufacturing category with fast cost-reduction curves, an advanced technology route (1550 nm) priced far above the mainstream (905 nm) at scale will be commercially overwhelmed even if technically superior. VHS over Betamax, Android over Symbian, LFP over high-nickel NMC — the same manufacturing-economics law applies.
Chapter 7: Industrial Belts — Cluster Map and Tianxia Gongchang Perspective
1. Shanghai Jiading: LiDAR's Core Engine
Hesai's headquarters and main R&D center are in Jiading District, with its Shanghai factory handling ATX/AT128 mass-production assembly at monthly capacity of tens of thousands of units. RoboSense has important R&D teams in Shanghai. Upstream optical components, scanning elements, and drive chips cluster in Jiading and neighboring Songjiang and Minhang.
Jiading has been a traditional hub for automotive components (SAIC Group, Volkswagen China headquarters), with a mature automotive Tier 1/Tier 2 supply chain providing natural infrastructure for LiDAR front-install supply chains. factory data platforms factory data shows above-average density of LiDAR-supply-chain factories in Shanghai Jiading and surrounding areas. Valeo Automotive Shanghai is representative of foreign LiDAR localized manufacturing.
2. Suzhou: Laser Chip and Upstream Components Hub
Changguang Huaxin (Suzhou, 688048) in Xiangcheng District is China's primary VCSEL laser chip mass-production center. Changguang Huaxin Suzhou is the core vehicle for China's 905 nm self-sufficient supply. Suzhou Industrial Park's semiconductor and photonics cluster provides powerful upstream support. Zhixin Technology (Jiangsu) is active in SPAD chip development.
3. Shenzhen: RoboSense HQ + Optical Elements + Robotics LiDAR
Shenzhen is RoboSense's founding city and headquarters, with core R&D and MX product development carried out there. Shenzhen also has China's leading precision optical element manufacturing for LiDAR (lens assemblies, narrow-band filters). Shenzhen Ledong Robotics and Shenzhen Kereiwo cover consumer-robotics and industrial LiDAR sensing applications.
4. Wuhan: Laser Chips and Photonic Industry Depth
Wuhan (Optics Valley — East Lake Hi-Tech Zone) is China's traditional laser industry hub with the most comprehensive laser R&D and manufacturing cluster. Ruijing Laser (Wuhan) covers high-peak-power EEL chips; Wuhan Yuwei Optics supplies optical components.
5. Beijing: Robotaxi Ecosystem and Research
Beijing leads China's Robotaxi commercialization — Luobo Kuaipao (Baidu Apollo) in Yizhuang Economic and Technological Development Zone is one of the world's most advanced commercial Robotaxi operations. Benewake Beijing Photonics focuses on robot and short-range LiDAR. Shanghai Slamtec focuses on robot navigation LiDAR.
6. factory data platforms Data: The Manufacturing Network
factory data platforms covers 4.8 million verified, in-production factories across China. In LiDAR sensor and upstream supply chain factory data, Shanghai, Shenzhen, Suzhou, Beijing, and Wuhan form China's five major LiDAR industry nodes. Typical factory profiles: Benewake Beijing (autonomous driving + robotics LiDAR), Slamtec Shanghai (robot navigation LiDAR), Changguang Huaxin Suzhou (VCSEL production), Valeo Shanghai (foreign LiDAR localization), Wuhan Yuwei Optics (optical components), Ningbo Vasa (LiDAR sensors), Shanghai Bolei Intelligent (autonomous-driving LiDAR).
China's LiDAR industry's real foundation lies not in the reported shipment volumes of a few top-tier system integrators, but in these hundreds of component, module, and sub-system factories distributed across five industrial belts — weaving a dense collaborative manufacturing network that no newcomer can rapidly replicate through capital investment alone.
Chapter 8: Sub-Market Focus
1. L2+ NOA Passenger Vehicles: 2025's Biggest Increment
L2+ assisted driving (NOA/Navigate-on-Autopilot) is the largest single LiDAR shipment driver in 2025. Overall NOA penetration in Chinese passenger cars reached ~20% by end of 2025. The RMB 200,000–250,000 segment jumped from 2.1% (January 2024) to 24.7% (October 2024). Hesai ATX landed in BYD Han/Tang/Seal/Sea Lion, Xiaomi SU7/YU7, Li Auto L9/L8/L7 — all high-volume models.
2. L3 Commercial Readiness: Regulatory Milestone Catalyzes Premium Market
First-batch L3 approvals (December 2025) are the trigger, not the explosion. Broader impact: more automakers enter L3 approval queue in 2026–2027; single-car LiDAR count rising from ~1 (L2+) to 2–3 (L3); regulatory framework institutionally mandates LiDAR for L3+ compliance.
3. Robotaxi: Commercial-Scale Era Opens
Luobo Kuaipao (Baidu Apollo), WeRide (WRDW, Nasdaq), Pony.ai (PONY, Nasdaq). Robotaxi vehicles typically carry 3–5 LiDAR units; single-car LiDAR BOM ~RMB 20,000–100,000 — far above passenger-car front-install. If China's Robotaxi fleet reaches 300,000–500,000 vehicles by 2028 (optimistic scenario), LiDAR demand from this segment alone would be ~1.2–2.0 million units/year.
4. Commercial Vehicle Robotruck: 1550 nm's Main Battlefield
Autonomous trucks (Robotruck) require detection ranges >250 m, large-vehicle identification precision, all-weather operation — advantages that align with 1550 nm LiDAR. Seyond's partnerships with DeepWay, Yutong Tech (Yuchai), and Shaanxi Heavy Truck are based precisely on this technical fit.
5. Robotics LiDAR: Second Growth Curve Has Materialized
RoboSense 2025 full-year robot shipments: 303,000 units (+1,141.8% YoY). Q1 2026 robot shipments first surpassed ADAS business (+1,458.8% YoY). Hesai JT series H1 2025 +649.1% YoY. Four major drivers: (1) premium sweeping robots (annual shipments at millions-of-units scale, ~RMB 200–500/unit); (2) warehouse AMR/AGV scale deployment; (3) humanoid robots production ramp (Tesla Optimus, Unitree H1/G1, Zhiyuan, others); (4) service robot scenario expansion.
6. Industrial Surveying and Roadside Sensing
Industrial mapping (precision 3D modeling) serves niche professional markets. Roadside (Wanjit Technology, Huawei Smart Road) serves B2G (government) customers — smart highways, V2X. Both segments have higher per-unit ASPs but smaller total shipment volumes.
Chapter 9: Technology Evolution — Semi-Solid to Fully Solid, Sub-RMB-1,500 to FMCW
1. MEMS Micromirror: 2025 Mainstream; Mechanical Spinning Retired
Semi-solid MEMS/prism solutions dominate 2024–2025 mass-production automotive products (>95% of front-install). Mechanical spinning has effectively exited passenger-car front-install, remaining only in roadside sensing and some high-end Robotaxi configurations.
2. Flash LiDAR: Blind-Spot Applications Mature in 2025
Flash LiDAR (VCSEL array face-emitting + SPAD array receiving) is now in mature mass production for short-range blind-spot coverage (3–20 m near-field objects). Single-unit cost has fallen to RMB 500–1,000. Flash replacing MEMS primary radar remains distant — the physics of eye-safe power constraints limit Flash to <30 m effective range for now.
3. 1550 nm FMCW: Pre-2027 Large-Scale Mass Production Remains Infeasible
FMCW technology represents the highest-form next-generation sensing: simultaneous distance + velocity "4D sensing" with strong anti-interference. But manufacturing constraints are severe: laser coherence requirements (linewidth <1 MHz), immature photonic integrated circuit (PIC) processes, and signal-processing complexity far exceeding ToF keep mass-production costs at RMB 5,000–20,000/unit as of 2025. Industry consensus: FMCW high-volume front-install will not arrive before 2027–2028; initial pricing will far exceed dToF alternatives.
4. SPAD CMOS Integration: The Most Revolutionary Cost Reduction in the Receive Chain
2025's most important technical advance is large-scale SPAD integration on standard CMOS processes (28 nm/55 nm at TSMC/SMIC). This breaks the cost barrier of proprietary SPAD processes, allowing mass-production of receiver chips using the same fabs as consumer electronics. SPAD CMOS integration evolution path: Gen 1 (SPAD array + separate TIA chips, current mid-range products) → Gen 2 (SPAD + TIA monolithic integration, enabling MX/ATX sub-RMB-1,500 pricing) → Gen 3 (SPAD + TIA + DSP 3D stacked integration, expected 2026–2027, further 20–30% BOM reduction).
5. The Sub-RMB-1,500 LiDAR: The Historical Inflection of 2025
Three technology milestones simultaneously landing: MEMS SoC chipification, SPAD standard CMOS manufacturing, and self-developed SoC cost control. Together they supported RoboSense MX's ~RMB 1,500 pricing and Hesai ATX's sub-RMB-2,000 target — marking the transition from "luxury automotive sensor" to "mass-produced consumer-electronics-grade sensor," comparable to the inflection when LCD displays hit consumer-grade pricing in the mid-2000s or when camera modules hit mass-smartphone pricing in the early 2010s.
6. SoC Chipification and On-Vehicle Mass Production
LiDAR system electronics are evolving from discrete DSP+FPGA solutions to self-developed single SoCs. Self-development brings three advantages: BOM cost reduction (fewer discrete components), improved reliability (fewer solder joints), and algorithm closed-loop capability (sensor data processed in-chip with lower system latency).
Chapter 10: Risk Analysis
1. Tesla Pure-Vision Route: Limited Threat to China Market
Tesla FSD's pure-vision route has limited influence in China's market. Chinese consumers have a stronger "LiDAR = safer" preference; Xiaomi returned to LiDAR in 2025; L3 regulations mandate sensor redundancy; Chinese OEM competitive strategies center on differentiating by intelligent-driving capability. Assessment: direct substitution risk <5% market share impact in 2025–2028. Primary indirect effect: forces continued LiDAR cost reduction.
2. 4D Millimeter-Wave Substitution: Real Competitive Pressure Sub-RMB-200,000
In the cost-sensitive sub-RMB-200,000 segment, 4D mmWave vs. LiDAR competition is real — 53.94% of sub-RMB-200,000 vehicles chose 4D mmWave in January–May 2025. Resolution: continued LiDAR cost reduction toward RMB 1,000–1,200/unit by 2026–2027.
3. Price War and Gross Margin Erosion
RoboSense MX's sub-RMB-1,500 pricing is itself a price war signal. Key risk: can margins be maintained as ASPs decline? Evidence is positive: Hesai's Non-GAAP profitability in 2024 and full-year 2025 net profit guidance of RMB 450 million show that cost-reduction speed is outpacing price decline speed. RoboSense gross margin rose from ~8% to 17.2% over 2022–2024, confirming scale economics are working.
4. Overseas Financing Environment Tightening
Luminar's bankruptcy impaired global capital markets' confidence in the LiDAR sector. For companies still incurring deep losses (Seyond, Benewake), the financing window and valuation are much tighter than the 2021–2022 peak. Differentiated impact: already-profitable Hesai and near-profitable RoboSense face limited financing pressure; unprofitable companies must prove a credible profitability pathway or face narrowing capital access.
5. Automotive-Grade Certification and Production Ramp Risk
ISO 26262 functional safety and AEC-Q100 automotive-grade chip certification cycles require 18–24 months. ATX and MX have passed primary certifications. Newly developed products (for L3/FMCW next-generation routes) will face new certification timelines. Mass-production yield rate during initial production ramps remains a risk point.
6. Export Control Escalation Risk
Current US EAR has not explicitly restricted complete LiDAR systems. However, upstream 1550 nm InP epitaxial wafers and photonic integrated circuit processes are under discussion for potential control expansion. Impact is primarily on Seyond's 1550 nm supply chain. 905 nm mainstream route (Hesai/RoboSense) has relatively high domestic supply rates for upstream chips, making supply security risk more manageable.
Chapter 11: Forecasts for 2026–2030 — Sub-RMB-1,500 Universalization, Robotaxi Commercialization, Robotics Explosion
1. Shipment Forecast: From 2.5 Million to 12 Million+
| Year | China (10k units) | Global (10k units) | Passenger-Car Penetration |
|---|---|---|---|
| 2025E | ~240 | ~270 | 21% |
| 2026E | ~380 | ~430 | 27% |
| 2027E | ~550 | ~630 | 35% |
| 2028E | ~750 | ~870 | 45% |
| 2029E | ~950 | ~1,100 | 52% |
| 2030E | ~1,200+ | ~1,400+ | 60%+ |
5-year CAGR (2025–2030E): China ~38%, Global ~40% (including robotics + roadside + Robotaxi)
2. Market Size Forecast: RMB Trillion by 2030
| Year | China Market (RMB 100M) | Global Market ($100M) |
|---|---|---|
| 2025E | ~241 | ~50 |
| 2026E | ~432 | ~90 |
| 2027E | ~550 | ~115 |
| 2028E | ~650 | ~140 |
| 2030E | ~1,000+ | ~200+ |
3. Sub-RMB-1,500 Universalization: 2026–2027, Penetrating RMB 100,000–150,000 Vehicles
As scale economies and SPAD 3D stacking integration come on stream, primary forward automotive LiDAR ASP is expected to fall further to RMB 1,000–1,200 by 2026–2027. This makes LiDAR economically viable as standard equipment in vehicles priced at RMB 120,000–150,000 — opening a new market segment that represents ~25–30% of total Chinese passenger-vehicle sales.
4. L3 Scale Commercialization: 2026–2028 Gradual Rollout
2025 first-batch L3 approval is the starting point. 2026–2027: more automakers enter L3 approval queue; per-vehicle LiDAR count rising from ~1 (L2+) to 2–3 (L3); if 2027 L3 annual sales reach 1 million vehicles (single car 2.5 units average), L3 alone contributes ~2.5 million units additional annual demand.
5. Robotaxi Scale: 2026–2028
Luobo Kuaipao, WeRide, Pony.ai commercialization enters acceleration phase. If domestic Robotaxi fleet reaches 300,000–500,000 vehicles by 2028 (optimistic scenario), at ~4 units/vehicle, annual demand ~1.2–2.0 million units — a significant high-end LiDAR demand segment. Seyond and Hesai are the primary beneficiaries through Robotaxi designations.
6. Robotics LiDAR: Catches Up With Automotive by 2028
RoboSense Q1 2026 robot shipments already exceed ADAS business — the signal flare for acceleration. Four growth drivers: premium sweeping robot full-line standardization (millions of units/year); warehouse AMR/AGV scale deployment; humanoid robot production ramp (Tesla Optimus/Unitree/Zhiyuan); service robot multi-scenario expansion. By 2028, robot LiDAR may reach 3–5 million units/year, rivaling or exceeding automotive front-install.
7. Competitive Landscape: China Four-Vendor Dominance Deepens
2026–2030: China's four-vendor market share is expected to rise from ~88% to 92–95% as smaller overseas companies are further marginalized. Within the four: Hesai leads in mid-mainstream passenger vehicles through ATX + JT robotics series; RoboSense targets 2026 full-year profitability, with robot LiDAR business potentially becoming the largest revenue source by 2027; Huawei deepens ADS ecosystem + L3 policy tailwinds; Seyond scales Robotaxi/commercial vehicle customer base to reduce NIO concentration risk.
Chapter 12: Conclusion — After the Sub-RMB-1,500 Inflection, Chinese LiDAR Enters the Foundation of Global Autonomous Driving
In 2025, China's LiDAR industry completed three historically significant milestones:
First: Shipment volume dominance; industry structure locked. Global automotive LiDAR shipments surpassed 2.5 million units; Chinese manufacturers captured >80%; the four-vendor bloc holds 88%. Luminar's bankruptcy — despite technically praised 1550 nm technology — declared the systemic commercial failure of the Western LiDAR camp. Technology credentials cannot substitute for scalable mass-production commercial capability; this is Luminar's deepest lesson.
Second: Sub-RMB-1,500 mass production lands; democratization begins. RoboSense MX (~RMB 1,500) broke the historical assumption that "LiDAR cannot enter the sub-RMB-2,000 range." This is not a simple price cut — it is the result of chipification technology (MEMS SoC + standard-CMOS SPAD + self-developed SoC) completing its transition from lab to production line. China's manufacturing ecosystem, replete with consumer-electronics and new-energy-vehicle scale efficiencies, once again demonstrated its ability to democratize a category of high-end sensors — as it did for LCD displays, camera modules, and power batteries before.
Third: Robot LiDAR second-growth-curve materializes as current revenue. RoboSense's 303,000-unit robot shipments in 2025 and Q1 2026 robot shipments exceeding ADAS business confirm that "robot LiDAR" is no longer a future expectation but a measurable current output — visible in quarterly financial reports.
Looking ahead to 2026–2030, four mutually reinforcing drivers will collectively push the industry toward 12 million+ units/year and a trillion-RMB market: passenger-vehicle penetration climbing from 21% to 60%+ (sub-RMB-1,500 pricing opens the RMB 120,000–150,000 market; L3 regulations catalyze premium standard equipment; NOA popularization drives mid-range adoption); robotics LiDAR from millions to tens of millions (humanoid robot production ramp + premium sweeping-robot standardization + warehouse AMR scale-up); Robotaxi/L4 per-vehicle multi-unit leverage (2028: 300,000–500,000 fleet scale, ~4 units/vehicle, ~1.2–2.0 million units/year); commercial Robotruck high-speed highway scale (1550 nm/long-range LiDAR's primary battlefield).
Tianxia Gongchang Industry Research Institute, in long-term tracking of China's manufacturing pathways, consistently observes the same pattern: when a product category migrates from high-end niche to mass-market standard, the decisive factor is never which technology is most advanced, but which players first achieve scalable mass production, deep customer ecosystem binding, and chipified cost control. Chinese LiDAR companies have systematically built advantages on all three dimensions — not since 2025, but from the product accumulation of 2018–2020 and the mass-production validation of 2021–2023 that preceded it.
factory data platforms's 4.8 million verified, in-production factories are the substrate of China's manufacturing ecosystem — among them, the factories directly tied to the LiDAR supply chain, from 905 nm laser chip plants in Suzhou and Shenzhen to optical elements plants in Hangzhou and Suzhou, from system-integration facilities in Shanghai Jiading to robot application factories in Beijing Yizhuang, form a unique industrial collaboration network. No other country currently possesses a comparably competitive factory network across all these tiers simultaneously — this is the ultimate foundation for China's LiDAR industry maintaining global leadership through the next decade.
The sub-RMB-1,500 inflection has landed. Over the next five years, this "eye of autonomous driving" will complete its final journey from flagship-exclusive to universal standard — at which point every new vehicle driving on Chinese roads may be equipped with this sensing node manufactured in Jiading, Shenzhen, or Suzhou.
Data Sources and Principal References
This report was compiled and analyzed by Tianxia Gongchang Industry Research Institute based on factory and industrial chain data from the Tianxia Gongchang platform (www.tianxiagongchang.com), supplemented by public information, official disclosures, and reports from authoritative media. Principal data sources include:
- factory data platforms platform Chinese factory database and industrial belt data (www.tianxiagongchang.com)
- Hesai Technology (HSAI.O/HSAI.HK) 2024 Annual Report, 2025 Q1/Q2/Q3 earnings releases, and 2025 performance guidance announcements
- RoboSense (2498.HK) 2024 Annual Report, 2025 full-year results announcement, and Q1 2026 financial report
- Seyond Hong Kong IPO prospectus (filed February 2025 and updated version) and listing announcement
- Luminar Technologies (LAZR.O) Q2 2025 earnings release and Chapter 11 bankruptcy filing announcement
- Gasgoo Research Institute: "2024 January–November China LiDAR Installation Volume Statistics"
- Qianzhan Research Institute: "2025 China LiDAR Industry Panoramic Atlas"
- Zhiyan Consulting: "2025 China Automotive LiDAR Industry Market Size and Competitive Landscape"
- MIIT: "Regulations on Admission and Road-Testing Pilot Management for Intelligent Connected Vehicles" (2023)
- MIIT First-Batch L3 Conditional Autonomous Driving Vehicle Admission Announcement (December 2025)
- Qbitai, 36kr, Sina Tech, Every Daily Economic News, and other authoritative media enterprise interviews and industry reports
- OFweek Laser, Gasgoo, IT Home, and other sector media for real-time shipment and market data
- SEC/HKEX public disclosure documents from Luminar, Ouster, Innoviz, and other overseas companies
- CITIC Securities, Huatai Securities, China International Capital Corporation (CICC), and other investment bank LiDAR industry research reports