Abstract
In the modern industrial ecosystem, one type of chip is embedded in almost everything that needs to "see" — smartphone cameras, automotive front-view cameras, factory vision inspection benches, hospital endoscopes, and the security cameras at every door. It is called the CMOS Image Sensor (CIS). Not the most expensive chip, nor the most complex, but among the most widely deployed visual perception chips in use today. Global annual shipments reach approximately 70 billion units, with a market value of around USD 23.7 billion in 2024.
This market has long been dominated by a single Japanese company. Sony Semiconductor Solutions commands approximately 52% of global market share, erecting a barrier in the image sensor space that no rival has been able to breach — virtually every flagship smartphone priced above the equivalent of USD 700, and every premium imaging chip behind it, uses a Sony sensor. Samsung Electronics holds approximately 19% share in second place, while OmniVision — U.S.-founded and now a subsidiary of China-listed Will Semiconductor (603501) — holds approximately 11% in third. The combined share of these three leaders (CR3) exceeds 82%, making CIS one of the most concentrated sub-sectors in semiconductors.
China is the world's largest CIS-consuming market. Smartphones are the primary downstream, with China accounting for more than 70% of the global handset supply chain's value. Automotive electrification and intelligence, which drives camera-count explosions in ADAS, combined with the world's densest deployment of security infrastructure, continuously add to domestic CIS demand. Yet this giant consumer nation's domestic supply rate — measuring what fraction of CIS consumed in China is designed by mainland Chinese companies — stood at only around 30% as of 2024. Within that 30%, the main battleground is mid-to-low-end handsets, security cameras, and parts of automotive cameras; the 1-inch large-format sensors of flagship phones and the high-end automotive CIS for L2+/L3 ADAS remain inaccessible to domestic brands.
This report uses 2026 as its observational vantage point to systematically map the global and Chinese CIS industry: market scale, competitive landscape, supply-chain structure, segment markets, technology evolution, risks, and the outlook to 2030.
Key Findings at a Glance:
- Global CIS: ~USD 23.7 billion in 2024, over USD 30 billion by 2030; automotive ADAS is the fastest-growing segment (CAGR ~9.4%).
- CR3 exceeds 82%. Sony's ~52% share cannot be dislodged in the near term; SK Hynix's exit creates space for Chinese players; onsemi faces competitive pressure in automotive.
- China's Big Three combined global share ~23% (2024): Will Semiconductor FY2024 revenue RMB 25.7 billion, net profit RMB 3.3 billion (+498%); SmartSens revenue RMB 6.0 billion (+109%, net profit +2,663%); Galaxycore revenue RMB 6.4 billion (+36%).
- The 30% localization ceiling is a function not only of technology gaps but of brand trust and supply-chain switching costs.
- Upgrade path: security (base consolidation) → automotive ADAS (highest potential, top priority) → mid-to-high-end handsets (long-term goal, limited breakthrough before 2028).
- Technology frontier: hybrid bonding 3-layer stacking is the core process battleground for 2025–2027; global shutter will fully permeate automotive by 2026–2028; SPAD LiDAR is the next-decade strategic variable; AI+ISP on-chip integration is the near-term roadmap.
I. The Revolution of Light: CIS Definition, Classifications, and Supply Chain Overview
1.1 From Photons to Pixels — How CIS Works
In the world of cameras, light is a particle flow that needs to be "counted."
When light passes through a lens and strikes the silicon surface of an image sensor, every pixel in the sensor does the same thing: uses the photoelectric effect to convert photons into electrons, and then quantifies that pile of electrons into a numerical value — brightness. Millions of pixels completing this conversion simultaneously produce one frame of image. This process is the foundation of all digital imaging.
Image sensors come in two main technology families: CCD (Charge-Coupled Device) and CMOS (Complementary Metal-Oxide-Semiconductor). Both convert light into electrical signals, but differ in readout method and manufacturing process.
CCD readout is like a relay race: each row of pixels passes its charge to the next row, which is ultimately converted to a voltage signal by amplifiers at the edge. The design delivers excellent image quality and signal-to-noise ratios, but requires dedicated fabs, consumes significant power, reads slowly, and costs more. CCD dominated during the early era of film-to-digital conversion, particularly in professional cameras and broadcast equipment.
CMOS readout is "every pixel handles its own accounting": each pixel has its own built-in amplifier and analog-to-digital converter, eliminating the need to pass charges, and outputting digital signals directly. This makes CMOS sensors faster, lower power, compatible with standard semiconductor processes, and dramatically cheaper to manufacture. The cost was that early CMOS had worse signal-to-noise ratios than CCD, with noticeable image quality differences in some scenarios.
The turning point came in the mid-to-late 2000s. With breakthroughs in stacking processes, back-illuminated (BSI) structures, and on-chip noise suppression algorithms, CMOS's image quality weaknesses were addressed one by one, while its cost, integration, and power advantages grew ever more prominent. Around 2008, CMOS comprehensively surpassed CCD in most applications. Today, except for niche scientific and astronomical applications, CCD has essentially disappeared from consumer markets — smartphones, automobiles, security cameras, action cameras, and drone cameras almost universally use CMOS Image Sensors.
1.2 BSI, Stacking, Global Shutter: Three Levels of Technology Advancement
Understanding the competitive landscape of CIS requires understanding these three technology tiers, which roughly correspond to the three most critical technical leaps in the CIS industry over the past twenty years.
First Tier: Back-Side Illumination (BSI)
Traditional front-side illuminated (FSI) sensors have the metal interconnect layer on the pixel surface — wiring, transistors — which blocks some incoming light, reducing the "fill factor," meaning the ratio of actual light-sensitive area to total pixel area.
BSI's solution is "flip the entire wafer": after fabricating the circuits, face the circuit side downward and let light enter the photodiodes directly from the back side (where there is no wiring). Wiring no longer blocks light; fill factor approaches 100%; low-light sensitivity improves dramatically. Sony was first to mass-produce BSI sensors in 2009 under the Exmor R brand. Samsung and OmniVision quickly followed; by the mid-2010s BSI became standard in smartphone CIS. By 2024, BSI structures accounted for approximately 65% of global CIS market share.
Second Tier: Stacked CMOS
BSI solved the light-collection problem but not the readout speed or area efficiency problem. Stacked CMOS places two separate wafers together: one dedicated to the pixel array (using optimized imaging process nodes), another dedicated to logic circuits (A/D conversion, ISP unit, memory buffers, using the most advanced logic process), connected via Through Silicon Vias (TSV). Readout speed improves 3–10× compared to non-stacked BSI, enabling 4K/8K high-frame-rate video, rapid burst shooting, and video stabilization.
Sony introduced stacked BSI CIS (Exmor RS) in 2012; Samsung followed in 2013; OmniVision came later. In 2023, Sony debuted a three-layer stacked sensor in the Xperia 1 V. Stacked CIS now accounts for the vast majority of flagship smartphone sensors.
Third Tier: Global Shutter (GS) and SPAD
Rolling shutter is the default mode in most CMOS sensors today — the pixel array exposes row by row, not simultaneously. For ADAS cameras, this is unacceptable: a fast-moving vehicle or pedestrian captured with rolling shutter can appear distorted, potentially misleading driver assistance systems.
Global shutter lets all pixels expose simultaneously, completely eliminating rolling-shutter distortion — a core technical requirement for automotive ADAS cameras. onsemi's Hyperlux series is the representative automotive global-shutter CIS product.
SPAD (Single-Photon Avalanche Diode) is another technology path: instead of sensing light intensity, it senses the moment of each photon's arrival; combined with laser pulses, it measures distance — this is the core of dToF (direct Time-of-Flight) LiDAR sensors. Sony's IMX459/IMX516 SPAD sensors are used in smartphones' 3D sensing capabilities and will be strategically important for automotive solid-state LiDAR.
1.3 Pixel Size and Application Segment Classification
Pixel size is another critical CIS performance dimension, typically measured in μm (micrometers). Larger pixels capture more photons per unit area, delivering better signal-to-noise ratios; smaller pixels enable higher resolution but degrade low-light performance.
Application segments include:
- Smartphone CIS: Largest volume (~60–65% of CIS market); fastest technology iteration; high-end dominated by Sony/Samsung, mid-to-low-end is the domestic battleground.
- Automotive CIS: Fastest growing (CAGR ~9.4%); AEC-Q100 automotive qualification required; global shutter, high dynamic range (≥120 dB HDR), and wide temperature operation are core requirements; each L2+ vehicle needs 4–8 cameras, L3+ needs 8–12.
- Security CIS: High volume, lower ASP; resolutions upgrading from 2MP to 4MP/8MP; AI camera replacement cycle is the core driver; Hikvision and Dahua are the main buyers.
- Industrial Vision CIS: High precision, high frame rate, global shutter universal; OmniVision and Sony have dedicated product lines.
- Medical Endoscopy CIS: Extremely small pixels (for capsule cameras); high sensitivity; regulatory certification barriers are extremely high.
- Consumer Electronics (tablets/PCs/VR/AR): ToF sensors, ultra-compact modules; steady growth.
1.4 Supply Chain Overview: From Silicon Wafer to Camera Module
The CIS supply chain runs from upstream silicon wafers through design, manufacturing (wafer foundry), packaging and testing, to module integration, and finally into end devices.
Upstream: Advanced BSI stacked CIS manufacturing requires special silicon wafers (bonded SOI wafers) and advanced bonding processes (hybrid bonding). This core process is controlled by TSMC (advanced-node BSI stacking), Sony's proprietary fabs (Nagasaki/Kumamoto), and Samsung's proprietary fabs. Domestically, SMIC and Hua Hong handle mid-to-low-end CIS foundry work.
Midstream Design: Fabless is the main model for Chinese CIS: Will Semiconductor (OmniVision), Galaxycore, and SmartSens are all Fabless, commissioning foundries to manufacture. Sony and Samsung are IDMs (Integrated Device Manufacturers), designing and producing themselves.
Midstream Packaging: CIS packaging primarily uses CSP and WLCSP. Domestic WLCSP specialist Jcet (WLCSP, 603005) is the leading CIS packaging service provider. Changdian Technology and Tongfu Microelectronics also participate.
Downstream: Lens suppliers (Sunny Optical, Largan, Genius), module suppliers (O-Film, Q Technology, Sunny module division) and end device manufacturers form the direct downstream.
1.5 CIS and LiDAR: A Converging Technology Path
In the wave of automotive intelligence, cameras (CIS) and LiDAR are the two core sensors in perception systems. SPAD technology is beginning to create a technical intersection between CIS and LiDAR. Sony's SPAD CIS portfolio positions it not only to improve smartphone ToF ranging but also to enter the automotive LiDAR market. Lidar companies like Robosense (2498 HK) and Hesai (HSAI) currently rely largely on imported SPAD sensors (mainly from Sony); domestic SPAD CIS design and manufacturing capability represents the next major frontier in CIS localization.
1.6 CIS Economic Properties: The Overlay of Growth and Cyclicality
The CIS industry is structurally a growth sector (long-term growth rate ~5–7%), but at the micro level it displays significant cyclicality. These two coexist, requiring industry participants to "think long, act short."
Growth comes from three structural drivers: (1) rising camera density per device (handsets moving from single to five cameras, cars from one rearview to 12 environmental cameras); (2) structural ASP improvement (high-end 1-inch sensors, automotive global-shutter, AI security starlight sensors drive average price upward); (3) continuously emerging new application scenarios (ToF depth sensing, in-cabin driver monitoring, endoscopy micro-CIS).
Cyclicality comes from supply-demand mismatches: wafer foundry capacity additions take 18–24 months, making the supply side unable to respond quickly to demand fluctuations. China's three major CIS players saw combined profit collapse from approximately RMB 8 billion in 2021 to approximately RMB 600 million in 2023, then rebound to approximately RMB 4 billion in 2024 — a dramatic illustration of this cycle.
II. Global Competitive Landscape: The Sony Empire and Other Players
2.1 CR3 Exceeding 82%: An Extremely Concentrated Market
In the global semiconductor industry, CIS is one of the few sub-sectors where market concentration is comparable to memory chips. In 2024, Sony Semiconductor Solutions, Samsung Electronics' System LSI division, and OmniVision collectively captured more than 82% of global market share.
This level of concentration is the product of capital and technology intensity: BSI stacking design iteration requires hundreds of specialized R&D engineers and more than a decade of process accumulation; on the capital side, Sony alone operates multiple wafer fabs in Kyushu (Fab 4, Fab 5, etc.), each expansion costing tens of billions of dollars.
The highly concentrated landscape is not static. SK Hynix actively withdrew from CIS to concentrate capital on HBM (High Bandwidth Memory) for AI demand. This retreat vacated portions of the mid-tier CIS market, objectively enabling share gains by SmartSens and Galaxycore. onsemi focuses strategically on automotive global-shutter CIS rather than competing directly with Sony and Samsung in the handset market.
2.2 Sony Semiconductor Solutions: The Unassailable Number One
Sony entered the image sensor field in the 1980s. In 2009, it pioneered BSI sensor mass production (Exmor R brand), establishing technical leadership; in 2012, it introduced stacked BSI CIS (Exmor RS brand), widening the gap further. This timeline of leading at every technology transition has cemented Sony's near-monopoly in the premium CIS market.
Sony's global CIS share is approximately 52%, even higher in premium flagship smartphones — Huawei, Samsung (external sales), Apple, and Xiaomi flagships have historically used Sony sensors overwhelmingly. Sony's consumer-grade CIS brand LYTIA (launched 2023) has become a "quality certification badge" for flagship phone cameras in China, with "equipped with Sony LYTIA flagship sensor" as a standard marketing phrase at product launches.
Sony's manufacturing system is another moat. Sony operates dedicated CIS wafer fabs in Nagasaki, Kumamoto, and Oita, Japan. Fab 5, Sony's newest large-scale CIS production facility, is dedicated to stacked BSI production and reportedly involves total investment exceeding ¥500 billion (approximately USD 3.5 billion). This IDM model gives Sony complete autonomy over key processes.
In automotive CIS, Sony is also actively attacking, with automotive CIS revenue reportedly recording high double-digit growth in FY2024–2025. Sony's I&SS segment achieved approximately JPY 14.7 trillion (approximately USD 9.8 billion) in FY2024, with operating profit margins typically 15–20%, making it Sony Group's most important profit source.
2.3 Samsung System LSI (ISOCELL): The Second Place Dilemma
Samsung occupies a unique position: as one of the world's largest phone manufacturers, it is both the largest CIS buyer (self-use in Galaxy series) and Sony's most direct competitor. This "athlete and judge" dual role is a technological advantage but a commercial constraint — third-party phone brands worry about competitive relationships and are reluctant to source heavily from Samsung.
Samsung's ISOCELL brand was established in 2013. Its 200MP ISOCELL HP9 represents the 2024 flagship direction. Samsung's CIS share is approximately 19%, primarily in its own Galaxy series and a few Chinese premium brands. In automotive CIS, Samsung is rapidly expanding its ISOCELL automotive line.
2.4 OmniVision: The Third-Place Global Expansion
OmniVision Technologies was founded in 1995 in Silicon Valley. It was taken private in 2016 by a Chinese consortium at approximately USD 1.9 billion, then acquired by Will Semiconductor (603501) for approximately RMB 8.6 billion in 2019, becoming the core of Will's CIS business. OmniVision's global share is approximately 11%, making it the world's third-largest CIS design company.
OmniVision's competitive advantage is the breadth of its product portfolio across handsets, automotive, security, industrial, and medical applications. In automotive CIS, OX08B (8MP, global shutter, AEC-Q100 certified) has entered global Tier 1 supplier chains (Bosch, Aptiv, Magna). Will Semiconductor 2024 CIS revenue was approximately RMB 19.19 billion, with automotive CIS at approximately RMB 5.9 billion (+29.85%).
Will Semiconductor announced plans in 2025 to rename itself "OmniVision Group," strategically focusing on the CIS track.
2.5 SK Hynix: Strategic Retreat from CIS to HBM
SK Hynix was formerly a top-5 CIS supplier but made a strategic decision to reallocate capital from CIS to HBM for AI demand. This retreat objectively cleared market space for domestic competitors, particularly benefiting SmartSens's rapid share gains in security CIS.
2.6 onsemi: The Automotive Global Shutter Expert
onsemi's global CIS share is approximately 3.6% (2025), but in the automotive ADAS global-shutter sensor niche, its historical position was higher. The Hyperlux AR0823AT (4.86 μm pixels, 8MP, global shutter, ASIL-C certified) is the benchmark automotive ADAS sensor. As Sony, Samsung, and OmniVision increase automotive CIS investment, onsemi faces mounting competitive pressure. onsemi's response is to focus on ASIL-D level L3/L4 autonomous driving sensors.
2.7 Global CIS Competitive Landscape Summary (2024)
| Company | Global Share | Positioning | Headquarters |
|---|---|---|---|
| Sony Semiconductor | ~52% | Premium handset + automotive + industrial | Japan |
| Samsung ISOCELL | ~19% | Handsets (primarily self-use) + automotive | South Korea |
| OmniVision (Will Subsidiary) | ~11% | Mid-range handset + automotive + security + industrial | US/China |
| Galaxycore | ~8% | Mid-to-low-end handset + display drivers | China (Shanghai) |
| SmartSens | ~4% | Security + handset + automotive | China (Shanghai) |
| SK Hynix | <3% (exiting) | Shifting to HBM | South Korea |
| onsemi | ~3.6% | Automotive global shutter | US |
| ST | ~1–2% | Industrial + embedded | Europe |
III. PEST Analysis: Drivers and External Pressures
3.1 Political Environment: The Dual Chessboard of Semiconductor Self-Sufficiency and Export Controls
China's "14th Five-Year Plan" listed integrated circuits as one of seven frontier technologies. The National IC Industry Investment Fund (Big Fund) Phases I and II raised over RMB 340 billion combined, with significant amounts flowing into CIS design companies, foundries, and packaging. Science and Technology Innovation Board (STAR Market) listings gave domestic CIS companies favorable valuations and listing conditions.
The U.S. Export Administration Regulations (EAR) have progressively tightened, with indirect transmission channels to CIS: TSMC's advanced-node service to specific Chinese entities is restricted; EDA tool export controls could affect advanced CIS design; key manufacturing equipment (AMAT CVD/ALD, Lam Research etch) is on the control list. Japan followed with its own 2023 export controls on advanced semiconductor manufacturing equipment.
3.2 Economic Environment: Handset Cycle Recovery and Automotive New Volumes
Global smartphone shipments in 2024 reached approximately 1.23 billion units (+5–6%), recovering modestly. The AI smartphone concept drove premium upgrade demand. China's NEV sales in 2024 reached approximately 9.5 million units (+35%), with 40% EV penetration rate — premium configurations on EVs directly drove automotive CIS demand. Security market AI camera replacement created a structural upgrade demand cycle.
3.3 Social Environment: The Imaging Arms Race and Security Infrastructure
Cameras have become the most important smartphone selling point in China's premium market. DxOMark evaluation has become a quasi-authoritative third-party certification, maintaining strong consumer focus on camera performance. Security cameras are now social infrastructure, with China having the world's highest deployment density and ongoing AI upgrade wave driving replacements.
3.4 Technology Environment: Stacking Advances and AI Vision Fusion
Hybrid bonding represents the next-generation stacking core process. Global shutter popularization will make ADAS cameras standard. AI and CIS fusion through in-sensor computing begins commercializing. Quantum dot and organic photoelectric sensor technologies represent 2030+ remote prospects.
IV. China Market Scale and Localization Rate
4.1 China: The World's Largest CIS Consumer
China is the world's largest CIS consuming market, accounting for approximately 30–31% of global CIS value (~USD 7.3 billion or ~RMB 52.7 billion in 2024), supported by: (1) China accounting for 70%+ of global smartphone supply chain value; (2) Hikvision and Dahua together holding over 40% of global security camera market share with predominantly domestic CIS procurement; (3) China being the largest NEV market with the highest electrification rate globally.
4.2 Localization Rate: The 30% Reality
China's domestic CIS localization rate is approximately 30% (2024), concentrated in low-end handsets, security cameras, and parts of automotive cameras. The 30% is structurally distributed:
- Low-end handsets (≤8MP): >70% domestic
- Mid-range handsets (13–50MP): ~40–50% domestic
- Flagship handsets (≥50MP, 1-inch sensors): <5% domestic
- Security CIS: ~60% domestic
- Automotive CIS: ~20–30% domestic
- Industrial/medical: ~20% domestic
4.3 Path from 30% to 50%
The three main incremental sources are: (1) Automotive CIS domestic substitution — domestic NEV brands have strong motivation for supply chain localization; (2) Security AI camera replacement — SmartSens and OmniVision benefit from upgrade cycle; (3) Mid-range handset penetration — Galaxycore's 50MP mass production creates opportunity.
4.4 The Manufacturing Dependency Hard Constraint
The current 30% localization rate is fundamentally a design-level metric, not a manufacturing one. High-end CIS logic layers still rely on TSMC's advanced nodes. Genuine "full-chain domestic" manufacturing of high-end CIS remains unachievable before 2028–2030.
V. Supply Chain Breakdown: From Silicon Wafer to Camera Module
5.1 Wafer Foundry
TSMC is the primary advanced CIS foundry. OmniVision's high-end products (OV50X series) use TSMC N7/N5 nodes. Sony's proprietary fabs represent IDM mode's extreme — complete self-control of key processes. SMIC is the key domestic foundry partner for mid-to-low-end CIS, primarily serving Galaxycore with 28nm/40nm processes. Hua Hong Semiconductor handles portions of domestic mid-to-low-end CIS foundry work.
5.2 CIS Design: The Three Domestic Fabless Players
| Dimension | Will/OmniVision | Galaxycore | SmartSens |
|---|---|---|---|
| Main Markets | Mid-high handset + automotive + security + industrial medical | Mid-low handset (primary) + display drivers | Security (base) + handset + automotive (growth) |
| Technical Advantage | Full portfolio + vehicle-grade certified + medical | Cost competitiveness + low-end scale | Low-light/HDR + security customization |
| Global Share (2024) | ~11% | ~8% | ~4% |
| Foundry Partner | TSMC (high-end) + UMC | SMIC (mid-low-end) | TSMC + SMIC |
5.3 Packaging & Testing: Jcet's WLCSP Specialization
WLCSP (Wafer Level Chip Scale Package) is the mainstream CIS packaging technology. Jcet (WLCSP, 603005) in Suzhou Industrial Park is the most important domestic CIS WLCSP packaging service provider. Its core capabilities include TSV interconnect processes and glass-base wafer-level packaging (G-WLP). In 2024, Jcet achieved revenue of RMB 1.13 billion (+23.7%), gross margin 43.28%, with automotive CIS packaging as the fastest-growing segment.
5.4 Lenses: Sunny Leads, Largan Guards the High End
Sunny Optical (2382 HK) is China's largest handset lens and camera module supplier, with 2024 revenue ~RMB 38.3 billion (+20.9%), smartphone lens global number one, and automotive lens shipments exceeding 100 million units for the first time. Largan Precision (3008 TW) dominates Apple iPhone and Samsung Galaxy flagship main camera lenses with gross margins exceeding 55%.
5.5 Modules: O-Film and Q Technology's Scale Competition
The global smartphone camera module market saw approximately 4.4 billion units shipped in 2024, with the top three (O-Film, Q Technology, Sunny module division) accounting for over 60% combined. Module companies occupy the "mid-stream pressure" position — squeezed by both upstream CIS/lens pricing and downstream brand customers' price pressure.
5.6 Supply Chain Competitive Advantage Assessment
The value chain is highly unequal: Sony (40–50% gross margin), TSMC (55–60%), Largan (55–65%), and Jcet (~43%) occupy the top; Chinese CIS design companies (net margins 1–13%) are in the middle; module companies (net margins 1–4%) are at the bottom. This structure reflects the Fabless model's structural limitations: with 40–60% of revenue flowing to foundries, design companies face a gross margin ceiling of approximately 30–40%.
VI. Key Company Deep Dives
6.1 Will Semiconductor (603501): From Distributor to Global CIS Third Place
Will Semiconductor was founded in 2007 as a semiconductor component distributor. The 2019 acquisition of OmniVision for approximately RMB 8.6 billion transformed it into the world's third-largest CIS design company in one step.
2024 Financial Performance: Total revenue RMB 25.731 billion (+22.41%), record high; net profit attributable to parent RMB 3.323 billion (+498.11%); CIS (image sensor) revenue RMB 19.19 billion (74.76% of revenue), of which automotive CIS approximately RMB 5.905 billion (+29.85%); R&D investment approximately RMB 3.245 billion (15% of CIS revenue).
OmniVision's OX series automotive products (OX08B, OX09B) have entered global Tier 1 supply chains (Bosch, Aptiv, Magna). The planned rename to "OmniVision Group" signals strategic focus on the global CIS brand.
6.2 Galaxycore (688728): Scale King of Mid-to-Low End
Galaxycore was founded in 2003 in Shanghai, pioneering domestic CIS design. Its competitive logic is "extreme cost + SMIC foundry": deep partnership with SMIC since 2008 built cost competitiveness unmatched by international rivals.
2024 Financial Performance: Revenue RMB 6.383 billion (+35.9%); net profit RMB 187 million (+287.2%); smartphone CIS revenue RMB 3.598 billion (+60.44%); critically, 50MP CIS achieved mass production in 2024 — the key milestone in penetrating mid-tier handset main cameras.
6.3 SmartSens (688213): From Security Breakthrough to Three-Track Expansion
SmartSens Technology was founded in 2017 in Shanghai, listing on STAR Market in 2023. The founding team came from Sony Semiconductor and OmniVision, bringing deep CIS architectural design experience.
2024 Financial Performance: Revenue RMB 5.968 billion (+108.87%); net profit RMB 393 million (+2,662.76%); smartphone CIS revenue RMB 3.291 billion (+269%); security CIS revenue RMB 2.150 billion (+28.6%); automotive CIS revenue RMB 527 million (+79.1%).
SmartSens's super-starlight (0.001 lux) and HDR technologies in security cameras are core competitive differentiators, directly translating to automotive ADAS requirements.
6.4 Jcet (603005): The Hidden Champion of CIS Packaging
Jcet (Suzhou WLCSP) plays an indispensable role in the CIS supply chain. Founded in 2001 in Suzhou Industrial Park, it specializes in WLCSP packaging for image sensor chips.
2024 Financial Performance: Revenue RMB 1.13 billion (+23.7%); parent-attributable net profit RMB 253 million (+68.4%); gross margin 43.28% (+5.13pp) — far above typical packaging companies (15–25%). Automotive CIS packaging drove the high gross margin improvement.
6.5 VeriSilicon (688521): CIS Value Chain IP Infrastructure
VeriSilicon (688521) is a semiconductor IP licensing and custom chip design services company. Its Image Signal Processor (ISP) IP cores are used by CIS designers and terminal SoC companies. With on-chip AI inference advancing, ISP IP evolution toward "ISP + NPU fusion" is a strategic opportunity for VeriSilicon.
6.6 Overseas Companies' China Strategy
Sony continues to deepen its China relationships with all major handset brands and has accelerated supply to Chinese NEV brands (NIO, Li Auto, AITO). Samsung's CIS in China mainly serves non-flagship handset models. onsemi's China focus has shifted to automotive electronics, supplying automotive CIS and SiC power devices to BYD, Geely, SAIC, and GAC.
VII. China CIS Industrial Clusters: Shanghai at Core, Suzhou, Beijing, and Shenzhen as Satellites
7.1 Shanghai: The Heart of China's CIS Design Industry
Galaxycore, SmartSens, and Will Semiconductor — all three of China's most important CIS design companies are headquartered in Shanghai. This is no coincidence. Shanghai's semiconductor industry has deep historical accumulation: SMIC's establishment, abundant overseas Chinese semiconductor engineer returnees, and mature venture capital ecosystems created China's unparalleled semiconductor innovation soil.
Galaxycore's Shanghai roots: Founded in Minhang and Pudong by Dr. Zhao Lixin (Stanford PhD), leveraging SMIC's 12-inch foundry capability to build a 30–50% cost advantage over imports.
SmartSens's Zhangjiang High-Tech Park: Zhangjiang concentrates over 400 semiconductor companies, providing complete resources for CIS startups. The founding team's Sony/OmniVision background enabled rapid progress.
Will Semiconductor's Shanghai headquarters: A listed company body in Shanghai, with domestic R&D centered on automotive CIS, security CIS, and mid-to-high-end handset CIS.
7.2 Suzhou: Packaging and Testing Hub
Suzhou Industrial Park (SIP) is China's core packaging and testing cluster. Jcet's location here benefits from: proximity to Shanghai (20-minute high-speed train), efficient SIP management, and the legacy of Taiwanese packaging companies (ASE Suzhou branch) that trained a deep talent base.
7.3 Beijing: Technical Rear of Universities, Research Institutes, and IP Design
Tsinghua University's Electronic Engineering Department, Peking University, and the CAS Institute of Microelectronics long-term conduct CIS fundamental research, supplying high-end R&D talent. VeriSilicon's Beijing R&D center and multiple SPAD and 3D sensing startups cluster around Zhongguancun Science City.
7.4 Shenzhen: Supply Chain Terminal and Market Linkage Hub
Huawei, OPPO (Dongguan, adjacent), and vivo (Dongguan) headquarters are in the Shenzhen-Dongguan area — the most important flagship phone customer cluster. O-Film's Shenzhen factory, Hikvision's Shenzhen R&D center, and vast numbers of FPC/motor/structural component factories make Shenzhen the "market end" and final integration hub.
7.5 factory data platforms Data Perspective
factory data platforms's data on 4.8 million active factories shows that CIS-related optical module processing, precision manufacturing, and FPC factories are concentrated in Guangdong (Shenzhen/Dongguan), Jiangsu (Suzhou/Kunshan), and Zhejiang (Ningbo). These factories — while not doing CIS design — are the direct suppliers to module companies and the indispensable physical foundation of China's CIS supply chain operation.
7.6 China CIS Industrial Cluster vs. Global Comparison
Japan's Kyushu CIS cluster (Sony-led): deep technology, complete self-control. Korea's Gyeonggi CIS cluster (Samsung-led): tight integration with Samsung's semiconductor ecosystem. China's CIS cluster (Shanghai+Suzhou+Shenzhen): largest market, most engineers, most complete supply chain — but dependent on overseas advanced manufacturing.
The implication: China's CIS industrial cluster's competitive advantage is exactly what Japan/Korea lack — massive domestic downstream market and relatively complete supply chain configuration. China's main disadvantage is what Japan/Korea excel at — high-end process manufacturing capability. A complementary division of labor will persist through 2030.
VIII. Segment Market Deep Dives
8.1 Flagship vs. Mid-to-Low End: One Market, Two Worlds
Flagship Market: Sony's Three-Lock Moat
Sony's dominance in the smartphone flagship CIS market rests on three reinforcing advantages: (1) technological generation gap (dual-layer transistor pixel, hybrid bonding in development); (2) brand premium (LYTIA brand as quality certification); (3) engineering validation depth (12–18 months of joint customization with handset brands, creating high switching costs).
Mid-to-Low End: Domestic Brand Battleground
The 1300MP–50MP mid-range segment (RMB 800–3,000 price band) is the direct battleground for Galaxycore, SmartSens, and OmniVision. In sub-800 price handsets, Galaxycore dominates with >70% domestic share. Front cameras (selfie lenses) are a key entry point for domestic CIS companies into handset brands' supply chains.
8.2 Automotive ADAS: Fastest Growing, Highest Barriers
Automotive CIS CAGR is approximately 9.4% (2024–2030). Technical barriers include: AEC-Q100 automotive qualification (12–18 months certification); ISO 26262 functional safety (ASIL B–D); global shutter; HDR (≥120 dB); supply chain long-term stability (10–15 year supply commitment).
In China, automotive CIS competition is shifting from foreign-dominated to "foreign + domestic co-existing." Domestic NEV brands have stronger motivation for domestic supply, creating systematic localization opportunities for OmniVision (the leader) and SmartSens (fast ramping).
Driver Monitoring Systems (DMS) and Occupant Monitoring Systems (OMS) are rapidly growing segments driven by EU GSR2 regulations requiring DMS from 2024.
8.3 Security: AI Replacement Structural Opportunity
Security CIS is SmartSens's base, with China's domestic localization rate exceeding 60%. Hikvision and Dahua's combined global market share exceeds 40%, preferring deep custom cooperation with domestic CIS companies. The AI camera upgrade cycle (from 2MP to 8MP+AI) creates sustained structural replacement demand. SmartSens's super-starlight CIS and OmniVision's wide dynamic range low-light products are the core beneficiaries.
8.4 Industrial Vision and Medical: High-Unit-Price Niche Markets
Industrial machine vision cameras require extremely high frame rates (100fps+), universal global shutter, pixel uniformity, and wide operating temperature ranges. OmniVision has dedicated industrial CIS product lines. Medical endoscopy requires ultra-small pixels, ultra-low power, and biocompatibility — OmniVision is a global top-three supplier of endoscopic CIS with high margins.
IX. Technology Evolution Path: From BSI Stacking to the Single-Photon Era
9.1 The Underlying Logic of Technology Evolution
The core physical constraint — "the smaller the pixel, the worse the light sensitivity" — drives all major CIS technology innovations: BSI, stacking, global shutter, pixel binning, hybrid bonding, and SPAD all represent attempts to circumvent this constraint.
9.2 BSI Stacking: The Current Mainstream Platform
Three technology tiers have emerged:
- Tier 1 (2012–2018): Two-layer stacking (pixel layer + logic layer via TSV); 3–5× readout speed improvement
- Tier 2 (2019–2023): Three-layer/multi-layer stacking (+ DRAM buffer); supports 4K 120fps video
- Tier 3 (2024+): Hybrid bonding — Cu-Cu direct bonding replaces TSV; interconnect density improves 100–2,500×; expected to achieve broader mass production in 2025–2027
For domestic CIS design companies, achieving hybrid bonding faces both technological and manufacturing challenges — SMIC and Hua Hong currently lack mature hybrid bonding CIS production processes.
9.3 Global Shutter Cost Evolution
Global shutter CIS costs approximately 2–3× rolling shutter equivalents in 2024; expected to narrow to 1.2–1.5× premium by 2030, making global shutter the automotive ADAS standard configuration.
9.4 SPAD: From Laser Ranging to Single-Photon Imaging
SPAD enables digital photon counting with ~100-picosecond time resolution, making it ideal for dToF LiDAR. Sony's IMX459/IMX516 SPAD sensors are commercially deployed. Domestically, SPAD CIS is almost entirely absent from commercial production — the next major frontier in CIS localization.
9.5 Pixel Binning and Computational Photography
Samsung's 200MP 16:1 pixel binning and Sony's large-format single-sensor approaches represent two competing philosophies. AI-enhanced CIS (in-sensor computing) is becoming the next competitive frontier.
9.6 AI-Integrated Sensors
Near-term (2024–2027): On-chip ISP upgrade with AI noise reduction and HDR synthesis moving into sensors. Longer-term (2027–2030+): Full in-sensor computing with embedded lightweight NPU outputting structured data rather than raw pixels. VeriSilicon's AI IP has relevant synergies.
9.7 Glass-Substrate Stacking: Post-2028 Outlook
Glass Core Substrate replacing silicon in advanced packaging promises lower dielectric constant, better thermal properties, and higher interconnect density. Expected in prototype validation 2028–2030.
9.8 CIS Technology Roadmap Timeline
| Milestone | Year | Leader | China Status |
|---|---|---|---|
| BSI mass production | 2009 | Sony | — |
| 2-layer stacked BSI | 2012–2013 | Sony/Samsung | OmniVision ~2016+ |
| 3-layer + DRAM stacked | 2019–2021 | Sony | OmniVision mid-tier |
| Dual-layer transistor pixel; 200MP | 2023 | Sony/Samsung | Galaxycore 50MP (2024) |
| Hybrid bonding mass production | 2025–2026 | Sony/TSMC | 5 years from production |
| Global shutter permeates ADAS | 2026–2028 | onsemi/OmniVision | OmniVision OX08B mass production |
| In-sensor AI commercializes | 2027–2029 | Sony IMX500 | VeriSilicon IP + domestic CIS |
| SPAD solid-state LiDAR mass production | 2028–2030 | Sony/startups | Early stage, major gaps |
| Glass-substrate CIS validation | 2028–2030 | TSMC/Sony | Jcet G-WLP foundation |
9.9 Linkage with Semiconductor Equipment and Laser Industry
Key CIS manufacturing equipment: wafer bonders (EV Group, SUSS MicroTec — both subject to export controls), TSV etch systems (Lam Research, SPTS), lithography tools (ASML, Nikon). For the automotive LiDAR connection: SPAD arrays combined with pulsed lasers (905nm NIR or 1550nm SWIR) and TDC circuits form complete solid-state LiDAR systems. If China achieves coordinated progress in CIS + SPAD LiDAR + mmWave MMIC radar, the full autonomous driving perception stack could become domestic, a strategic achievement far more significant than any single sensor's market share number.
X. Risk Matrix: Cycle, Monopoly, Price Wars, and Controls
10.1 Handset Cycle Dependency: The Industry's Largest Systemic Risk
The 2022–2023 industry deep freeze validated this risk concretely: combined profit of China's three major CIS players collapsed from ~RMB 8 billion (2021 peak) to ~RMB 600 million (2023 trough). The fundamental response strategy is business diversification — accelerating the proportion of automotive, security, and industrial revenue, which are less correlated with handset cycle volatility.
10.2 Sony-Samsung High-End Monopoly: Structural Barriers
The less-than-5% domestic share in flagship handset CIS reflects not inadequate technology but three overlapping commercial barriers: brand trust, engineering validation depth, and supply chain switching costs. Even if domestic CIS technically catches up to Sony specs, large-scale adoption in flagship supply chains requires years of consecutive successful models, consumer validation, and third-party evaluation recognition — a 3–5 year engineering project, not a procurement decision.
10.3 Price Wars: Mutual Attrition Among China's Three Players
Galaxycore's de-facto net margin of ~1%, SmartSens's ~6.6%, and OmniVision's ~13% reveal the pressure concentration. Sustained price wars compress R&D investment, slow technology upgrade, and create a vicious cycle where companies struggle to escape low-end competition. The only escape is finding differentiated market segments with defensible margins — automotive being the most important current opportunity.
10.4 U.S. Export Controls: Long-Term Unresolved Risk
Current direct impact on CIS is limited — controls primarily target AI accelerators and HBM. Indirect transmission channels include: TSMC advanced-node restrictions, EDA tool controls, and manufacturing equipment controls. Short-term (2025–2026): low risk. Medium-term (2027–2028): moderate risk as domestic players push toward advanced products. Long-term (2029–2030): high uncertainty around OmniVision's U.S. entity compliance status.
10.5 Geopolitical Supply Chain Risks
Taiwan hosts the core advanced CIS foundry capacity. Any major change in Taiwan Strait dynamics would create non-linear disruptions to the CIS supply chain. TSMC's overseas fab expansion (Japan, Arizona, Germany) provides gradual diversification, but advanced-node capacity in Taiwan will remain near-total through 2027.
XI. 2026–2030 Forecast: The Climb from USD 23.7 Billion to Over USD 30 Billion
11.1 Global CIS Market Scale Forecast
Comprehensive prediction combining Yole Group (CAGR ~4.4%) and Mordor Intelligence (CAGR ~7.1%) middle-ground estimates:
| Year | Market Scale (USD billion) | YoY Growth | Key Driver |
|---|---|---|---|
| 2024 (actual) | ~23.7 | +6.4% | Handset recovery + automotive ramp |
| 2025E | ~24.8–26.2 | +5–10% | AI handset upgrades + ADAS penetration |
| 2026E | ~26.2–28.2 | +5–8% | Automotive GS permeation + security AI replacement |
| 2027E | ~27.8–30.2 | +5–7% | L3 AV scale deployment prep |
| 2028E | ~29.3–32.0 | +5–6% | SPAD LiDAR mass production |
| 2030E | ~31.0–35.5 | +3–5% | Market maturity with automotive still high-growth |
11.2 Handset CIS: From Dominant to Relatively Declining
Projected 2030: ~USD 17.5–20 billion, but share declining from ~63% to ~55–58%, as automotive and security grow faster. Volume driver: increasing cameras per phone; price driver: premium ASP improvement.
11.3 Automotive CIS: The Most Certain High-Growth Track
Automotive CIS 2030 projected: ~USD 4.5–5.5 billion (CAGR ~9.4%), rising from ~11% to ~15–17% of global CIS. Key drivers: ADAS regulatory mandates globally; per-vehicle camera count rising from ~4 to ~7–8; China NEV exports bringing automotive CIS to global markets.
11.4 Security CIS: AI Replacement Continuous Incremental
Security CIS 2030 projected: ~USD 3.5–4.2 billion (CAGR ~10–13%), driven by AI camera replacement upgrades in China and global security infrastructure investment.
11.5 China Market and Localization Rate Forecast
China CIS market from ~RMB 52.7 billion (2024) to ~RMB 85–100 billion (2030), CAGR ~9–11%. Localization rate from ~30% to ~45–50% (automotive + security as main increments).
| Track | 2024 Rate | 2027E | 2030E |
|---|---|---|---|
| Flagship handsets | <5% | <8% | 10–15% |
| Mid-to-low handsets | ~50–70% | ~65–80% | ~70–85% |
| Automotive CIS | ~20–30% | ~30–40% | ~45–55% |
| Security CIS | ~60% | ~68% | ~75–80% |
| Industrial/medical | ~20% | ~25% | ~30–35% |
| Composite | ~30% | ~37% | ~45–50% |
11.6 Scenario Analysis: Optimistic, Base, Pessimistic
Optimistic (probability ~25%): No new TSMC restrictions; China NEV global exports drive automotive CIS; SPAD domesticated; SMIC hybrid bonding breakthrough. Result: 2030 composite localization ~52–55%; Will/OmniVision automotive CIS global share ~17–20%; combined Big Three revenue >RMB 75 billion.
Base (probability ~55%): Status quo maintained; automotive CIS grows at projected CAGR; limited flagship handset breakthrough (10–15%). Result: composite localization ~45–50%; combined Big Three revenue ~RMB 60–65 billion. (This is the central forecast in §11.5.)
Pessimistic (probability ~20%): New TSMC restrictions cut off 7nm below for mainland entities; OmniVision U.S. entity listed on Entity List; another handset downcycle. Result: composite localization ~32–38%; combined Big Three revenue ~RMB 40–50 billion.
XII. Conclusion: The Localization Road from 30% to 50% Is Not a Frontal Assault
12.1 Structural Insight: Sony Is Unassailable, but the Battlefield Extends Beyond
China CIS localization success should not be measured by whether it can beat Sony. Sony's dominance in flagship handset CIS is the product of twenty-plus years of technical accumulation, brand building, and supply chain investment — not achievable by simply increasing R&D spending within five years. The truly winnable battlegrounds are automotive ADAS and security cameras, where Sony and Samsung are not absolute rulers, domestic players have sufficient technical competitiveness, China is the largest market, and local advantages are clear.
12.2 Three Structural Observations
Technology gaps and commercial barriers are separate: Even when specs converge, brand trust and switching costs remain higher barriers than technical performance. The transition from "specs matched" to "mainstream adoption" requires 3–5 years of consecutive engineering validation.
The automotive time window has an expiry date: If OmniVision and SmartSens don't achieve automotive CIS scale ramp within 2026–2028, the window will narrow as Sony and Samsung intensify their push. The "10–15 year long-term automotive supply contracts" mechanism means whoever wins early adoption locks in a decade of stable share.
Manufacturing capability is the ceiling of localization depth: The current 30% is design-layer localization, not manufacturing-layer. Advancing SMIC/Hua Hong CIS advanced process nodes is the fundamental variable determining how deep localization can go.
12.3 Five Key Milestones (2025–2030)
- 2025–2026: Galaxycore's 50MP enters mainstream flagship supply chain
- 2026–2027: SmartSens automotive CIS achieves mass production in 2–3 major domestic NEV models
- 2027–2028: OmniVision automotive CIS revenue share exceeds 40%, completing strategic transformation
- 2028–2029: Domestic SPAD CIS first mass production
- 2029–2030: Composite domestic localization rate reaches ~45%
12.4 factory data platforms Factory Perspective
Within the supply chain from chip design to camera modules to end devices, thousands of manufacturing enterprises participate, forming the real factory foundation of China's CIS industry. factory data platforms's B2B platform covers 4.8 million active factories — a dataset that makes visible where optical module processing, precision CIS packaging, FPC flexible boards, and lens assembly capacity is actually concentrated across China.
12.5 Final Judgments
- Market scale is unambiguously upward: Global CIS 2024 ~USD 23.7 billion, 2030 >USD 30 billion.
- Sony's 50% share will not be dislodged before 2030: LYTIA brand deepening + Fab 5 expansion ensures leadership.
- China's Big Three growth is real, not a bubble: Supported by actual revenue and order data from annual reports.
- 30%→50% localization path is automotive + security, not flagship handsets: Automotive contributes the largest and most certain increments.
- Full-chain manufacturing domestication is a 2030+ long-term proposition: Design localization is achieved; manufacturing localization (hybrid bonding, advanced nodes) needs 5–10+ years of sustained R&D.
The race is still on. Flagship handsets are Everest for China's CIS industry; automotive and security are the North Face route currently being climbed — not a lesser objective, but the path that builds strength for the summit attempt.
factory data platforms Factory Database
The B2B data platform covering 4.8 million active Chinese factories. Unlike enterprise lookup tools (Qichacha, Tianyancha), factory data platforms specializes in identifying "real, active factories." Query camera module, optical component, precision CIS packaging, and FPC manufacturers in the CIS supply chain — with cross-filter by industry and region.
Find Factories on factory data platforms →Data Sources and References
- Will Semiconductor (Will Semiconductor Co., Ltd.) 2024 Annual Report (April 2025, Cninfo)
- Galaxycore (688728) 2024 Annual Report (April 2025, STAR Market)
- SmartSens Technology (688213) 2024 Annual Report (April 2025)
- Jcet (Suzhou WLCSP, 603005) 2024 Annual Report (April 2025)
- Sunny Optical Technology Group 2024 Annual Results (March 2025)
- O-Film Group (002456) 2024 Annual Report (April 2025)
- Yole Group, Status of the CMOS Image Sensor Industry 2024/2025
- Yole Group, "CIS Market to Reach More than $30B by 2030" (July 2025)
- TechInsights, Smartphone Image Sensor Market Forecast 2024–2029
- Mordor Intelligence, CMOS Image Sensors Market Size, Growth Report
- Grand View Research, CMOS Image Sensor Market Size, Industry Report, 2030
- Digitimes, "Sony Accelerates Image Sensor Growth in Automotive" (2025)
- Securities Market Weekly, "Domestic Image Sensors Race Forward" (June 2024)
- OFweek Sensor Network, "Two International Giants Retreat — Global CIS Reshuffles" (March 2025)