Top 30 Domestic Leading AI Chip Companies and Their Profiles
-
This list is compiled based on the selection results announced at the 2023 Global AI Chip Summit.
01 Axera
Axera is a company dedicated to building world-leading artificial intelligence perception and edge computing chips, founded in May 2019. Axera focuses on the research and development of high-performance, low-power AI vision processor chips and independently develops neural network processors for inference acceleration. Integrating powerful computing capabilities with ultra-low power consumption, the company enhances image quality through neural networks and supports various AI vision tasks such as object detection and facial recognition.
Axera's first product, the AX630A, was launched in January 2022 with a well-planned product roadmap and received enthusiastic market response. In addition, Axera has introduced a series of other products, including the AX620A, AX620U, and AX170A, all of which possess industry-leading advantages such as "high computing power, low power consumption, and excellent image processing capabilities."
Axera Semiconductor (Shanghai) Co., Ltd. and Axera Semiconductor (Ningbo) Co., Ltd. are both subsidiaries of Axera. The Shanghai subsidiary primarily focuses on AI chip R&D and the construction of basic computing platforms, while the Ningbo subsidiary targets application scenarios such as autonomous driving, robotics, and AR/VR.
02 Cambricon
Cambricon was founded in 2016, specializing in the research and development of artificial intelligence chip products and technological innovation. It is committed to creating core processor chips in the field of artificial intelligence, enabling machines to better understand and serve humans. Cambricon provides a series of intelligent chip products and platform-based basic system software that integrate cloud-edge-device, hardware-software collaboration, training-inference fusion, and a unified ecosystem. In 2021, Cambricon announced the first official appearance of the Siyuan 290 intelligent chip and accelerator card, the Xuanxi 1000 intelligent accelerator after mass production, and the intelligent edge computing device—Tenglong STAU-R100-C—released in collaboration with Super Vision. The company has subsidiaries in Beijing, Shanghai, Shenzhen, Hefei, Kunshan, Nanjing, and other locations.
03 Kunlunxin Technology
Kunlunxin (Beijing) Technology Co., Ltd. was formerly the Baidu Intelligent Chip and Architecture Department. It completed independent financing in April 2021, with a first-round valuation of approximately 13 billion yuan. Kunlunxin Technology is an AI chip company with deep expertise in architecture, chip implementation, software systems, and application scenarios, focusing on creating general-purpose AI chips with strong versatility, ease of use, and high performance. The company has large-scale deployments in fields such as the internet, smart industry, and smart finance, empowering various industries with AI computing power. Kunlunxin Technology has strong R&D capabilities, with team members possessing top-tier academic backgrounds globally. Most members come from leading companies such as Baidu, Qualcomm, Marvell, and Tesla. The company has proposed the 100% self-developed XPU core architecture for general artificial intelligence computing, and its research results have been successfully published in top international academic conferences such as Hotchips and ISSCC.
04 Novumind
NovuMind was founded in 2021 and specializes in the design and development of edge artificial intelligence chips. It is the world's first chip company to embed CIS into AI edge computing SOCs. Unlike traditional algorithms in the AI industry, NovuMind's products, combined with highly integrated machine image recognition sensors, use streamlined and real-time internal computing to achieve independent operation of image detection, tracking, and recognition without the need for connectivity, any peripheral components, or internet access, all while consuming extremely low power. This provides low-cost, low-power edge computing solutions for smart doorbells, surveillance systems, drones, machine vision, smart homes, and smart toys.
05 Enflame Technology
Enflame Technology is a provider of cloud and edge computing power products in the artificial intelligence field, dedicated to building a computing power foundation for general artificial intelligence. It offers original innovations and independently developed AI acceleration cards, system clusters, and hardware-software solutions. Its products can be widely applied in various industries and scenarios, including the internet, intelligent computing centers, smart cities, smart finance, scientific computing, and autonomous driving.
The founding team of Enflame Technology is led by Mr. Zhao Lidong, who established the company in March 2018 and is responsible for strategic planning, financing, and business operations. Before joining Tsinghua Unigroup, Mr. Zhao worked in Silicon Valley for over 20 years, holding various positions in management, R&D, and marketing. From 2007 to 2014, he served at AMD, responsible for CPU/APU product planning, market analysis, and expansion. As a Senior Director of the Product Engineering Department, he led the R&D of CPU/GPU/APU and multiple related core IPs, managing a team of over a thousand people and participating in the establishment of the China R&D Center. He also worked at Juniper Networks (formerly Netscreen), responsible for the development of network security chips, and at S3 Inc., engaged in the R&D of GPU graphics processor chips.
06 Siengine Technology
Siengine Technology is an intelligent technology company jointly invested by Geely Holding Group, Zhejiang ECARX Technology Co., Ltd., and Arm China. It is registered in Wuhan Economic and Technological Development Zone, Hubei Province, with branches in Beijing and Shanghai.
The company primarily focuses on the R&D, production, and sales of automotive processors and automotive electronic semiconductors. Its main products include integrated solutions for automotive electronic chips such as smart cockpits, autonomous driving, and electronic/electrical architectures. In 2023, Siengine Technology was awarded "Top 10 Autonomous Driving Chips" and "Annual Local Intelligent Driving Solution." The same year, its "Dragonhawk-1" chip officially entered mass production.
07 Biren Technology
Biren Technology is a tech company dedicated to developing original general computing architectures, establishing efficient software/hardware platforms, and providing integrated solutions in the field of intelligent computing.
The company is committed to first focusing on cloud-based general-purpose intelligent computing, gradually surpassing existing solutions in multiple fields such as AI training and inference, and graphics rendering, to achieve breakthroughs in domestic high-end general-purpose intelligent computing chips. In its development path, Biren Technology will initially concentrate on cloud-based general-purpose intelligent computing, progressively surpassing existing solutions in areas like AI training and inference, and graphics rendering, to realize breakthroughs in domestic high-end general-purpose intelligent computing chips.
08 SemiDrive
SemiDrive is a self-developed high-end chip and solution provider, founded in December 2018 and headquartered in Shanghai, with R&D centers in Beijing, Shenzhen, Xi'an, Chengdu, and Toronto, Canada.
The company's core technical team consists of veterans from international giants like AMD, with an average of over 18 years of experience in chip and software design. They have led the design and mass production of more than 100 chips, including the industry's first 7nm GPU. SemiDrive primarily focuses on cutting-edge fields such as artificial intelligence, data centers, graphics rendering, smart cities, smart transportation, edge computing, industrial quality inspection, the metaverse, and digital twins, aiming to provide vast computing power for the pixel world.
Currently, the company has launched a series of cloud-based general AI inference and video acceleration card products based on its VUCA unified computing architecture and has completed the design and development of a domestic 7nm cloud GPU chip.
09 Kunyun Technology
Kunyun Technology is a leading AI computing power provider dedicated to delivering next-generation AI computing platforms with high performance, low latency, and cost-effective computing power. With its groundbreaking dataflow AI chip technology at the core, Kunyun offers integrated intelligent solutions combining computing power, algorithms, and platforms for industries such as energy, chemicals, power, and smart cities, facilitating digital transformation across sectors. The company's advantages include hardware-software integration, over 100+ AI recognition algorithms, cost reduction through legacy system utilization, and user-friendly solutions. Its service achievements include 1,000+ benchmark cases and coverage in 82.3% of regions nationwide.
10 Pingtouge Semiconductor
Pingtouge Semiconductor Co., Ltd. was established in September 2018 as Alibaba Group's wholly-owned semiconductor chip business entity. Pingtouge offers a full-stack product series spanning data center chips, IoT chips, and processor IP licensing, achieving end-to-end chip design coverage. It has released product series such as Xuantie, Yitian, and Hanguang.
11 Iluvatar CoreX
Iluvatar Corex is a high-tech company co-founded by Silicon Valley experts and domestic industry talents, specializing in intelligent computing with a focus on high-performance data processing and deep learning applications. Established on December 29, 2015 with Diao Shijing as legal representative, the company achieved several milestones in 2022: the "Tiankai 100" series products received nearly 200 million RMB in cumulative orders in April; the "Zhikai 100 chip" was successfully powered on in May; the DeepSpark open-source community launched in August along with the Hundred Applications Open Platform; and the "Zhikai 100 Series Acceleration Card" was officially released in December.
EZchip
EZchip is an AI chip design company founded in 2017, specializing in providing general computing power for edge/end-side AI applications with core competencies in machine vision algorithms and SoC chip design.
EZchip's SoC integrates self-developed analog and digital IP with outstanding energy efficiency. Their products are widely used in smart home appliances, intelligent office equipment, and smart education systems, providing AIoT solutions. The company possesses strong chip design capabilities and excellent AI algorithm implementation, with years of industry experience enabling rapid mass production to facilitate edge/end-side AI applications across various industries.
Horizon
With the mission of "Empowering Machines for a Safer and Better Life," Horizon Robotics is a leading provider of high-performance intelligent driving computing solutions. Based on its innovative Brain Processing Unit (BPU) architecture, Horizon has outlined a comprehensive roadmap for self-developed AI chips. In 2017, it launched its first-generation chips, Journey 1 and Sunrise 1. By 2019, Horizon introduced China's first automotive-grade AI chip, Journey 2, and the next-generation smart application acceleration engine, Sunrise 2. In 2020, Horizon accelerated its chip iteration with the release of Journey 3, a high-performance automotive AI chip, and Sunrise 3, a new-generation AIoT chip.
In July 2021, Horizon unveiled Journey 5, the industry's first full-scenario central computing chip integrating autonomous driving and intelligent interaction, boasting a single-chip computing power of 128 TOPS. With this launch, Horizon became the only AI computing company capable of providing comprehensive chip solutions covering L2 to L4 autonomous driving scenarios.
Horizon has established deep collaborations with numerous automakers (listed alphabetically), including Audi, BAIC Group, BYD, Changan Auto, Great Wall Motors, Dongfeng Motors, Voyah, Venucia, Volkswagen Group, GAC Group, GAC Aion, JAC Motors, Geely Auto, Li Auto, NIO, Chery Auto, SAIC Motor, SAIC Maxus, SAIC Roewe, SAIC-GM-Wuling, FAW Group, Hongqi, and Ziyoujia, as well as Tier 1 suppliers like Desay SV, Neusoft Reach, Continental, Freetech, Faurecia, Huayang, APG, and In-Driving, rapidly building an open and win-win ecosystem for smart vehicles.
14 Black Sesame Technologies
Black Sesame Intelligent is a leading supplier of automotive-grade intelligent vehicle computing chips and chip-based solutions. The company started with the Huashan series of high-performance chips for autonomous driving and recently launched the Wudang series of cross-domain computing chips to meet the more diverse and complex needs of intelligent vehicles' advanced features. The company's proprietary automotive-grade products and technologies equip intelligent vehicles with mission-critical capabilities, including autonomous driving, smart cockpits, advanced imaging, and connectivity. Through its self-developed IP cores, algorithms, and software-driven SoCs and SoC-based solutions, it provides full-stack autonomous driving capabilities to meet a wide range of customer needs.
15 ListenAI
ListenAI is a high-tech company focused on providing system-level (SoC) chips for intelligent terminals. Its products can be widely used in smart homes, home appliances, education and office, consumer electronics, smart vehicles, and other fields. Since its establishment, the company has been committed to deeply coupling cutting-edge global AI algorithms with its leading independent chip design to create cost-effective single-chip solutions, promoting the true arrival of the era of intelligent connectivity of all things.
16 TsingMicro
TsingMicro Intelligent Technology was founded in 2018 and is positioned as a leader in reconfigurable computing chips. Its core technologies and products are based on a reconfigurable computing architecture (such as voice processing chips and video processing chips), providing chip products and solutions that start from the edge and extend to the cloud.
17 Simm Computing
Simm Computing was established in April 2019. From its inception, the company has adhered to a technological roadmap based on the RISC-V instruction set for designing high-performance chips for data centers. Its development strategy involves evolving from domain-specific architecture (DSA) to general-purpose computing (CPU), with a product plan that starts with AI chips, focusing first on inference and later on training. The company currently employs 230 people, over 60% of whom hold postgraduate degrees or higher, with R&D personnel accounting for 90% of the workforce. It has filed 194 patent applications, 96% of which are invention patents.
The company's products include 12nm AI inference chips, with self-developed computing cores, chip microarchitecture, and upper-layer software compiler toolchains. Simm Computing possesses independent and controllable technological innovation advantages in chip architecture and integrated hardware and software, which can support the computing power platforms in the data center industry.
Simm Computing is not only a member of the RISC-V International Foundation but also the chair unit in the AI direction. Globally, Simm Computing's high-end AI acceleration chips are among the leading commercial implementations of the RISC-V instruction set in the data center field.
18 YunTian LiFei
Shenzhen Intellifusion Technologies Co., Ltd. was established in August 2014 as an artificial intelligence enterprise with full-stack capabilities in algorithms, chips, and big data. Through its core competency of "algorithm chipification" and the technological approach of "end-cloud collaboration," Intellifusion has built a comprehensive core capability platform encompassing IoT perception convergence, algorithm empowerment services, and knowledge graph construction. The company went public on the STAR Market in 2023.
In the fields of digital cities and human settlements, Intellifusion has successfully implemented a series of benchmark solutions, bringing safe, intelligent, and convenient AI experiences to various industries. As a leading domestic AI enterprise, Intellifusion was the first to propose the "1+1+N" self-evolving urban intelligence practice framework, actively constructing a self-learning, self-evolving digital city intelligent collaborative development system to accelerate the growth of the AI industry.
19 Denglin Technology
Denglin Technology, a leader in high-performance general-purpose computing, was founded at the end of 2017. It focuses on chip R&D and technological innovation for high-performance general-purpose computing platforms, aiming to create cutting-edge chip products and platform-based foundational system software that integrate cloud-edge-end, hardware-software collaboration, and training-inference fusion. The company's independently innovated GPU+ (a software-defined on-chip heterogeneous computing architecture based on GPGPU) perfectly addresses the dual challenges of universality and high efficiency through architectural innovation while maintaining compatibility with programming models and software ecosystems including CUDA/OpenCL. Extensive customer product tests have proven that for AI computing, GPU+ significantly outperforms traditional GPUs in performance, especially energy efficiency.
As the first domestic GPU company to achieve large-scale commercial implementation, Denglin's first innovative AI computing accelerator based on GPU+, Goldwasser, has been widely deployed in various application scenarios from edge to cloud computing, successfully filling the technological, product, and commercial gaps in the domestic high-performance GPGPU field.
20 Reexen Technology
Shenzhen Reexen Technology Co., Ltd. was co-founded by several PhD graduates from top universities worldwide. The neuromorphic sensing-memory-computing integrated architecture represents one of the new architectures in the post-Moore era of integrated circuits. Reexen's brain-inspired chips adopt analog-digital hybrid forms to achieve the R&D and mass production of sensing-memory-computing integrated chips.
21 Moffett AI
Moffett provides AI chip acceleration solutions for both cloud and edge computing. By optimizing computing paradigms, Moffett supports fully sparse neural network development, delivering ultra-high computing power and ultra-low power consumption for general AI computing platforms.
Moffett is an AI chip designer committed to disruptive innovation, adopting world-leading sparse algorithms to create the next generation of AI chips and become a global leader in the AI chip 2.0 era. Founded in Silicon Valley in 2018, Moffett currently has its headquarters in Shenzhen. The founding team consists of top AI scientists from Carnegie Mellon University and core members from world-leading semiconductor companies (such as Intel, Marvell, and Oracle) with extensive experience in high-volume chip development.
22 Timechi Intelligence
Timechi Intelligence Technology (Shanghai) Co., Ltd. was established on July 25, 2018, specializing in the research and development of various edge-side artificial intelligence processing chips. In the AIoT era of intelligent connectivity, we are committed to starting from the needs of real-world scenarios, providing chip products and complete system-level solutions for a wide range of edge-side devices that support multi-modal intelligent human-computer interactions such as voice, vision, imaging, and display, as well as data processing, through domain-specific architectures and customized chip designs. In July 2023, Timechi Intelligence completed its B+ and B++ rounds of financing, with investors including New Vision Capital, Gaoxin Caijin under Jinan Gaoxin Holdings, Yonghui Capital, existing shareholder Bangming Capital, and some individual investors.
23 Lightelligence
Lightelligence is an artificial intelligence infrastructure company originating from the Massachusetts Institute of Technology (MIT), designing faster and more energy-efficient processors and algorithms for AI computing using optical technology. Founded in 2017, Lightelligence is one of the earliest AI chip companies globally to focus on optical computing as its core technology. In 2017, Dr. Shen Yichen, founder and CEO of Lightelligence, and his team at MIT published a cover paper in the journal Nature Photonics, presenting the new starting point of integrated photonic computing to the world.
In April 2019, Lightelligence officially released the world's first photonic chip prototype board, successfully integrating the entire photonic computing system, which had previously occupied half a laboratory, into a conventional-sized board. In tests, over 95% of the entire model's computations were completed on the photonic chip, with processing accuracy approaching that of electronic chips (over 97%), while the time required to complete matrix multiplication could be reduced to less than 1/100th of that of electronic chips.
The company adheres to the corporate vision of "Creating Useful Technology," specializing in the design of photonic computing chips and the layout and accumulation of core technologies such as algorithms. It aims to build a comprehensive photonic computing ecosystem, providing optimized solutions for the next generation of computing paradigms (cloud computing).
24 NextVPU
NextVPU is a technology innovation company specializing in the design of AI computer vision processing chips and terminal applications. Headquartered in the Integrated Circuit Industrial Park of Shanghai Zhangjiang Science City, the company has over 300 employees, with R&D accounting for 80% of its workforce. It has secured multiple rounds of funding from renowned institutions, accumulating a total of 500 million RMB in financing, and has applied for over 200 domestic and international patents.
NextVPU has successfully launched its N-series, D-series, and V-series chips, which set world records in visual processing capabilities. These chips are widely applicable in professional security, assisted driving, robotics, smart retail, and other fields.
25 HiSilicon
HiSilicon is a semiconductor company founded in October 2004, with its predecessor being Huawei's Integrated Circuit Design Center established in 1991. Headquartered in Shenzhen, HiSilicon has design divisions in Beijing, Shanghai, Silicon Valley (USA), and Sweden.
HiSilicon primarily produces chips and solutions for wireless networks, fixed networks, and digital media, successfully applied in over 100 countries and regions worldwide. In the digital media sector, HiSilicon has launched SoC network surveillance chips and solutions, video phone chips and solutions, DVB chips and solutions, and IPTV chips and solutions. In 2019, HiSilicon's Q1 revenue reached $1.755 billion, a 41% year-on-year increase, significantly outpacing other semiconductor companies and rising to the 14th position in the rankings. In 2020, HiSilicon's sales reached $2.67 billion, marking its first entry into the global semiconductor TOP10 list.
26 CoolSemi Microelectronics
Hefei CoolSemi Microelectronics Co., Ltd. (formerly Shanghai CoolSemi Microelectronics Co., Ltd.) was established in July 2011, aiming to become a global leader in intelligent chips. The company leverages three core technologies—intelligent sensing, intelligent computing, and intelligent transmission—to develop proprietary chip architectures and core IP, providing high-performance, low-power chips and related toolchain solutions specifically for artificial intelligence. Its products are primarily used in smart security, smart hardware, smart automotive, and other fields.
CoolSemi Microelectronics is headquartered in Hefei High-Tech Zone, with branches and subsidiaries in Shanghai, Chengdu, Shenzhen, and other locations. Currently, the company has over 200 employees, with technical developers accounting for more than 80% of the workforce, all holding master's or doctoral degrees.
27 Moore Threads
Moore Threads Intelligent Technology (Shanghai) Co., Ltd. was established in Shanghai in September 2020, with wholly-owned subsidiaries and R&D centers established in Beijing, Nanjing, Chengdu, Hangzhou, Shenzhen, Wuhan, and Changsha. Moore Threads is dedicated to providing full-stack GPU chip solutions for heterogeneous computing, widely applicable in cutting-edge fields such as artificial intelligence, smart cities, data centers, cloud computing, autonomous driving, digital twins, and the metaverse, offering robust computing power support for the development of the digital economy.
Moore Threads develops a full-stack GPU chip product line, including the Xisi
N-series GPUs for AI inference, the Xiyun C-series GPUs for training AI models with hundreds of billions of parameters and general-purpose computing, and the Xicai G-series GPUs for graphics rendering, meeting the data center demands for "high energy efficiency" and "high versatility." All Moore Threads products utilize fully self-developed GPU IP, featuring proprietary instruction sets and architectures, complemented by a complete software stack (MXMACA) compatible with mainstream GPU ecosystems. With inherent advantages in energy efficiency and versatility, the company provides customers with comprehensive ecosystem solutions integrating both hardware and software, serving as the computational cornerstone for data center construction and the digital and intelligent transformation of industries under the "dual carbon" policy.28 Sophon
SuanNeng is committed to becoming a global leader in general computing power, focusing on the research, development, and application of computing power products such as AI and RISC-V processors. Following the philosophy of comprehensive open-source and open ecosystems, the company collaborates with ecosystem partners to create a full-scenario product matrix covering "cloud, edge, and terminal," providing computing power products and integrated solutions for applications such as the internet, urban operations, intelligent computing centers, general security, smart manufacturing, and AIGC. The company has R&D centers in over 10 cities in China, including Beijing, Shanghai, Shenzhen, Qingdao, and Xiamen, as well as in countries like the United States and Singapore. Since 2016, its SOPHON series products have undergone multiple iterations, with each generation achieving exponential improvements in energy efficiency compared to the previous one.
29 XinChi Technology
Founded in 2018, XinChi Technology specializes in providing high-performance and highly reliable automotive-grade chips. The company has R&D centers in Shanghai, Beijing, Nanjing, Shenzhen, and Dalian, with offices in Changchun and Wuhan. As a leader in full-scenario smart automotive chips, XinChi focuses on delivering high-performance, reliable automotive-grade chip products and solutions for future smart mobility, covering smart cabins, autonomous driving, central gateways, and high-performance MCUs. These encompass the most critical chip categories in future automotive electronic and electrical architectures, achieving the vision of "Four Chips in One, Empowering the Soul of the Car."
In terms of automotive certification, XinChi has obtained the German TÜV ISO 26262 ASIL D functional safety process certification, AEC-Q100 Grade 1/Grade 2 reliability certification, German TÜV ISO 26262 ASIL B functional safety product certification, and dual certifications from the State Administration for Market Regulation and the State Cryptography Administration for national security and information security. This makes XinChi a certified automotive-grade chip company with all four certifications.
In terms of capital, Xinchi has received investments from funds including SAIC Jinshi, Walden International, Matrix Partners China, Lenovo Capital, Sequoia Capital, Vertex Ventures, CDB Equipment Fund, CITIC Securities, Morning Capital, Shanghai Science and Technology Innovation, Yunhui Capital, Hechuang Capital, Jiangsu Jinshi Transportation Technology Industry Fund, Anhui Transportation Control Jinshi Investment, Guozhong Capital, Huatai Insurance, Qianhai Saisei, Zhangjiang Hi-Tech, Chuangtu Capital, and Lanpu Capital.
Currently, Xinchi has completed the tape-out of four series of chips, obtained the highest-level automotive-grade certification, and achieved mass production. The company serves over 260 clients and has nearly 200 designated projects, covering more than 90% of domestic automakers and some international mainstream car manufacturers, including SAIC, Chery, Changan, Dongfeng, FAW, Nissan, Honda, Volkswagen, and Li Auto.
30 Zhichun Technology
Zhichun Technology was founded in 2017 and specializes in the field of in-memory computing chips. It innovatively uses Flash memory to complete the storage and computation of neural networks, addressing the AI memory wall problem, improving computational efficiency, and reducing costs.
The R&D team was co-founded by Dr. Wang Shaodi and Dr. Guo Xinjie, along with multiple scholars and industry practitioners, with an average of over 10 years of industry experience.
The company's WTM2101 chip is optimized for low-power AIoT applications, capable of executing large-scale deep learning operations with power consumption ranging from microwatts to milliwatts. It can be applied in markets such as smart voice and smart health, and has already entered mass production and commercial use. The WTM8 series chips are designed for real-time processing of 4K-8K video.
In January 2023, Zhicun Technology completed a 200 million yuan Series B2 financing round, bringing its total funding to nearly 800 million yuan. Moving forward, the company will continue to focus on the field of in-memory computing chips, leading the industrialization of in-memory computing.