Global AI Industry Transition

Currently, the global artificial intelligence industry is undergoing a critical transition from a “technological race” to a “productivity revolution.” According to IDC, the global AI market is expected to exceed $500 billion by 2026, with a compound annual growth rate (CAGR) of 32.1% from 2024 to 2028. As a core engine for global AI development, China is building a globally competitive AI innovation ecosystem driven by policy leadership, technological breakthroughs, and industrial integration. As of 2025, the Ministry of Industry and Information Technology reports that the number of AI companies in China has surpassed 4,800, with a core industry scale exceeding 680 billion RMB, covering key areas such as smart chips, large models, intelligent computing clusters, and industry applications. Following the comprehensive achievement of the goals set in the State Council’s “New Generation Artificial Intelligence Development Plan” by 2025, the “AI+” initiative is advancing deeply—leading the global open-source landscape with the DeepSeek series of large models and accelerating the industrialization of cutting-edge directions like embodied intelligence and AI agents. China is making a historic leap from following technology to leading innovation.

Forbes China AI Selection

Forbes China has long focused on the deep changes brought by AI to the business world and its power to reshape economic and social development models. Following the launch of the 2024 Forbes China AI Technology Company selection and the comprehensive upgrade of the 2025 selection system, Forbes China has once again partnered with FIRST MOMENTUM to launch the “2026 Forbes China AI Technology Companies Top 50 Selection.” This selection actively responds to the national “AI+” strategic deployment and upgrades the selection criteria—adding five new sub-categories: “AI Technology Breakthrough Companies,” “AI Business Implementation Demonstration Companies,” “AI Globalization Benchmark Companies,” “Sustainable Development Excellence Practice Companies,” and “Ecosystem Construction and Open Source Contribution Companies.” The aim is to comprehensively select outstanding companies and individuals in technology depth, business transformation, global expansion, social value, and ecosystem building, setting new coordinates for the high-quality development of China’s AI industry.

Selection Rules and Evaluation Dimensions

The selection process employs a dual-track nomination mechanism of “self-declaration + expert recommendation” to ensure openness, authority, and professionalism. The self-declaration channel is open to AI technology companies registered in China with independent legal status. Companies must submit detailed application materials based on their actual operating conditions, covering dimensions such as technological architecture, core products, commercialization achievements, intellectual property layout, and strategic development plans, subject to a comprehensive compliance review and information verification by the selection committee.

The evaluation committee consists of senior experts from academia, industry, investment, and industry associations, who recommend outstanding candidate companies based on their deep insights into scientific research frontiers, technology transfer, capital operations, and industrial ecosystems.

Based on the collected nominations, the selection committee conducts systematic background checks and multiple rounds of screening using big database cross-verification, public information checks, and on-site investigations of key companies, ultimately forming the list of selected companies.

Core Evaluation Dimensions

The core evaluation dimensions include:

• Technological Innovation Capability: Examining R&D investment intensity, patent quantity and quality, core technological breakthroughs, and technological leadership.
• Market Performance: Assessing revenue growth, market share, industry ranking, and commercialization progress.
• Application Results and Actual Impact: Analyzing the industry penetration rate and social benefits of innovative application cases.
• Investment and Financing Status: Examining financing rounds, amounts, investor backgrounds, and capital activity.
• Sustainable Development Potential: Evaluating alignment with national policies, ESG performance, and long-term competitiveness.
• Team and Culture: Assessing core team backgrounds, industry-academia-research collaboration capabilities, and corporate culture development.

Key Insights from the Selection Results

The selected companies exhibit a significant characteristic of coexisting “technological depth” and “scene evolution”—with leading companies mastering full-stack technological capabilities continuing to lead, and startup companies achieving breakthroughs in cutting-edge directions rapidly emerging. This confirms that China’s AI industry has entered a critical turning point from “application-driven” to “technology-driven.”

Industry Structure

In terms of industry structure, technology-oriented companies account for 61.2%, while application-oriented companies account for 34.1%. Technology-oriented categories include large models, embodied intelligence, AI chips, AI for Science, enterprise-level AI, and intelligent agents; application-oriented categories cover AI healthcare, AI education, AI marketing, intelligent transportation, and AI manufacturing. Forbes China found that more than half of the companies in this year’s selection are foundational technology companies, while application-level companies span over ten vertical fields including healthcare, finance, education, and manufacturing, with a ratio of approximately 6:4 between technology and application companies.

Track Distribution

The track structure presents a three-layer pattern of “head concentration, waist expansion, and long tail emergence.” In the head layer, there are 13 companies in embodied intelligence and robotics (15.3%) and 11 companies in large models (12.9%), together accounting for 28.2%. In the mid-waist, there are 9 companies in AI marketing and content (10.6%), 8 in AI healthcare and wellness (9.4%), and 8 in enterprise-level AI and intelligent agents (9.4%), together accounting for 29.4%. In the long tail layer, there are 25 companies in AI chips, AI education, intelligent transportation, AI manufacturing, AI for Science, digital humans, AI perception and IoT, frontier exploration, and other tracks, accounting for a total of 33.1%. This year’s selection covers over 20 sub-directions, significantly broadening the track coverage compared to previous years.

Regional Distribution

The regional pattern exhibits the dual characteristics of “high concentration in the head and significant internal differentiation.” Beijing leads with 21 companies (24.4%), with 15 in the main selection and 6 in sub-selections, covering 15 directions including large models, AI chips, and enterprise-level AI. Shanghai and Shenzhen each have 12 companies (14.0%), with significant structural differences—Shanghai’s selected companies focus on embodied intelligence, AI chips, and physical AI simulation, while Shenzhen’s are distributed across AI marketing, intelligent device manufacturing, and embodied intelligence. Hangzhou has 9 companies (10.5%), with 7 in the main selection and 2 in sub-selections, leading in the number of large model companies. Chengdu has 11 companies (12.8%), with 2 in the main selection and 9 in sub-selections, focusing on AI education, AI agents, and AIGC film applications. Other cities like Wuhan (3), Guangzhou (3), Changsha (3), Chongqing (2), Hefei (2), Nanjing (2), Qingdao (1), Suzhou (1), Dongguan (1), Changzhou (1), Sanya (1), and Wenzhou (1) also have selected companies, showcasing the flourishing development of AI across various cities.

Selected Companies List

Sub-selection Companies List

Core Discoveries: From “Chinese Innovation” to “World Engine”

Frontier Exploration

As Ground AI enters deeper waters, the most ambitious Chinese companies have set their sights on broader territories—from near-Earth orbit to fundamental scientific laboratories. Space computing and AI for Science are two of the most forward-looking directions, transitioning from science fiction concepts to engineering practices, representing the ambition and capability of China’s AI industry to explore unknown boundaries.

In the field of space computing, Guoxing Aerospace successfully launched a space computing constellation with “12 stars in one arrow,” achieving a single star computing power of 744 TOPS, enabling the deployment of general large models in orbit. They plan to build an artificial intelligence space infrastructure consisting of 2,800 computing satellites, expanding “computing power as a service” from the ground to space. Xingfan Intelligent focuses on the “algorithm-defined chip” route, developing intelligent agent chips and intelligent computing products, building integrated hardware and software capabilities for space computing, embodied intelligence, and token factories, having delivered thousands of P computing power products, providing a key piece for China’s self-controlled computing power.

In the AI for Science field, Daina Technology deeply integrates multimodal embodied intelligence with industry-specific AI models, creating an “AI + darkroom laboratory” fully autonomous unmanned system, freeing researchers from repetitive physical labor and promoting the transition of scientific research from “labor-intensive” to “intelligent autonomous.”

Breakthroughs in Core Technologies

From the “efficiency revolution” of domestic large models to the “mass production” of AI chips, from the “self-controlled” intelligent computing clusters to the “precision breakthroughs” in industrial vision, China’s foundational AI technologies are undergoing a critical transition from “usable” to “user-friendly.” By 2026, the market share of domestic AI chips is steadily increasing, and domestic large models are achieving top global levels in multiple international evaluations. “Vertical integration”—mastering the full stack of models, chips, and frameworks—is becoming a core competitive barrier for leading companies.

In the large model and intelligent computing infrastructure field, China’s large model cohort has achieved a leap from “following” to “leading.” DeepSeek released DeepSeek-V4 in May 2026, realizing a full-stack migration from Nvidia CUDA to domestic computing power, supporting ultra-long contexts of one million tokens, with its sparse attention mechanism and fine-grained expert parallel technology reducing long-context reasoning costs by over 50%. Kimi, known for its long context and programming capabilities, supports 256K tokens of context and can dynamically expand to 300 sub-agents, achieving an accuracy rate of 80.2% in the SWE-Bench programming test, accelerating its transformation from consumer-level dialogue products to productivity tools. StepStar has developed a comprehensive model matrix with leading performance, releasing 42 models cumulatively: the base model Step 3.5 Flash topped the OpenRouter platform’s OpenClaw call volume, favored by global users of small crayfish; StepAudio 2.5 TTS ranks high in the Artificial Analysis evaluation list. Tongyi Qianwen launched the Qwen3.6 flagship model, continuously achieving “Day 0” adaptation support for DeepSeek, becoming a core force in China’s open-source large models going global. Volcano Engine released the Doubao large model 2.0 series in February 2026, achieving global first-tier capabilities in multimodal understanding and agent tool invocation, with daily token usage exceeding 120 trillion.

In the AI chip field, domestic chips are transitioning from “usable” to “user-friendly.” Cambricon’s SiYuan 590 chip has achieved full-scene mass shipment, and the new generation flagship SiYuan 690 is set for mass production in early 2026, with FP16 computing power exceeding 700 TFLOPS, continuously achieving “Day 0” adaptation for DeepSeek-V3.2 and DeepSeek-V4. Birun Technology’s BR166 series products have achieved full-form mass production and large-scale delivery, successfully delivering multiple thousands of computing cluster projects, with the next-generation BR20X chip planned for official launch in 2026, achieving breakthroughs in both performance and efficiency in training and inference scenarios.

In the intelligent manufacturing field, Innovation Qizhi is a leading provider of “AI + manufacturing” solutions, with its AI solutions having achieved large-scale implementation in heavy industries such as steel, energy, panels, and semiconductors. Zhongke Yaolu is known as the “national team of intelligent perception,” having developed a new generation of spatial perception methods with its self-developed spatial intelligent camera; relying on a matrix of onboard, airborne, and miniature products, it is laying out intelligent driving, low-altitude economy, and core tracks of embodied intelligence. Lide Vision has created an embedded FPGA + AI edge computing intelligent detection architecture, with its self-developed industrial detection large model achieving over 99% classification accuracy, processing over 300 defects per second, breaking foreign monopolies with its “integrated chip computing.” Suli Technology focuses on AIoT technology R&D and implementation, creating a soft and hard integrated product system to empower cities, agriculture, shipping, motor control, and high-end equipment.

In the cognitive intelligence and multimodal interaction field, AI is transitioning from “perception” to “decision-making.” Zhongke Wenge, born from the Institute of Automation of the Chinese Academy of Sciences, has built a decision-making intelligence system centered on DOMA, ranking among the top in China’s enterprise-level large model-driven decision-making intelligence market in 2024. Huayuan Computing is a pioneer in cognitive intelligence technology in China, developing a new generation of cognitive intelligence engine platform based on the deep integration of cognitive science and intelligent science. Xiaoshi Technology positions itself as the “definer and leader of industrial visual intelligence agents,” winning championships in international competitions such as CVPR and NIST, and has completed Hong Kong stock registration, planning to go public in 2026. Jiu Si Intelligent’s self-developed “Poly Voice” voice large model ranks among the global top tier, building a global AI Voice Agent platform, redefining human-machine communication through voice interaction.

In the intelligent terminal and infrastructure field, selected companies are constructing a complete intelligent infrastructure stack from “cloud brain” to “edge terminal.” iFLYTEK, as a representative of original technology innovation in artificial intelligence, has always transformed voice, translation, and cognitive large model capabilities into products that solve practical needs. The upcoming iFLYTEK AI glasses are its latest achievement in promoting AI applications across multiple scenarios through innovative hardware—this product not only enriches iFLYTEK’s hardware matrix but also objectively verifies its continuous promotion of AI’s penetration into lighter and more diverse scenarios based on over 20 years of accumulated technical foundation. In the intelligent voice interaction and cognitive terminal layer, Future Intelligence’s viaim iFLYTEK AI conference headset has created a new category of conference headsets, leading sales for four consecutive years, with over 1.5 million global registered users, serving as a typical case of the penetration of cutting-edge AI technology into consumer-grade hardware. In the industrial intelligent agent platform layer, Kaiyuan Cloud Technology aims to become a global leading AI intelligent agent service operator with the mission of “making AI ubiquitous and making intelligence so simple.” As a specialized and innovative enterprise and a national high-tech enterprise, the company focuses on educational research, engineering transportation, port water affairs, biomedicine, and other scenarios, collaborating with top institutions to build full-stack solutions relying on its self-developed AI intelligent agent factory, empowering intelligent upgrades across various industries. In the intelligent consumer terminal layer, Kusai Intelligent, as a global leading ODM smartphone manufacturer, has been deeply involved in the industry for 20 years, with core businesses covering smart devices, printed circuit board components, and internet services (including AI agents), with operations spanning over 70 countries and regions, serving as a key link in the transition of AI terminals from “concept prototypes” to “mass production of millions.”

Rise of Embodied Intelligence

2026 is viewed by the industry as the year of commercialization for embodied intelligence. In this selection, the number of companies in the embodied intelligence and robotics track has increased by over 40% year-on-year, becoming one of the fastest-growing sub-directions. IDC predicts that China’s spending on embodied intelligence users will exceed $11 billion in 2026. Selected companies are forming a closed loop across the entire chain from “body development” to “brain construction” to “scene implementation,” with humanoid robots transitioning from “demo showcases” to “factory mass production,” as China’s embodied intelligence industry gains global competitiveness.

In the humanoid robot direction, UBTECH Robotics is one of the few companies globally to achieve commercial mass production of humanoid robots, delivering over 500 units of the industrial humanoid robot Walker S2 in 2025, with annual production capacity exceeding 1,000 units and total orders nearing 1.4 billion RMB, covering five major scenarios including aerospace manufacturing, automotive manufacturing, and 3C electronics manufacturing, accelerating towards a target of 10,000 units in 2026. Yushu Technology is known for its extreme hardware cost reduction capabilities and motion control algorithms, achieving a global market share of 69.75% for its quadruped robots in 2024, ranking among the top in humanoid robot shipments in 2025; with annual revenue reaching 1.7 billion RMB in 2025 and a net profit of 600 million RMB, it has become one of the few profitable embodied intelligence companies globally, having submitted an IPO application to the Sci-Tech Innovation Board in 2026.

In the direction of embodied intelligence technology and applications, Variable Robotics focuses on the development of foundational models and general robots for embodied intelligence. Its WALL-B is an embodied intelligence foundational model based on the World Unified Model Architecture (WUM), achieving native multimodal fusion of vision, language, motion, and physical prediction, and has advanced commercial implementation in industrial logistics, home services, elderly care, and hotel scenarios. Songying Technology has developed a physical AI simulation system ORCA, fully replacing Nvidia Omniverse, providing millimeter-level precision physical simulation and large-scale synthetic data generation (cost reduced to 1% of traditional methods, efficiency improved by a hundredfold), adapting to domestic GPUs such as Muxi, Moore, and Hanbo, providing full-link capabilities of “scene construction—data synthesis—collaborative training” for embodied intelligence companies, serving as a key infrastructure for AI robots transitioning from “virtual training” to “real-world implementation.” Quantitative Party has completed a strategic advancement from consumer technology to intelligent hardware and embodied intelligence, focusing on general AI and physical intelligence as core technological foundations, building a comprehensive intelligent ecosystem across family, work, and life scenarios, integrating “hardware carriers + data-driven + ecological collaboration.” Tuosda serves as a benchmark enterprise transitioning from traditional industrial robots to embodied intelligence, adhering to the principle of “defining robot products by real industrial scenarios.” Leveraging its accumulated expertise and application scenarios in injection molding equipment and CNC machine tools, it constructs a closed loop of “scene + robot + data + AI,” targeting high-value, technologically accessible industrial scenarios, steadily expanding from injection molding scenarios to industrial multi-scenarios and commercial all-scenarios, promoting the implementation of embodied intelligence in the physical world. HintonAI focuses on AI decision-making engines and intelligent robot clusters, aiming to become a comprehensive intelligent hub driving efficiency revolutions from research to industry, promoting efficiency revolutions from “research → laboratory → small trials → medium trials → mass production.”

In the intelligent transportation and logistics field, selected companies are redefining transportation infrastructure. Jiadu Technology launched the transportation industry large model “Jiadu Zhixing Transportation Large Model,” with innovative smart subway achievements implemented in 48 cities globally, covering over 150 lines and 3,500 stations. Xidi Zhijia delves into the field of intelligent agents for transportation, reconstructing heavy-load autonomous driving technology systems with a physical AI frontier paradigm, accurately adapting to essential scenarios such as mining areas and hazardous chemical parks. New Stone Technology has introduced an unmanned vehicle instant delivery RaaS model, covering over 300 cities in 20 countries, with a cumulative driving distance exceeding 150 million kilometers, operating the world’s leading single-city unmanned delivery fleet.

AI Agents and Enterprise Intelligence

AI Agents are rapidly transitioning from proof of concept to enterprise-level deployment. The Institute of Computing Technology predicts that by 2028, over 15% of global work decisions will be autonomously completed by AI Agents. Companies mastering the capabilities of “vertical large models + engineering of intelligent agents” are becoming core suppliers for digital transformation.

In the enterprise-level AI platform direction, Zhongguancun Kejin has developed an enterprise intelligent agent and digital employee product matrix covering five major scenarios: intelligent marketing, intelligent sales, intelligent customer service, intelligent operations, and intelligent office, based on its self-developed Dezhuzhu large model platform, delivering RaaS services to over 3,000 leading enterprises and government clients globally. Mingtu Technology, as a professional provider of intelligent agents and digital humans, constructs a product matrix integrating intelligent agents with OPC supply-demand matching, cloud applications for digital humans, and capability development tools, developing over 1,000 industry partners and over 2 million collective intelligent agents, while simultaneously exporting intelligent agent standards to the international market along the Belt and Road, becoming one of the companies with extensive applications of industrial intelligent agents. Zhuoshi Technology, founded in 2018, has a core team from leading AI companies like Baidu and Alibaba, possessing industry-leading capabilities in large model algorithms, industry model development, and embodied intelligent AI brains, with its self-developed trillion-parameter “Xuanji Yuheng” model at its core, creating an expandable, ready-to-use, cost-effective industry large model platform, applications, and terminal products serving industrial manufacturing, enterprise services, health care, and education sectors. Dipu Technology went public on the Hong Kong Stock Exchange in 2025, forming a foundation for an “AI-level enterprise operating system” centered on the Deepexi enterprise large model and FastAGI enterprise intelligent agent platform. Lingyi Wanwu, as the first to go all-in on B2B as one of the “six small tigers” in large models, focuses on lightweight industrial large models and enterprise-level multi-agent implementation, covering five major industries: government, finance, industry, office, and sales, serving benchmark clients like international energy giants and AIA insurance. As a benchmark for China’s industrial AI going global, the company has collaborated with Kazakhstan to create the Alem Kazakh large model, meeting with President Tokayev and promoting the president’s order to adopt AI-enabled education pilot suggestions, pushing for digital transformation across various industries through a “top-down project” model.

In the vertical application direction, as a demonstrative enterprise for the landing of AI Agent commercial scenarios, Tuoduo Delivery integrates artificial intelligence, big data, and blockchain technology into the full lifecycle of asset management services, constructing a full-stack digital service system for the asset management industry, driving financial digital transformation with a dual strategy of “horizontal chain compliance and vertical intelligent service.” Shuishou Intelligent, established by Waterdrop Company in 2021, empowers innovation in insurance technology, healthcare, and general office scenarios. The company has applied for over 100 AI patents, with its AI customer service “Bao Xiaohui” achieving an average response time of just 1.5 seconds, while the AI underwriting expert “KEYI” boasts an accuracy rate of 99.8%, and the AI assistant “ClawSquare” enables autonomous communication and collaboration between different functional AI Agents, driving industry intelligent transformation with an AI-native strategy.

Diverse Tracks and Deep Evolution

From medical diagnosis to smart education, from embodied intelligence to space computing, from digital publishing to cross-border marketing, the technological capabilities of selected companies have embedded themselves in over 20 sub-fields of the national economy. This breadth of coverage indicates that artificial intelligence is transitioning from a technical tool for specific scenarios to a systemic infrastructure that spans industries and chains.

In Healthcare and Wellness

AI is moving from “auxiliary diagnosis” to “full lifecycle health management.” Weining Health serves over 6,000 healthcare institutions, launching the desktop-level AI medical product WiNBOT doctor-specific AI workstation in 2025, consistently leading the frontier of AI medical innovation. Shukun Technology is a pioneer in medical AI large models, creating the “Digital Human” platform and the “Shukun Kun” large model, becoming an AI company covering all modalities and scenarios in healthcare, winning the Science and Technology Progress Award in both Beijing and Jiangsu provinces in 2025, with products covering over 5,000 hospitals and multiple platforms, entering over 20 countries and regions. Fubao Intelligent focuses on the “robot + medical care” track, becoming an industry benchmark with its self-developed medical care robots and full-scene intelligent care solutions, serving over one million users across multiple scenarios including homes, communities, and hospitals, leading the innovation of the silver economy. Shanmu focuses on “personal health computing,” embedding medical-grade devices into users’ daily lives for continuous biological data collection, pioneering a new health management paradigm of “biological Physical AI.” Weimei Health empowers grassroots healthcare with the “Singularity Medical Inquiry” large model, constructing a full-cycle service system covering health consultation, early screening, auxiliary diagnosis, and chronic disease management, bridging the healthcare resource gap between urban and rural areas with AI.

In Education

Educational AI is transitioning from “tool assistance” to “teaching reconstruction.” Tianli International operates nearly 60 schools across 19 provinces and cities in China, serving over 145,000 students, with its Qiming Darun brand leading industry change through a dual-driven educational model of “moral education + AI.” Its self-developed “Tianli Qiming AI Companion” is an educational large model achieving large-scale application in campus teaching, trained on a billion-level dataset of teaching behaviors, deeply integrating knowledge graphs, intelligent diagnostics, and personalized learning path planning capabilities, deployed in 107 schools and serving over 250,000 teachers and students. The company plans to serve over 1,000 schools and more than 2 million students with AI capabilities over the next five years, fully unleashing the composite advantages of “scenes + data + educational genes.”

In Marketing and Content Creation

AIGC is reshaping the production relationships of the creative economy. Dongxin Times is a leading AI marketing company in China, with a revenue of 15.1 billion RMB in 2025, producing over 50,000 AI short videos per day, and pioneering the construction of AI marketing infrastructure based on “media traffic × scene tokens × marketing big data.” Quzhi Group is a leading AI interactive marketing service provider, dedicated to building a lifestyle platform of “AI + consumer scenarios.” The company has developed the AI-OMNI multimodal engine, launching products such as AI digital human guides, AI holographic marketing, AIGC middle platforms, and AI Agent workstations, providing intelligent marketing solutions for brands in food and beverage, new energy, and is expanding AI indoor entertainment spaces to overseas markets like Dubai and Singapore. DCG Data Transmission Group focuses on AI technology services in the publishing and reading industry, based on its self-developed “BOOKSGPT” large model, covering publishing business, distribution, and reading, achieving AI empowerment throughout the entire publishing process. Adhering to technological innovation, it has built a complete development system integrating production, education, and research, providing AI solutions for over 460 publishers and tens of thousands of editors, generating industry value exceeding 10 billion RMB and serving over 300 million readers. Chengdu Haiyi AI (SeaArt AI) is a global multimodal AI interactive entertainment and creation platform, ranking among the top in global AI image generation and AI product export, with a cumulative global user base exceeding 200 million, monthly active users exceeding 25 million, and this year launching the OPC “one-person company” AIGC incubation base, driving digital cultural upgrades with self-developed technology. Mika Vision Science is a pioneer in the AI era of digital imaging, a leading AIGC film production agency covering all categories of film content, with a team honored with national-level awards such as the “Five One Project” from the Publicity Department and the AI Application Innovation Award from the National Radio and Television Administration; in 2026, it won the Tencent Video “Creative Breakthrough AI Director Award,” promoting the integration of AIGC technology and film industrialization through collaboration with top universities to build an AIGC talent industry academy. Ying Sai AI, as a leading domestic AI marketing video service provider, focuses on the “AI + marketing” scene tokens. Its core product “inSaiHilight” reconstructs the e-commerce content production paradigm with “1 link generating 1 minute of video,” praised as the “DeepSeek moment of marketing videos”; its Ying Sai large model has completed the first batch of “algorithm + large model” dual registration, becoming a benchmark model in the AI marketing vertical.

In Public Services

AI is becoming a new infrastructure for social governance. Shengtong Technology deeply engages in the intelligentization of enterprise communication scenarios, achieving a response to service demands from 170 million people globally last year, facilitating better communication worldwide. Suker Intelligent, recognized as a specialized and innovative “little giant” and a benchmark enterprise for military-civilian integration, focuses on “AI + public safety,” relying on an integrated system of sky, land, and sea to build a multi-source remote sensing data foundation, creating a vertical large model for infrastructure risk assessment, achieving millimeter-level deformation monitoring over 800 kilometers in space and millisecond-level disaster warnings, empowering the protection of nearly 600,000 kilometers of transportation, water conservancy, and pipeline facilities, leading the “satellite + AI” integration application field. Leiming Technology drives the entire business chain of human resources with “service + technology + AI,” self-developing an integrated HRM management platform and AI intelligent training products, reducing recruitment cycles by 50%, improving HR management efficiency by 60%, and increasing new employee retention rates by 60%, not only helping enterprises achieve standardized personnel management and normalized talent reserve but also assisting enterprises in achieving sustainable and healthy operations.

Influential Figures

Building Industry-Academia-Research Closed Loops with Technological Depth and Cross-Field Integration

In a world where technology, business, and capital intertwine, innovators and leading figures are core elements, brave reformers, and future leaders. The influential figures selected this year have constructed a diverse landscape of “ecological leadership—academic depth—industrial reshaping—technology for good” through their unique paths, jointly promoting the leap of China’s AI cause to higher dimensions.

In the direction of cross-field leaders and ecological construction, Dr. Kai-Fu Lee, CEO of Lingyi Wanwu and Chairman of Innovation Works, has been engaged in the AI field for over 40 years, founding the Microsoft Research Institute in China (later upgraded to the Asia Research Institute), and was named a 2023 global “AI leader” by TIME. His Innovation Works has incubated over ten AI unicorns using the “VC + AI” model, and in 2023, he led the team to establish Lingyi Wanwu to create a “large model operating system,” while promoting the global expansion of Chinese AI technology into countries along the Belt and Road, continuously defining the evolution path of industrial AI through cross-field resource integration.

In the direction of academic depth and foundational innovation, Zhu Songchun, a globally renowned expert in computer vision, currently serves as the director of the Beijing Institute of General Artificial Intelligence and is the chairman of the Chongqing Artificial Intelligence Academy and the Chongqing General Artificial Intelligence Research Institute. He has received top international academic honors such as the Sloan Award, Marr Prize, and Helmholtz Award, leading China’s AGI foundational innovation efforts with the paradigm of “small data, big tasks.”

In the direction of industry-academia-research integration, Jiang Hongbo, an academician of the European Academy and Vice Dean of the School of Computer Science at Hunan University/Chief Digital Transformation Officer of China State Construction International Group, has delved into intelligent IoT and major infrastructure intelligent monitoring fields, pioneering a technical architecture of “AI algorithms + ubiquitous perception + cross-platform integration,” with research results applied to over 1,500 major infrastructure projects across 27 provinces in China, strengthening the national infrastructure safety defense with AI.

In the direction of industry leadership: technological entrepreneurship and track reshaping, selected figures are opening up new tracks with forward-looking insights, driving the large-scale implementation of AI technology in both the physical and digital worlds. Wu Minghui, founder of Minglue Technology, has a dual background in mathematics and AI from Peking University. He has led Minglue Technology to become the “first stock of Agentic AI,” with its GUI large model Mano topping the global dual lists of OSWorld and Mind2Web. He advocates that “Scaling Out is better than Scaling Up”—multiple agents collaborating over the network are more credible and secure than a single giant model. Mao Xingsheng, founder and chairman of Shukun Technology, has served as CTO of IBM China Research Center and global head of IBM Cloud Platform R&D, leveraging nearly 20 years of technical experience in both China and the US to build the “Digital Human” AI technology platform, with products covering over 5,000 medical institutions globally, driving the transition of medical imaging from “manual reading” to “intelligent diagnosis.” Ma Xiao, founder and CEO of Future Intelligence, adheres to the business philosophy that “AI is more useful the more vertical it is,” precisely targeting high-frequency essential workplace scenarios, opening up a new category of AI conference headsets, transforming AI technology into new productivity in the workplace, with products winning over 1.5 million high-net-worth core users globally, truly achieving large-scale commercial implementation of AI hardware. Tu Jing, founder and CEO of Zhuoshi Technology, was the first to propose the concept of “industry large models,” leading the construction of a self-controlled “Xuanji Yuheng” large model system, not only building a matrix of multi-domain industry large models but also deeply extending technology into the frontier of embodied intelligence. Qiu Jianhua, a member of the Medical AI Committee of the International Academy of Artificial Intelligence and Chairman of Zhihui Eye, leads the construction of the medical foundational large model “Bianshi,” reconstructing the underlying logic of healthcare services with multimodal AI.

In the direction of technology for good, selected figures practice social responsibility with the power of AI. Si Mahuapeng, founder, chairman, and CEO of Silicon-based Intelligence Group, proposed a series of AI innovation concepts such as silicon-based civilization, silicon-based life, and silicon-based labor globally in 2017, applying them to the Silicon-based Intelligence Group he founded. In 2025, he used AI to create the “Silicon-based Da Sima” series of technology self-media accounts, showcasing China’s hardcore technological prowess, which has become a technology content brand covering science and technology, finance, and people’s livelihoods. Luo Shi, chairman of Tianli International Holdings, founded Tianli Education in 2002. In 2025, he launched the “Tianli Qiming AI Companion”—an educational large model applied on a large scale in campuses, covering 107 schools and serving over 250,000 teachers and students, leading the intelligent transformation of education with the dual drive of “moral education + AI.” Wang Jun, co-founder of Weimei Health and a PhD in computer science from Peking University, leads the development of the “Singularity Medical Inquiry” large model, providing AI solutions for over 4,000 medical institutions, with over 75% in county and grassroots areas, bridging the healthcare resource gap between urban and rural areas with AI.

Conclusion: Anchoring Benchmarks, Igniting the Future

The release of the results of the 2026 Forbes China AI Technology Companies Top 50 Selection is not only a systematic review of the development trajectory of China’s AI industry over the past three years but also a profound response to the historical proposition of “from technological innovation to productivity revolution.”

Over the past three years, we have witnessed the evolution of the AI industry from “large model parameter competition” to “value-driven outcomes,” from “single-point technology applications” to “full-stack ecosystem construction,” and from “innovation in the Chinese market” to “participation in global rules.” The comprehensive renewal of this selection system is a proactive response to these profound changes—we not only focus on the height of technology but also on the warmth of technology; we not only measure the breadth of business but also the thickness of value.

Standing at the strategic height of developing new-quality productivity, China’s artificial intelligence industry is ushering in its best era. On the policy level, the “AI+” initiative is advancing deeply, with local industrial policies being intensively introduced, forming a policy resonance between the central and local governments; on the technological level, the full-stack independent innovation capability continues to strengthen, with key directions such as domestic large models, intelligent chips, and embodied intelligence achieving collective breakthroughs; on the industrial level, the integration and penetration of AI and the real economy are accelerating, injecting strong momentum into high-quality development. Forbes China will continue to use this selection as a benchmark to record the leap trajectory of China’s AI industry, honoring every explorer and practitioner who drives technology towards good, industry towards new, and society forward.

The future is here; only by leading with intelligence can we empower all things with AI.