China’s core AI industry is already worth more than $175 billion, and the National Development and Reform Commission has set an ambitious goal of growing AI-related industries to over $1.45 trillion by 2030.
As of 2025, the country already had more than 6,200 AI companies operating across foundation models, cloud computing, computer vision, robotics, AI chips, autonomous driving, enterprise software, and consumer applications.
The speed of adoption is equally striking. China had 602 million generative-AI users by the end of 2025, representing 42.8% of the country’s internet-user population. [1]
The following top Chinese companies represent very different approaches to winning the AI race. Some are technology giants with enormous distribution networks; some are AI-native startups focused exclusively on foundation models; and some specialize in computer vision or speech recognition.
Did you know?The US-China AI performance gap has fallen to just 2.7%. Chinese models such as DeepSeek have repeatedly reached or briefly matched the performance of leading US systems. [2]
Table of Contents
12. Megvii
Founded: 2011Key AI Products: Brain++, MegEngine
Competitive Edge: Over 1,300 business-related authorized patents
Megvii first gained global attention with Face++, its facial-recognition platform. The company has since expanded into AI infrastructure, robotics, autonomous logistics, smart-city systems, and deep-learning technologies.
In 2015, Megvii introduced Brain++, its proprietary AI development platform for model training, optimization, and deployment, and later open-sourced MegEngine, a deep-learning framework designed for efficient AI research and production deployment.
It has developed several influential open-source AI architectures, including YOLOX, RepVGG, and RepLKNet, which have been widely adopted by AI researchers and developers worldwide.
In 2024, Megvii partnered with automotive giant Geely Holding Group to build AI-driven smart mobility platforms and advanced autonomous driving hardware and software. [3]
11. SenseTime
Founded: 2014Revenue: $700 million+
Key AI Products: SenseNova multimodal models
Competitive Edge: Decade-long leadership in computer vision
SenseTime has undergone a major transformation from conventional computer vision into generative AI and multimodal foundation models. In 2024, it launched SenseNova, a unified multimodal design that handles understanding, reasoning, and visual generation simultaneously.
It has deep expertise in computer vision, including facial recognition, object detection, image segmentation, video analytics, medical imaging, and autonomous-driving perception. In fact, the company has accumulated one of the largest proprietary vision AI datasets and research portfolios in China.
In 2026, the company released SenseNova U1 that combines text understanding, reasoning, and image generation into a single system using a custom NEO-Unify architecture. This eliminates the translation layers typically used between separate language and image models. [4]
They have also developed SenseNova-Skills, an open-source AI skill framework for autonomous agents, as well as lightweight multimodal models that significantly reduce token consumption and improve enterprise deployment efficiency.
SenseTime has already vast experience deploying AI in finance, healthcare, retail, government, and smart-city environments. It serves 660+ enterprise customers and has a 57% repeat-purchase rate. [5]
10. MiniMax
Founded: 2021Revenue: $80 billion+
Key AI Products: Talkie, Hailuo AI, MiniMax Agent
Competitive Edge: Rapid international expansion
MiniMax combines frontier multimodal AI research with globally distributed consumer products.
Its MiniMax-01 research series introduced Lightning Attention, an architecture designed for extremely long-context processing and efficient scaling.
The company’s models combine language, vision, audio, video, and agent capabilities, enabling products such as MiniMax Agent, MiniMax Audio, MiniMax Code, and Hailuo AI to operate on a shared foundation-model platform.
MiniMax products reach more than 200 million users across over 200 countries and regions, including North America, Europe, and Asia. Its AI companion app, Talkie, is among the most downloaded AI entertainment apps globally, while Hailuo AI has become a major text-to-video platform competing with leading global services.
Financially, MiniMax has experienced explosive growth. In 2025, it generated nearly $79 million in revenue, representing 159% year-on-year growth. More than 70% of that revenue came from overseas markets, differentiating it from many domestic-focused Chinese AI companies. [6]
In 2026, the company spent roughly $2 billion on AI infrastructure, frontier-model research and development, and global expansion. [7]
9. Moonshot AI
Founded: 2023Valuation: $30 billion+
Key AI Products: Kimi, Kimi K-series models
Competitive Edge: Long-context intelligence leadership
Moonshot became famous through its Kimi large language model and AI productivity platform.
Kimi can process extremely large amounts of information. It is designed for long-context reasoning, allowing users to analyze books, research papers, legal documents, financial reports, codebases, and large enterprise knowledge repositories.
The platform now includes Docs, Sheets, Slides, website generation, coding tools, research workflows, and autonomous AI agents,
Moonshot released Kimi K2 Thinking, a 1-trillion-parameter mixture-of-experts reasoning model, and later introduced Kimi K3, a 2.8-trillion-parameter open-weight model.
K3 has shown particularly strong performance in coding, reasoning, and agent-style tasks, while maintaining substantially lower operating costs than many proprietary AI rivals.
By 2026, Moonshot AI reached $300 million in annualized recurring revenue and a valuation of around $30 billion. [8]
8. DeepSeek
Valuation: $70 billion+
Key AI Products: Open-weight LLMs, reasoning AI
Competitive Edge: Radical cost efficiency
DeepSeek is the Chinese AI startup that shocked the global technology industry by proving that frontier-level AI models could be trained and deployed at a fraction of the cost of many leading Western systems.
It became internationally famous after the release of DeepSeek-V3 and DeepSeek-R1. The models delivered performance comparable to leading closed-source AI systems in reasoning, coding, mathematics, and general language tasks, while offering open weights that developers could download, modify, and deploy.
These models use techniques such as Multi-head Latent Attention (MLA), DeepSeekMoE sparse architectures, Multi-Token Prediction, and Group Relative Policy Optimization (GRPO). These innovations help reduce memory use, training costs, and inference expenses while maintaining strong model performance. [9]
In 2026, DeepSeek launched V4-Pro and V4-Flash. V4-Pro contains about 1.6 trillion parameters, and both models support 1 million-token context windows.
7. iFlytek
Revenue: $4 billion+
Key AI Products: Spark (Xinghuo), iFlyCode
Competitive Edge: Dominance in Chinese speech and language AI
iFlytek built its reputation on speech recognition, speech synthesis, natural language processing, machine translation, and educational AI long before the current generative AI boom.
Today, its flagship Spark (Xinghuo) large language model powers AI assistants, coding tools, healthcare AI systems, educational platforms, and enterprise AI solutions across China.
Spark is designed for Chinese-language intelligence and professional applications, with a strong focus on education and medical AI.
Its technologies support real-time transcription, multilingual translation, pronunciation assessment, educational tutoring, medical documentation, voice biometrics, and digital humans, and its open platform provides 600+ AI capabilities to more than 11 million developer teams. [10]
Plus, Spark is trained entirely on Huawei’s Ascend AI platform, making it one of China’s leading LLMs built on a fully domestic AI infrastructure stack.
6. Zhipu AI
Founded: 2019Enterprise Customers: 12,000++
Key AI Products: ChatGLM, CodeGeeX
Competitive Edge: Enterprise AI & government deployment
Zhipu has built the GLM (General Language Model) family, which powers conversational AI, coding assistants, AI agents, multimodal systems, speech generation, and enterprise AI platforms.
Technically, the ChatGLM and GLM-4 series were trained on more than 10 trillion tokens across Chinese, English, and other languages. Academic evaluations have shown strong performance in reasoning, coding, mathematics, long-context understanding, and tool use.
Zhipu’s other products, including AutoGLM and AutoClaw, highlight its focus on AI agents capable of using external tools, browsing the web, writing code, and completing complex tasks autonomously.
In fact, Zhipu has grown into China’s largest independent large language model developer, with more than 12,000 enterprise customers, 45 million developers, and deployments across over 80 million end-user devices.
The company has also seen rapid commercial growth. Its open API platform reached $250 million in annual recurring revenue by 2026, representing roughly 60-fold year-over-year growth. [11]
Big Tech AI Divisions
5. Huawei (Ascend/Pangu)
Revenue: $127 billion+
Key AI Products: Pangu-Σ model, Ascend processors
Competitive Edge: Sovereign AI and domestic technology stack
Huawei is building the entire AI stack: Ascend AI chips, AI supercomputing clusters, the MindSpore AI framework, CANN software tools, Huawei Cloud, and the Pangu foundation model family.
Unlike most AI companies that focus primarily on models or applications, Huawei designs AI hardware, networking, cloud infrastructure, operating systems, developer tools, and industry-specific AI models.
However, their core AI strategy revolves around Ascend processors and Pangu foundation models. The Ascend chip family is Huawei’s answer to NVIDIA’s AI accelerators, while Pangu provides large language, multimodal, predictive, scientific computing, and industry-specific AI models for sectors like government, finance, and energy.
Pangu-Σ, introduced in 2023, reached 1.085 trillion parameters, while newer models such as Pangu Ultra and Pangu Pro MoE have demonstrated competitive reasoning, coding, and enterprise performance.
As for the Ascend chip ecosystem, Huawei has announced a roadmap that includes the Ascend 950, 960, and 970, alongside next-generation AI supercomputing systems such as the Atlas 950 and Atlas 960 SuperPods. [12]
4. Tencent (Hunyuan)
Founded: 1998Revenue: $108 billion+
Key AI Products: Hunyuan LLM, Hy3, WorkBuddy
Competitive Edge: WeChat ecosystem and gaming leadership
Tencent embeds AI across WeChat, Weixin, QQ, gaming, advertising, fintech, cloud computing, enterprise software, and developer platforms, allowing new AI capabilities to reach over a billion users almost instantly.
At the center of its AI strategy is the Hunyuan large language model family, which has been expanded into reasoning, coding, multimodal, image, video, and 3D generation models.
The company has built a broad frontier AI portfolio including Hunyuan-Large, HunyuanImage, HunyuanVideo, and Hunyuan3D. These models are particularly strong in image generation, video generation, 3D generation, multimodal understanding, and long-context reasoning,
The company has also expanded its Yuanbao AI assistant and WorkBuddy platform, integrating Hunyuan directly into consumer messaging and enterprise productivity workflows.
Plus, Tencent’s massive gaming business gives it access to huge amounts of real-time player and interaction data. Hunyuan can use this ecosystem to improve NPCs, game development, content creation, moderation, and live-service operations, creating a valuable link between AI and gaming. [13]
3. Baidu (ERNIE)
Revenue: $18.6 billion+
Key AI Products: ERNIE Bot (Wenxiaoyan), Baidu AI Cloud
Competitive Edge: Operates China’s largest search ecosystem
Baidu began investing heavily in deep learning and natural language processing years before the launch of ChatGPT, and its ERNIE (Enhanced Representation through Knowledge Integration) research program started in 2019.
ERNIE models were designed around “knowledge integration”, combining large-scale language modeling with knowledge graphs, entity relationships, and structured information. This gives ERNIE particularly strong performance in Chinese-language understanding, factual reasoning, and search-related tasks.
The launch of ERNIE Bot (Wenxiaoyan) in 2023 marked Baidu’s entry into the generative AI race. The chatbot grew rapidly after receiving regulatory approval, reaching 300 million users by 2024,
By the same year, daily interactions on the ERNIE platform had climbed to 1.65 billion, highlighting the rapid growth of AI usage across Baidu’s ecosystem.
By 2026, Baidu AI Cloud had been ranked China’s No. 1 AI cloud provider by IDC for several consecutive years and held a 19.9% share of China’s LLM cloud market.
Beyond AI models and cloud services, Baidu has put AI into commercial use through its Apollo Go robotaxi service. The service has surpassed 20 million cumulative rides and reached more than 350,000 rides in a single week. [14]
2. Alibaba (Qwen)
Revenue: $148.4 billion+
Key AI Products: Qwen (Tongyi Qianwen)
Competitive Edge: Open-source leadership at massive scale
Alibaba combines one of Asia’s largest cloud platforms with one of the world’s most widely downloaded open-source AI model families. This allows the company to monetize AI through cloud services, enterprise deployments, and AI infrastructure, rather than relying only on consumer subscriptions.
Alibaba launched Qwen (Tongyi Qianwen) in 2023, and it has quickly become a global open-source AI phenomenon. The Qwen family has nearly 1 billion downloads worldwide, with more than 1.1 million downloads each day.
Its consumer Qwen Chat app has reached about 203 million monthly active users. Qwen is also being used by more than 90,000 businesses and 2.2 million corporate users on DingTalk for everything from customer support and coding to analytics and productivity.
Moreover, Qwen models are trained on up to 18 trillion tokens and achieve competitive performance across reasoning, coding, mathematics, and multilingual benchmarks.
Alibaba plans to invest more than $53 billion in AI and cloud infrastructure over three years, making it one of the largest AI investment programs in Asia. [15]
1. ByteDance (Doubao/Seed)
Revenue: $186 billion+
Key AI Products: Doubao AI assistant, Seedance
Competitive Edge: Massive consumer distribution through TikTok and Douyin
ByteDance has evolved from a social-media company into a full-stack AI company with consumer applications, enterprise cloud AI, video generation, coding models, multimodal research, and frontier foundation models.
The launch of Doubao in 2023 marked the beginning of this shift, and the chatbot rapidly became one of the most widely used AI apps in China. Today, Doubao serves more than 345 million monthly users. [16]
Behind Doubao is ByteDance’s Seed AI division, which was established in 2023 to focus on artificial general intelligence and cutting-edge AI research. Seed has developed advanced models for reasoning, vision-language tasks, coding, and AI agents, including Seed-Thinking, Seed1.8, and Seed2.0.
ByteDance is reportedly training a frontier AI model with up to 10 trillion parameters, while its models now process more than 180 trillion tokens per day.
The company is also investing heavily in AI infrastructure, with reports estimating more than $12 billion spent in 2025 on data centers, GPU clusters, and model training. [17]
Read More
- 14 Largest Data Center Companies
- 9+ Different Types of AI With Examples
- 16 Leading Chinese Semiconductor Companies
- Generative AI users in China reach 602 million, Global Times
- AI’s influence on society has never been more pronounced, Stanford University
- Megvii signed a deal with Geely and a local tech hub, South China Morning Post
- Unifying multimodal understanding and generation with NEO-unify architecture, arXiv
- SenseTime’s revenue exceeded market expectations, PR Newswire
- MiniMax founder’s net worth more than doubles amid rapid growth, Forbes
- MiniMax says it has closed $2 billion funding round, Investing
- China’s Moonshot pauses Kimi subscriptions amid hot demand, Reuters
- A strong, economical, and efficient mixture-of-experts language model, arXiv
- iFLYTEK launches open platform in Central Asia, Global News Wire
- China AI developer Zhipu hits record $112 billion valuation, Caixin Global
- Key products in Huawei’s AI chips and computing power roadmap, Reuters
- Chinese tech giant Tencent sees spending surge, CNBC
- Apollo Go reaches 20 million trips, 190 million fully driverless kilometers, CleanTechnica
- Alibaba plans to spend $53 billion on AI in a major pivot, Yahoo Finance
- Doubao has 345 million monthly active users, Binance Square
- TikTok owner ByteDance plans to spend $12 billion on AI chips, Reuters
