Thousands of Firms Fuel China’s AI Surge, with a Government Registry Monitoring Their Progress.

20 January 2026
Artificial Intelligence, Blog
Thousands of Firms Fuel China's AI Surge, with a Government Registry Monitoring Their Progress.

When DeepSeek made its debut on the international scene in January 2025, it seemed to come out of nowhere. However, this large language model was merely one among the thousands of generative AI applications released in China since 2023—and there’s a public archive documenting each of them.

The leading internet regulator in the country, the Cyberspace Administration of China (CAC), mandates that any company launching an AI tool with “public opinion properties or social mobilization capabilities” must first register it in a public database known as the algorithm registry. To complete this submission, developers must demonstrate how their products mitigate 31 risk categories, including age and gender discrimination, psychological harm, and “violating core socialist values.”

Applicants submit their filings to their local CAC (for instance, the Shanghai CAC for firms registered in Shanghai), which then forwards the applications to the central CAC for final approval. Only after this step will a tool appear in the algorithm registry. While the European Union is working towards a comprehensive AI Act, notes Matt Sheehan, a research scholar at the Carnegie Endowment for International Peace, China’s regulatory approach is more fragmented, focusing on individual algorithms and developing iterative standards. (The US lacks a similar registration framework or central regulatory entity.)

Over time, the CAC has unintentionally created the most detailed map of a nation’s AI ecosystem globally.

*Data current as of April 2025, includes both “generative AI” and “deep synthesis” algorithms

Referencing the CAC’s update from August 2024, you would find DeepSeek listed as entry 152, a single entry in a neatly organized table. In the table, you can discover an AI managing homestays and another drafting patents. One assists obstetricians and gynecologists in a Shanghai maternity ward; another helps oversee state power grids. Kendra Schaefer and her team at Trivium China, a policy consulting firm based in Beijing, have been compiling these CAC updates into a detailed database, enhanced by their own research.

A Broad View of the Boom

Almost 80 percent of China’s generative AI registrations are concentrated in and around its leading tech hubs—Beijing, Shenzhen, Shanghai, and Hangzhou. Each city has its unique strengths: Beijing’s prestigious universities, national laboratories, and political influence provide an advantage in large-scale innovation; Shenzhen (in Guangdong) boasts a dense hardware supply chain and a vast talent pool; Shanghai, close to multinational corporations, excels in commercialization; and Hangzhou (in Zhejiang) thrives thanks to Alibaba’s e-commerce dominance.

However, innovation extends well beyond the coastal regions. Chongqing is establishing itself as an AI manufacturing and logistics center, while significant state investment has enabled Hefei, in Anhui Province, to earn the title “China’s speech valley” due to its cluster of speech-recognition companies, including iFlyTek. Filings are also emerging from less obvious areas such as Guizhou, known as “Big Data Valley,” where large data centers support Huawei’s Pangu model, and Inner Mongolia, where state enterprises are integrating AI into mining and agriculture.

*Data current as of April 2025

In the Trivium dataset, state-linked entries—which include state-owned enterprises and government-funded research institutes—account for 22 percent of all filings. Many state-affiliated companies collaborate with Big Tech to build their AI: for instance, PetroChina has partnered with Huawei and iFlyTek to develop oil and gas applications; State Grid used DeepSeek to create a model for optimizing power grids.

Foreign companies make up just 0.5 percent of the filings. Ikea, for example, developed a smart shopping algorithm that offers product recommendations. Yum China, the parent company for Kentucky Fried Chicken in China, filed a model that curates menus and promotional content.

Zeroing In on the Competition

*Data current as of April 2025

More than half of the entries in the algorithm registry are for what Schaefer describes as cross-sector technologies. These encompass foundational models, “general-purpose” text generators, and an array of multimedia tools—voice modulators, 3D renderers, and image creators. “No one wants to find themselves reliant on a competitor’s technology,” Schaefer states. Unlike in the US, where entities like OpenAI, Anthropic, and Google DeepMind dominate the landscape, competition in China is diverse and robust when it comes to foundational AI. However, developing these models is expensive, and the market is starting to consolidate. China’s six “AI tigers”—Moonshot, Minimax, Zhipu, Baichuan, 0.1AI, and Stepfun—are all backed by Alibaba or Tencent. Although ByteDance’s Doubao has overtaken DeepSeek as China’s most popular chatbot, its position at the top is not guaranteed.

Niche Natives

As the giants compete for chatbot dominance, startups are diligently working across numerous sectors.

Squirrel AI 松鼠

Founder Derek Li claims his 12-year-old company is far ahead of its ed-tech rivals. They have effectively “put wheels on a horse,” integrating AI into their previously stagnant software. Squirrel asserts it can identify knowledge gaps, track progress, and adapt lessons in real time.

Following China’s ban on for-profit tutoring in 2021, the company saw its revenues plummet overnight. It subsequently pivoted to licensing its platform to franchisees who also marketed the company’s AI-enhanced tablets. Squirrel’s network now comprises over 3,000 centers across China, catering to 1.2 million students. The company is currently contemplating an expansion into the US.

Li, who withdrew his sons from a private school in Shanghai to homeschool them using Squirrel’s platform, envisions a future where “teachers won’t teach knowledge.” Instead, he believes, “they’ll become data analysts who interpret learning reports and student abilities, and psychologists who understand emotions and shape personalities.”

AI Kanshe 看舌

AI Kanshe (translated as “AI Sees Tongue”) is a startup in traditional Chinese medicine that assesses health through images of the tongue, palms, and face. Founded by Li Wenhua, a former employee of Yaoshi Bang, one of China’s pioneer online pharmaceutical platforms, Li aimed to merge traditional diagnostic techniques with modern machine vision. The company services both consumers and healthcare professionals across clinics, pharmacies, and select hospitals, providing tools that aid in diagnosis and decision-making. Its model draws from over 100,000 annotated images of tongues, hands, and faces.

Zhongtan Puhui Cloud Technology 中碳普惠云科技

Established in 2024 by Wu Song, a former quant trader from Wall Street, Zhongtan Puhui Cloud Technology specializes in AI-driven tools for carbon accounting. Wu comments that the green transition still heavily relies on tedious human labor that could be automated.

Zhongtan Puhui develops AI agents that manage various carbon accounting tasks, including carbon footprint assessments and emissions audits. Its clientele spans from China Minmetals Group and DHL to small and medium-sized exporters in the Yangtze River Delta.

https://in.linkedin.com/in/rajat-media

Helping D2C Brands Scale with AI-Powered Marketing & Automation 🚀 | $15M+ in Client Revenue | Meta Ads Expert | D2C Performance Marketing Consultant

You may also like