Chinese AI strengthening economic resilience

The technological and pricing shock brought by Kimi k3 to the large-language model ecosystem continues to reverberate through capital markets, with some US media outlets even describing it as a "DeepSeek Moment 2.0". It is widely believed within the industry that the six-to-12-month innovation gap between China and the United States in large models has been largely closed. Although product iteration may lead to periodic fluctuations in specific technical indicators, China's artificial intelligence innovation has remained among the global leading group, with its overall capabilities comparable to those of any leading AI power.

However, China's AI progress should not be assessed only by isolated benchmark scores. Its distinctive path lies in broader technological coverage, richer application scenarios and stronger empowerment of the macroeconomy. By extending AI across industries and into the real economy, China is reinforcing economic resilience and improving development quality across multiple dimensions.

The recent 2026 World Artificial Intelligence Conference, held in Shanghai, offers a vivid demonstration of China's broad-based AI innovation. With three major venues — Expo, Zhangjiang and West Bund — the exhibition covered more than 100,000 square meters and brought together over 1,100 domestic and foreign enterprises. More than 3,000 cutting-edge achievements were on display, with over 300 hard-tech products making their global debuts or industry premieres. These exhibits covered the full industrial chain, from chips and foundation models to intelligent robots and terminal devices, presenting a comprehensive picture of AI and the emerging intelligent society.

US economists originally defined AI as a technological revolution mainly in the services sector and as an extension of the internet services revolution, which led to relatively concentrated innovation tracks, limited technological penetration and expectations that the main impact would fall on mid and low-skilled white-collar workers. Yet China's AI practice has already moved beyond this early assessment, building a complete innovation chain that covers computing power, infrastructure, general-purpose large models and intelligent applications.

More importantly, China is accelerating AI's movement from the digital world to the physical world. The speed of this transition and the richness of application scenarios both rank among the highest globally. Coordinated innovation across industries and scenarios is reshaping global understanding of AI's application scope, innovation breadth and social significance. Meanwhile, diverse application scenarios can improve enterprises' digital foundations and technological adaptability, lower cross-sector innovation expenses, strengthen synergies among different fields, and continuously raise industrial technological capacity and total factor productivity.

China's technological innovation has always been anchored in the real economy, with the high-end transformation of manufacturing as its core direction. The recommendations for the 15th Five-Year Plan (2026-30) period make clear that China should build a modern industrial system supported by advanced manufacturing and accelerate high-level technological self-reliance. This requires sustained efforts to strengthen foundational manufacturing technologies, address weaknesses in basic industrial capabilities, upgrade traditional manufacturing, expand emerging manufacturing sectors and make forward-looking arrangements for future manufacturing. Taken together, these efforts are expected to systematically enhance manufacturing capacity and strengthen the ability to respond to risks.

Unlike the US path, which places greater emphasis on AI applications in service-sector innovation and substitution, China's AI development focuses more on connecting with the physical world and deeply integrating with real industries. Its core effort is to empower advanced manufacturing and connect the entire manufacturing chain. Relying on 520,000 industrial enterprises above designated size, 39 major industrial categories and 191 medium industrial classifications, China's complete industrial system can fully unleash AI's value as a general-purpose technology, accelerating full-domain penetration and forming a virtuous cycle of technological iteration and efficiency gains.

Practical cases are already abundant. Sinopec's "Fenghuo" industrial intelligence agent integrates more than 1 billion pieces of industry expert knowledge and achieves four-layer technological coupling across platform, model, digital twin and intelligent agent, helping reduce energy consumption per unit of output. Zhenhua Heavy Industries' welding defect recognition model has achieved an accuracy rate of over 98 percent, replicating the practical experience and image-recognition logic of senior craftspeople, making it a benchmark for "AI + manufacturing". Waigaoqiao Shipbuilding uses AI to streamline shipowner compliance verification, while Shanghai Turbine Works relies on intelligent algorithms to control blade production parameters more precisely. These examples show how AI is being embedded in large manufacturing enterprises.

At the policy level, the 15th Five-Year industrial internet action plan proposes that by 2028, the penetration rate of industrial internet platforms should exceed 55 percent and the total number of connected devices should surpass 120 million. The aim is to build a new-generation industrial internet ecosystem featuring ubiquitous connectivity, digital-intelligent integration, deep collaboration and open-source openness, thereby providing long-term support for the intelligent upgrading of the whole manufacturing chain.

China's AI ecosystem also contributes to global development by providing inclusive intelligent public goods.

Cost-effective open-source models are already benefiting global markets. Leading Chinese large models generally follow open-source and open-weight strategies, featuring intensive and efficient research and development models and low deployment costs. Their usage prices are only about one-tenth of comparable overseas products. Grounded in the principles of peaceful development and serving the people, these technologies provide affordable and easily deployable AI tools for developing countries. In essence, their low-cost model is strongly inclusive and has accelerated the global diffusion of large-model technologies.

Industry estimates show that Chinese open-source and open-weight large models are priced 60 to 90 percent lower than similar US products from OpenAI and Anthropic, while still covering most routine office and production needs. Their cost-performance advantage has made them popular among global tech start-ups, especially enterprises and consumers in developing countries and the Global South. Data show that US companies' use of Chinese AI models, measured by token share, averaged more than 30 percent per week and peaked at 46 percent.

Beyond market access, China is also actively supplying global AI public goods and promoting collaborative governance. China's AI development path balances industrial application with public value and actively participates in global AI governance. Following President Xi Jinping's proposal of the Global AI Governance Initiative, China has continued to position itself as a provider of international public goods in the AI field. It has released the AI Capacity-Building Action Plan for Good and for All and the "AI Plus" International Cooperation Initiative, while advocating for the establishment of a World Artificial Intelligence Cooperation Organization.

Inclusive measures are also being implemented globally. China has announced the construction of an international AI application cooperation center, promoted the wider deployment of the MAZU intelligent meteorological early warning solution in more countries, and offered 5,000 specialized AI training opportunities for developing nations. These efforts help AI technologies take root more quickly across countries, improve production and public services and strengthen developing countries' ability to manage natural disaster risks.

Ultimately, general-purpose AI can help countries address gaps in industry and scientific research, thereby strengthening overall economic resilience. This is an important expression of AI's general-purpose capability and a core value of China's AI development. Meanwhile, China is improving AI-related laws, regulations and supervisory systems, using both legal constraints and technical controls to prevent potential risks and ensure that AI always serves humanity and remains under human control.

The writer is a research fellow at the Institute of Economics, Chinese Academy of Social Sciences, and deputy director of the Center for Xi Jinping Economic Thought Studies at the Chinese Academy of Social Sciences.

The views do not necessarily reflect those of China Daily.