Editor's note: Liu Ming is an assistant researcher at the Department of Economic Forecasting, State Information Center. The article reflects the author's opinions and not necessarily the views of CGTN.

In 2024, the UN General Assembly adopted by consensus a China-sponsored resolution on international cooperation in AI capacity building. The Global AI Governance Action Plan was released; and in July this year in Shanghai, 29 countries signed the agreement to establish the World AI Cooperation Organization. The initiative has moved from concept to action, and it is driving a deeper change in how the value of AI is distributed.

Open source ecosystems unlock the source of value creation

Today, AI value is concentrated in a few closed-source giants, and platform companies collect high rents by controlling ecosystem gateways. The Chinese approach instead supports a thriving open source ecosystem. Open weights lower barriers and put model capabilities in the hands of more developers. 

China-developed open-source AI models have passed 10 billion cumulative downloads globally, AtomGit has topped 11 million users, and China-led models took 17.1% of Hugging Face downloads in the year to August 2025 — the top country share.

Open source also lowers the cost of access. OpenAI GPT-5.5 charges 55 times as much as DeepSeek V4 Flash for input and 166 times as much for output, and on OpenRouter the share of tokens processed by open source models rose from 34% in January 2026 to 65% in June. Value creation is shifting from monopoly by a few firms to shared creation worldwide.

Capacity building reverses the one-way flow of value

The Global South has long served as a supplier of raw data. Data flows out cheaply while digital products flow back at high prices. The Chinese approach favors teaching people how to fish, treating capacity building as the fundamental way to bridge the intelligence divide. China has released the AI Capacity Building Action Plan for Good and for All, which sets out 10 actions under five visions, and will offer 5,000 AI training opportunities to developing countries over the next five years. When talent stays local and models are deployed locally, the added value of data processing also stays local, and value between North and South moves from one-way extraction to mutual benefit.

Regional cooperation turns technology into local gains

Many developing countries remain locked at the low end of global value chains with weak technology spillovers. The Chinese approach uses regional cooperation to link technology supply with local demand. China is building international AI application cooperation centers with ASEAN, the Arab League, the African Union, CELAC, the Shanghai Cooperation Organization and BRICS, and its Mazu intelligent weather warning system is in use in 30 countries. 

China has cumulatively provided African partners with over 70,000 free high-resolution remote sensing images, delivering robust spatial information support for their development planning. In Takeo Province, Cambodia, a smart farm project has raised annual income per hectare from about $1,000 dollars to over $3,000. Value stays and circulates within the region.

Multilateral mechanisms keep the rules fair

How value is shared depends on who writes the rules. When a few countries set the rules, a few companies take the gains and every country bears the risks, the arrangement cannot last. The Chinese approach calls for true multilateralism with the United Nations as the main channel, so that all countries enjoy equal rights, opportunities and rules in AI. As the first intergovernmental organization dedicated to AI, the World AI Cooperation Organization is open to all countries. The Action Plan for AI Cooperation and Development released at the July conference proposes eight actions, including inclusive computing power, shared open-source ecosystems, and joint standards. It addresses a common problem in the Global South — where advanced computing power cannot be bought, local language data is too costly to train on, and regulations are hard to draft. The Global South is now entering the rule-making of the intelligent economy and gaining its rightful place in the new division of labor.

The Chinese approach is turning the ideas of putting people first and AI-for-good into productivity gains that can be shared. As open source models become global infrastructure, the value of AI assets will depend more on real output from application scale, industrial adoption and better livelihoods. Whoever brings the benefits of technology to more countries and more people will lead the next stage of value distribution. Through open sharing, China is helping this new order take shape.