China Builds an AI Coalition With Open Models, Training and a New Global Organization

A Shanghai Agreement Gives Beijing a New AI Platform

Twenty nine countries signed an agreement on July 16, 2026, establishing the World Artificial Intelligence Cooperation Organization, or WAICO, with its headquarters planned for Shanghai. A day later, President Xi Jinping opened the World Artificial Intelligence Conference and the High Level Meeting on Global AI Governance with a call for a fairer international system.

China's commitments include 5,000 AI training and seminar opportunities for developing countries over five years, international cooperation centers and plans to promote its MAZU meteorological early warning system in 30 countries. Together, the organization and those programs give Beijing a diplomatic platform and practical proposals for countries seeking a larger role in artificial intelligence.

Xi placed developing countries at the center of his message, connecting access to computing, training and useful applications with international cooperation on the rules governing AI. He described the purpose of that cooperation in his opening speech.

All countries should take a people centered approach and develop AI for the positive and for good," Xi said. "We should ensure that AI is an important driver for shared prosperity and common security.

WAICO's founding members include China, Russia, Brazil, Indonesia, Malaysia, Pakistan, South Africa and Kazakhstan. The United States, European Union member states, Britain, Japan and other major Western economies are absent. India attended the Shanghai conference but was not among the organization's founding members.

The July developments bring together China's efforts to expand access to AI and its ambition to shape international governance:

• 2023: Xi proposed China's Global AI Governance Initiative.
• July 16, 2026: Twenty nine countries signed the agreement establishing WAICO.
• July 17, 2026: Xi opened the Shanghai conference and the High Level Meeting on Global AI Governance, presenting China's cooperation commitments.

Washington and its partners have concentrated on trusted supply chains, advanced chip controls and safety standards. Beijing is presenting open source models, affordable computing and development assistance as an alternative for countries with limited access to the most advanced systems.

What Powers Will WAICO Have?

WAICO is intended to promote international cooperation on AI development and governance. Its stated principles include a people centered approach, safety, fairness, shared benefit and respect for the United Nations Charter.

The organization is still at an early stage. As of July 21, public information does not fully explain how members will vote, how WAICO will be funded, which commitments will be binding or how decisions will be enforced. A charter and headquarters provide an institutional base; the ability to make governments follow common rules depends on the authority members give the organization.

WAICO's immediate role may therefore be diplomatic rather than regulatory. Developing nations could use it to coordinate positions before discussions at the United Nations and other international forums. China would also gain a channel for promoting standards connected to Chinese technology, training programs and infrastructure.

Humphrey Arnaldo Russel, head of the ASEAN China Research Center at the University of Indonesia, said AI brings challenges that cross borders. He identified safety, ethics, misuse, technical standards, fragmented regulation and unequal access to digital resources.

Russel said China's Global AI Governance Initiative could bring developing countries into discussions often led by wealthier economies. He described the potential role of a framework centered on the United Nations.

"The proposal is especially important for the Global South, whose interests remain underrepresented in global technology governance. A broadly accepted, United Nations centered framework could strengthen developing countries' participation in global AI governance while respecting different national conditions," Russel said.

WAICO's influence will depend on the uses its members make of it. Coordinating common standards would give it a different role from serving mainly as a route to Chinese technology. Practical programs and meaningful participation in decisions will be central to whether it becomes more than a discussion forum led by China.

Training and Weather Warnings Lead the Development Offer

Countries across Africa, Asia and Latin America face a basic imbalance in the AI race. Many have large populations, growing digital economies and urgent needs in health, agriculture, education and disaster response, while lacking the data centers, advanced chips, researchers and funding required to build major AI systems.

Xi's pledge of 5,000 training and seminar opportunities over the next five years addresses part of that gap. China also plans international AI application cooperation centers with ASEAN, the League of Arab States, the African Union, the Community of Latin American and Caribbean States, the Shanghai Cooperation Organization and BRICS.

The centers are intended to support cooperation on practical uses of AI. Their importance for participating countries will depend on whether the partnerships develop skills and institutions that can continue working locally, alongside access to foreign technology.

China's proposed expansion of MAZU offers a concrete application. The meteorological early warning system combines satellite observation, weather forecasting and automated warnings to help governments prepare for floods, droughts and other extreme weather. China plans to promote its use in 30 countries.

Philip Okello, principal meteorologist at the Kenya Meteorology Service, said the proposal fits Africa's need for stronger forecasting systems as climate risks become more frequent. He connected Xi's speech to the pressures countries already face.

"President Xi's speech is timely because climate change is no longer a future risk; it is a present day reality that is affecting countries across the globe," Okello said.

Training places, cooperation centers and MAZU deployment give partner governments specific commitments to assess. The value of those commitments will depend on how effectively people can use the technology in public services and how much expertise stays within participating countries.

Access Is Only Part of Building Local Capability

Gordon K'achola, founder and chief executive of the Africa Center for Diplomatic Affairs, said African countries want the ability to create, manage and adapt AI systems, alongside access to finished products. He described stronger local research and institutions as central to that goal.

"For Africa and the broader Global South, the goal is not just getting the latest tech. It is about building the capacity to create and control their own AI writing algorithms, growing research centers, and strengthening local institutions," K'achola said.

A country importing an AI service may gain benefits quickly while continuing to rely on foreign companies for updates, data processing and technical support. Training researchers, building computing facilities and developing local models give that country greater control over how government and businesses use AI.

Computing access and training can widen participation, but lasting independence also requires local rules for handling data, research institutions, technical standards and public oversight. Those capabilities determine how well a country can adapt systems to its own needs and make decisions about their use.

China's proposal will be judged partly by whether its partners gain that ability. The development offer is attractive because it combines technology with assistance, yet partner countries still need sustained investment and institutions capable of directing the work.

Why Open Models Are Central to China's Pitch

China is presenting open source and open weight AI models as ways to reduce adoption costs. Open source systems make parts of the code or model available for inspection, modification and reuse. Open weight models release the trained parameters that allow a system to generate results, although they may leave details of the training data and process undisclosed.

DeepSeek, Alibaba and Moonshot AI have gained attention with models that are often cheaper to run than leading proprietary systems from American companies. Proprietary systems remain under the control of their owners, while the access offered by open models can give other developers more room to adapt and reuse them.

Moonshot introduced its Kimi K3 model during the conference and described it as the largest open AI model by parameter count. Parameters are the trained values used by a model to produce results. Their number measures one aspect of a model's scale; performance also depends on how the system is designed and trained.

Affordable models can help Chinese firms attract users, developers and institutional partners across emerging markets. That commercial expansion also has a geopolitical dimension. Once businesses build products around a model, cloud platform or hardware system, moving to a competitor can become costly.

Maria Sun, co chair of the FIRA RoboWorld Cup 2026, said international exchanges could help more countries participate in AI development. She welcomed China's cooperation efforts.

"China's efforts to promote AI cooperation with developing countries are very encouraging. International exchanges like these help bring more countries into the AI era," Sun said.

Sun also described China as an important global AI player whose openness could support research and cooperation. China's approach combines international access to selected models and applications with national security rules that may limit foreign access to some advanced systems. Beijing criticizes American restrictions on advanced chips and AI technology while maintaining its own controls over data and technology.

Indonesia Seeks Investment and Technology Transfer

Indonesia's WAICO membership illustrates the appeal of China's offer. With a population of more than 280 million and a fast growing digital market, the country wants foreign investment while developing its own research and engineering capacity.

Indonesian officials have sought cooperation with Chinese companies in cloud computing, robotics, telecommunications, health care, agriculture and fisheries. Chinese firms have expanded data center and cloud services in Southeast Asia, allowing local businesses to use computing resources without building expensive facilities themselves.

Indonesia also wants technology transfer, university cooperation and local laboratories. Its ambition is to become a producer of digital services, rather than remain primarily a large market for applications developed abroad.

Shortages of AI researchers, engineers and investment make that task difficult. Indonesia's national AI policy remains divided among ministries, and many businesses are still experimenting with AI instead of using it in daily operations.

A clear policy framework and sustained investment in local skills will be necessary for Indonesia to direct those partnerships toward its own goals. Without them, foreign cooperation could leave the country dependent on systems developed elsewhere. The same concern applies across the Global South, where access to cloud services can advance faster than domestic research and oversight capacity.

The Competition With Washington Extends Beyond Models

Xi's appeal for openness also carried a message to Washington. He urged countries to oppose what he called the "overstretching" of national security concerns in AI and objected to putting one country's security above that of others. He did not name the United States, although his remarks were widely understood as criticism of American restrictions on advanced chips and other technologies.

The United States and China are competing over computing power, supply chains, technical standards, research talent and international partnerships, alongside model performance. American companies retain major advantages in private investment and advanced chip supply.

China has narrowed the performance gap between its strongest models and leading American systems. It has also gained strength in research output, patents, industrial robotics and open models with lower costs.

Washington has gathered countries around initiatives focused on AI opportunity, secure supply chains and cooperation among trusted partners. Beijing is seeking support from BRICS, ASEAN, African and Arab states, Latin American countries and members of the Shanghai Cooperation Organization.

India's attendance at the Shanghai conference, alongside its absence from WAICO's founding membership, makes it a key country to watch as both sides compete for support. Countries participating in Chinese cooperation may also seek to preserve relationships with other technology providers.

David Tuffley, an adjunct senior lecturer at Griffith University in Australia, said a people centered approach addresses the risk of AI gains becoming concentrated among a small number of companies and countries. He connected wider access to a greater voice in setting rules.

"For the international community, particularly developing countries, the commitment to capacity building and greater access to computing power is especially significant, as much of the world's population currently has little voice in shaping the rules that govern AI," Tuffley said.

Tuffley added that Australia and China share an interest in AI safety research, standards for testing advanced systems and preserving human judgment in health care and education. Those areas offer reasons for cooperation even as the wider technology contest continues.

Safety, Rights and Resources Remain Central Concerns

China's approach has received support from some international officials and developing country representatives, alongside criticism from analysts who question Beijing's motives and governance record.

Critics argue that China's call for openness partly reflects its need for access to global models, chips, research and talent while reducing the effects of American controls. They also warn that Chinese participation may bring standards that put state authority and information control ahead of privacy, political rights and independent oversight.

Developing countries still face practical choices about cost and access. Many cannot afford the most expensive Western systems and may prefer partners that provide hardware, software, training and infrastructure together. Governments must find workable ways to use AI for public services and economic growth while deciding which safeguards their partnerships require.

Stephen Jackson, the United Nations resident coordinator in China, has said AI can support every Sustainable Development Goal, from food security and health care to climate response. He has also warned that it can widen inequalities between countries and within them.

Jackson has pointed to cases in which Chinese satellites and AI analysis produced disaster maps for humanitarian teams within 48 hours. Sharing data and tools during emergencies can make AI a public resource, helping teams understand where assistance is needed.

AI systems also require large amounts of electricity and computing capacity. A fairer international system will need investment in energy, connectivity and digital literacy, alongside safeguards against misuse.

Governance must address privacy, cybersecurity, labor disruption and automated systems used in sensitive decisions. Wider access and stronger safety rules are connected tasks: governments need the resources to use AI and the institutions to oversee how it affects people.

Delivery and Member Influence Will Determine WAICO's Role

WAICO's first test is whether its founding agreement becomes a functioning institution with a secretariat, budget and funded projects. Its next test is whether members can influence decisions rather than simply endorse China's proposals.

The planned cooperation centers, deployment of MAZU in 30 countries and 5,000 training opportunities provide measurable targets. Governments can assess whether those programs create durable local expertise, while deciding how Chinese models fit their laws on data, privacy, national security and public procurement.

The United Nations is likely to remain the main forum for rules with global legitimacy. WAICO could work alongside that process or become a separate bloc promoting standards shaped largely by Beijing.

Indonesia, Brazil, South Africa and Kazakhstan may seek benefits from Chinese cooperation while resisting pressure to choose sides in the wider contest between the United States and China. Their ability to preserve choice will matter as technology controls and geopolitical rivalry divide access to systems and infrastructure.

China's bid combines practical assistance with a vision of international governance centered on sovereignty, development and cooperation between states. Supporters see a route to reducing the digital divide. Critics see an effort to export Chinese norms and gain influence over future standards.

The results will depend on what partner countries can build with the assistance, how transparent WAICO becomes and how much influence its members have. Training, affordable models and shared applications can expand access; durable local skills and meaningful decision making will determine the depth of that expansion.

Key Points

• Twenty nine countries signed WAICO's founding agreement on July 16, 2026, with its headquarters planned for Shanghai.
• Xi Jinping called for fair, people centered AI governance at the Shanghai conference on July 17.
• China pledged 5,000 AI training and seminar opportunities over five years and cooperation centers with major international groupings.
• China plans to promote its MAZU weather warning system in 30 countries.
• Open source and open weight Chinese models are central to Beijing's offer of more affordable AI access.
• Indonesia seeks investment, technology transfer and local research capacity while facing shortages of skills and funding.
• WAICO's voting, funding, binding commitments and enforcement arrangements remain insufficiently explained publicly.
• The organization's influence will depend on practical delivery, transparency and meaningful participation by its members.