The World AI Cooperation Organization Is Born — 29 Nations Build a New Order for AI Governance

On July 16, 2026, 29 countries signed the agreement establishing the World AI Cooperation Organization (WAICO) in Shanghai — with no United States, no major EU powers, and no Japan or South Korea. This is not merely the birth of another international organization, but the institutional starting point of "another option" within the global AI governance order. From the geopolitical configuration of the 29 founding members, to the technological shockwave of Kimi K3 released the very same day, to the triple echo one week later from Le Monde, the Associated Press, and Abu Dhabi — the founding of WAICO is an institutional experiment still unfolding.

On July 16, 2026, a new international order laid its first cornerstone in Shanghai. The agreement establishing the World AI Cooperation Organization (WAICO) was signed in Shanghai — representatives of 29 countries put pen to paper, and UN Secretary-General António Guterres attended the ceremony. This was no ordinary international-conference signing; it touched three planes at once: the institutional supply of global AI governance, the contest over rules in the US–China tech race, and the self-positioning of "Global South" countries within the technological order.

Twenty-Nine Signatures and One Absence

The 29 founding member states span a broad geography, from Southeast Asia (Laos, Pakistan, Indonesia) to Central Asia (Kazakhstan), Europe (Russia), Latin America, and Africa. In a follow-up commentary, Sputnik deliberately highlighted the geographic sweep "from Latin America to Southeast Asia," noting that "China and Russia have become the leaders of this agreement."

Set against this is the absence of the United States and its principal allies — no United States, no major EU powers, no Japan, no South Korea, no Australia. This was not a demonstration of "global consensus," but the self-organization of "the other half of the world."

This dichotomy clarifies WAICO's essential character — it is neither a "technical standards body" nor an "academic exchange platform," but a political choice about AI development. What the signatory countries share is not a common understanding of AI technology, but a common dissatisfaction with their place in the existing AI industrial chain — a refusal to remain forever users of APIs and consumers of closed-source models.

Headquartered in Shanghai: A Signal in Tech-Geopolitics

The agreement specifies that WAICO will be headquartered in Shanghai, China — no random choice. Signed on the eve of the opening of the World Artificial Intelligence Conference (WAIC 2026), Shanghai's dual role across these two events — host and institutional supplier — conveys a clear message: China is offering the world an option for AI governance that is independent of the US tech giants.

The appeal of that option is borne out empirically by the signatory list: 29 countries — a number larger than the founding membership of many international organizations. Kazakhstan's Deputy Prime Minister and Minister of Artificial Intelligence and Digital Development, Laos's Deputy Prime Minister and Minister of Foreign Affairs, Pakistan's Deputy Prime Minister and Minister of Foreign Affairs, the Deputy Chief of the Russian Presidential Executive Office, Indonesia's Coordinating Minister for Economic Affairs — the concentration of senior representatives in attendance shows that the signing was not a ceremonial photo opportunity, but a substantive commitment authorized by each country's decision-makers.

The AI-Sovereignty Consciousness of the "Global South"

WAICO's 29 member states are predominantly developing economies. Precisely because these countries sit on the receiving end of the current AI industrial chain — users of products rather than owners of technology — their need for "another order" is most acute. Sputnik called it "an important dimension of dispelling the remnants of colonial dependence" — a phrasing that points directly at a deeper structural problem: if the means of AI production (compute, data, talent, frameworks) are monopolized by a handful of companies and countries, technological dependency will translate into political subordination.

Seen from this angle, the significance of WAICO does not lie in its ability to immediately alter the balance of power in the global AI industry — most of the 29 member states lag far behind the United States in accumulated AI technology. Its significance lies in providing an institutional anchor — a framework that latecomers can join.

Three Events on One Day: The "Shanghai Moment" of AI Governance

Placing several events of July 16 side by side reveals an interesting superposition: the signing of the WAICO agreement (construction of an institutional framework), the arrival of Thai Prime Minister Anutin Charnvirakul in Shanghai to attend WAIC 2026 (a diplomatic move), and the disclosure of DeepSeek's valuation (350.8 billion yuan — a market validation of a Chinese AI firm). These three things happened on the same day, each pointing to a different layer of China's AI strategy: the institutional, the diplomatic, and the industrial. The synergy among them is not accidental, but the accumulated result of China's sustained push in the field of AI governance.

Echoes on Opening Day — Xi Jinping's Keynote and a New Round in the AI Industry

The day after the signing of the WAICO agreement, the 2026 World Artificial Intelligence Conference officially opened in Shanghai. On the morning of July 17, Xi Jinping delivered a keynote address at the opening ceremony, attended by Kazakh President Kassym-Jomart Tokayev, Cambodian Prime Minister Hun Manet, Thai Prime Minister Anutin Charnvirakul, and UN Secretary-General António Guterres. On the same day, the conference released a Chairman's Statement (15 articles), forming a complete policy framework running from technological ethics to global governance.

Four Questions of the Era, and a Four-Layered Answer

Xi's speech opened with a framing set of questions: "When machines begin to think, how should humanity get along with them? When algorithms take part in decision-making, how can safety be guaranteed? When technology challenges ethics, how can governance keep pace? When the divide keeps widening, how can inclusiveness be realized?" — these four questions point respectively to the humanistic, security, ethical, and distributive dimensions of AI development.

The four proposals put forward in the speech form the skeleton of China's AI-governance platform: openness and win-win (encouraging open-source openness), safe and controllable (building a system of regulations and technological monitoring), inclusiveness (shaping AI values with the common values of all humanity), and solidarity (practicing multilateralism and helping the Global South strengthen capacity building). The structural logic of these four points runs: innovation-driven (the development line) → a safety baseline (the risk line) → mutual learning among civilizations (the value line) → global governance (the institutional line).

Concrete Public Goods

Xi announced three quantifiable, concrete initiatives: over the next five years, 5,000 specialized AI training slots for developing countries; the construction of International AI Application Cooperation Centers for ASEAN, the Arab League, the African Union, CELAC, the SCO, and BRICS; and the rollout of the intelligent meteorological early-warning solution "Mazu" in 30 countries. The 5,000 training slots are "giving someone a fish" — short-term training; the cooperation centers are "teaching someone to fish" — building a network for capacity building; "Mazu" is "bulk delivery" — an already-packaged, scalable AI application. Together, the three constitute public-goods supply on different time scales.

The 15-Article Framework of the Chairman's Statement

The Chairman's Statement released by the conference covers 15 dimensions of AI governance. Several particularly operational articles are worth highlighting:

• Article 4, on data: emphasizes the principle of "manageable, controllable, and traceable," with property rights, circulation, and security pursued on three parallel tracks
• Article 8, on the protection of critical infrastructure: requires globally leading AI enterprises to install "necessary guardrails" in their models
• Article 9, on the governance of intelligent agents: clarifies decision-making authority and behavioral boundaries, and establishes a traceability mechanism — this is the first multilateral statement targeting the emerging risks of AI agents
• Article 13, on the Sustainable Development Goals: links AI capacity building to the 2030 Agenda

Industrial Signals Outside the Venue — Kimi K3 and a New Graduation on the US–China AI Race

On the same day WAIC opened, China's AI industry produced a heavily weighted data point: Kimi K3, released by Moonshot AI, with a parameter scale of 2.8 trillion, topped the Arena's front-end code leaderboard, overtaking Claude Fable 5 and GPT-5.6 Sol.

The industrial significance of this breakthrough needs to be understood on three planes:

Clearing the parameter hurdle. 2.8 trillion parameters is the largest scale among open-source models in the world today. Kimi K3 natively supports visual understanding and a 1-million-token context window — this is not an incremental improvement, but a pushing of the open-source-model capability ceiling to a position previously reachable only by closed-source models.

Anchoring the price. Kimi K3 costs roughly US$12 per million tokens, cheaper than the several high-end US models it challenges. Axios noted the point: "As Chinese firms keep delivering frontier models at lower cost, how long can US AI companies keep selling their services at premium prices?"

The cognition of mirror-image divergence. On the same day, Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, released its first model, Inkling. Forbes pointed out the rhetorical asymmetry: Inkling largely adopts the DeepSeek-V3 architecture and was trained on synthetic data distilled from Chinese open-source models such as Kimi K2.5 — yet "no one calls it theft."

This mirror image produced an even larger shockwave in the Korean direction. SK Hynix plunged 13.69% in a single day, and SpaceX closed below its issue price for the first time since its IPO — the market began to reprice the China risk (or China opportunity) of the "AI bubble."

The Agent Initiative — Standards Competition on a Parallel Track

Outside the WAICO institutional framework, another important tech-governance document was born on July 17. China's Cyberspace Administration released the Global Cooperation Initiative on Trust, Interconnection, and Interoperability of Intelligent Agents at WAIC's main forum. Composed of 10 principles covering technology R&D, mutual recognition of standards, joint construction of infrastructure, baseline governance of safety, and data flows, the initiative travels a different institutional path from WAICO: it does not involve the signing of inter-state agreements, but focuses on the technical standards and technical ethics of multi-agent collaboration.

One Week Later — Three Perspectives: Le Monde, the Associated Press, and the Abu Dhabi Desert

One week after the signing of the WAICO agreement, three media outlets from different standpoints looked back at the conference from different angles, assembling a more complete picture: what China's AI strategy is changing is not merely the technological competition landscape, but the basic coordinate system of global tech politics.

Le Monde: How Europe's Regulatory Advantage Is Offset by China's Cost Advantage

An in-depth analysis published by France's Le Monde on July 22 is perhaps the most systematic external reading of China's AI strategy from a European perspective to date. Its core judgment: what China displayed at WAIC in Shanghai was not a "progress report," but an integrated strategy aimed at reshaping the global technological landscape.

Le Monde sketches a triangular game among three actors:

• The United States — imposing rules through technological dominance, the extraterritorial application of law, and "kill switches." Export restrictions on Anthropic's most advanced model are the latest case of this pattern.
• China — taking a different path. Le Monde notes that China is following an industrial strategy already proven in practice: whether in solar panels, batteries, or electric vehicles, the goal is not to produce the most advanced technology, but to produce the most ubiquitous technology. The models developed by DeepSeek, Moonshot AI, and Zhipu AI "approach the performance of leading US firms at a fraction of the cost." For most of the world, "the balance of quality and price is likely more attractive than marginal gains in absolute performance."
• Europe — caught in the middle. Europe once believed it held a lead in regulatory standards — that its ambition to set ethical standards would make up for its industrial lag. But Le Monde's conclusion is sharp: "Beijing is threatening to neutralize this advantage — not through better regulation, but by combining competitive technological offerings with a political message espoused by the state." Having failed to carve out a genuine "third way," Europe is squeezed between two models — "possessing neither the United States' power nor China's appeal."

The Associated Press: The Depth of the Numbers

In a July 22 report, the Associated Press provided concrete magnitudes for WAIC 2026: more than 1,100 companies attended the July 17–20 conference, and Chinese firms have developed more than 400 humanoid-robot products, accounting for more than half of the world total. Four hundred humanoid robots — this figure is an important structural indicator. It means that China's deployment in humanoid robotics is not a single-point breakthrough by a few leading enterprises, but covers the entire product bandwidth from industrial to service to specialized applications. While global competitors are still debating "when will humanoid robots commercialize," China has already entered the "variety-filling" stage.

On the Dunes of Abu Dhabi: Strategy Lands

That same day (July 22), a video still showed Chinese robots building a solar farm in the Abu Dhabi desert. The value of this footage lies not in the technology demonstration itself, but in showing the overseas landing form of China's AI/robotics strategy: not weapons exports, not equipment sales, but the undertaking of infrastructure construction in a third country. Robots building a solar farm in the Abu Dhabi desert answer the question Le Monde posed — "what is the appeal of China's AI strategy?" — better than any slide deck or exhibition booth. The answer: it gets the work done, and it is cheaper than the alternatives.

Expansion of the Signatories — A Third Wave of Entrants in Late July

In the final two days of July, new names were added to WAICO’s founding roster. On July 31, Assistant Minister of Foreign Affairs Liu Bin met collectively with the envoys to China from Brunei, Togo, Iran, Sudan, and Vietnam, witnessing five countries sign the agreement. That same day, the ambassadors of Georgia and Tanzania signed on behalf of their governments, and on July 30, the Dominican ambassador to China signed. Counting the first batch of July 16 — China, Russia, and more than 25 other countries — the roster of founding members keeps expanding beyond the original twenty-nine.

The composition of this list is worth reading twice. Brunei and Vietnam are states ringing the South China Sea; Iran and Sudan are consequential players in the Middle East and Africa; Georgia sits at the hinge of Eurasia; and Tanzania and Dominica represent East Africa and the Caribbean, respectively. The geographic sweep runs from East Asia, Southeast Asia, the Middle East, and Africa all the way to the Caribbean — the Global South’s coverage has widened another ring beyond the twenty-nine-country list unveiled at the July 16 founding ceremony.

The pace of the expansion is itself a signal. Less than half a month after founding, a third batch of signatories has landed — evidence that the agreement’s signing channel stays permanently open rather than functioning as a one-off ceremony. For observers, this confirms the earlier judgment: WAICO is traveling a route of “intergovernmental multilateral mechanism plus co-builder identity,” and every additional signature adds another increment of network effect to the track. Set against the governance frameworks led by Western tech giants, these are two entirely different modes of growth.