Chinese President Xi Jinping opened the World Artificial Intelligence Conference in Shanghai on Friday, July 17, 2026, with a pointed call for multilateral AI governance — and an unmistakable rebuke of American-led technology restrictions. The speech, delivered to an audience that included the U.N. Secretary-General and the heads of state of Kazakhstan, Cambodia and Thailand, was more than ceremonial; it signalled Beijing’s intent to build an institutional architecture for AI that operates independently of Washington’s influence.
What Happened
Speaking at the opening ceremony of China’s annual World Artificial Intelligence Conference, Xi framed AI development as an endeavour that “should not be a solo performance by any single country but rather a symphony of global cooperation.” He specifically cautioned against what he called the “overstretching” of national security concepts — diplomatic shorthand for Washington’s export controls on advanced semiconductors and AI hardware that have curtailed China’s access to leading-edge computing technology.
Xi announced a series of concrete commitments aimed at the Global South: China will offer 5,000 AI training opportunities to developing nations over the next five years, and will share access to a Chinese-built AI meteorological tool — capable of providing early-warning disaster alerts — with 30 countries. He also pledged to expand AI cooperation with ASEAN, the Arab League, the African Union, the Community of Latin American and Caribbean States, the Shanghai Cooperation Organization and the BRICS grouping.
The most structurally significant development, however, came before the formal ceremony. Twenty-nine countries — including Russia, Pakistan and Kazakhstan — signed an agreement with China to establish the World Artificial Intelligence Cooperation Organization (WAICO), an intergovernmental body to be headquartered in Shanghai. Chinese state media described the organisation as dedicated to promoting global AI governance.
U.N. Secretary-General António Guterres also addressed the conference, lending the gathering a degree of international legitimacy that Beijing will likely amplify in its communications. His presence does not imply U.N. endorsement of WAICO, but it underscores the genuine appetite among multilateral institutions for a forum where AI governance can be discussed beyond the bilateral US-China frame.
Why It Matters
George Chen, partner and chair of digital practice at a consultancy cited in contemporaneous reporting, described WAICO as China’s answer to the U.S.-led “Pax Silica” initiative — a framing that captures the essential dynamic at play. The United States has spent the past several years building coalitions of like-minded democracies around shared AI standards and chip-supply controls. Beijing is now assembling a counter-coalition, and it is doing so through the institutional language of multilateralism rather than the bilateral dealmaking that characterises much of China’s diplomatic activity.
The strategic logic is legible. American export controls — including restrictions on Nvidia’s most advanced data-centre GPUs — have forced Chinese firms and researchers to develop domestic alternatives, a dynamic explored extensively in coverage of how China is challenging Nvidia’s AI chip supremacy. Those domestic efforts have yielded real results, as the emergence of models like DeepSeek demonstrated to a surprised Silicon Valley audience earlier this year. Xi’s speech at WAIC 2026 is, in part, a victory lap on that technological resilience — but it is also a forward-looking bid to set the terms of AI governance for countries that have no seat at the US-led table.
Read alongside Xi’s outreach to BRICS and ASEAN member states, the WAICO launch reveals a deliberate two-track strategy: harden China’s domestic AI stack against Western restrictions while simultaneously positioning Beijing as the natural partner for the roughly 100 nations that are neither close US allies nor close Chinese partners. If that middle tier of countries begins adopting Chinese AI infrastructure and governance norms — as the meteorological tool giveaway is designed to encourage — the US faces a world in which its own AI standards apply to a shrinking share of global compute and data flows.
This matters beyond geopolitics. AI governance is not abstract: it determines who decides how facial-recognition systems are deployed, what data-privacy standards large language models must respect, and whether AI-enabled surveillance tools are shared across borders. The fragmentation of AI governance into competing blocs would create a world in which the rules differ profoundly depending on which infrastructure a country has adopted — a structural challenge comparable to, but more consequential than, the divergence between GDPR and other data-protection regimes. Concerns about this kind of fragmentation are not new; analysts tracking the US-China AI race have long warned that the most durable risk is not a single dramatic breakthrough but a slow institutional bifurcation.
How China’s WAICO Compares to Western AI Governance Initiatives
To assess WAICO’s significance, it helps to place it alongside the governance frameworks already in circulation. The table below compares three initiatives on dimensions that matter to participating nations — bearing in mind that WAICO is newly formed and its operational details remain limited.
| Initiative | Lead Actor | Membership Basis | Primary Focus | Institutional Seat |
|---|---|---|---|---|
| WAICO (World AI Cooperation Org.) | China | 29 signatories (incl. Russia, Pakistan, Kazakhstan); largely Global South and SCO-aligned | AI governance, technology sharing, Global South inclusion | Shanghai |
| OECD AI Principles | OECD / Western democracies | 38 OECD members + adherents; predominantly developed economies | Trustworthy AI standards, human rights, transparency | Paris |
| Global Partnership on AI (GPAI) | G7 + invited partners | ~30 countries; democracies and like-minded states | Responsible AI research, working groups on data, future of work | Paris (OECD secretariat) |
| Note: WAICO operational details are based on state media reporting at launch. Comparative attributes for OECD and GPAI reflect publicly available frameworks. | ||||
The structural divergence is striking. OECD-aligned frameworks are built around democratic norms and are weighted toward developed economies. WAICO, by contrast, is designed to be attractive to countries that have historically felt excluded from technology governance discussions — and that have significant AI infrastructure needs China is well-positioned to meet. Whether WAICO produces substantive governance rules or remains largely a diplomatic vehicle will depend on the secretariat it builds in Shanghai and the binding commitments, if any, it extracts from members.
What Happens Next
The immediate question is whether WAICO graduates from a signing ceremony into a functioning institution. Intergovernmental bodies are notoriously slow to operationalise, and China’s track record with similar multilateral vehicles — from the Shanghai Cooperation Organization to the BRICS New Development Bank — is mixed: they produce communiqués, occasionally useful financing, and a diplomatic forum, but rarely the transformative rule-setting their founders advertise. Critics of Xi’s address will point out that China’s domestic AI governance is itself opaque and that Beijing’s record on technology transfer is driven by strategic interest rather than open-source generosity.
That said, the meteorological AI tool offer is worth watching closely. Sharing a working, early-warning system with 30 countries is a concrete deliverable that builds goodwill and, more importantly, creates infrastructure dependencies. Nations that integrate Chinese AI tools into their disaster-preparedness systems are likely to procure compatible hardware, adopt compatible data standards, and engage Chinese technical advisors — a pattern already visible in China’s export strategy for other technology sectors.
On the US side, Xi’s speech will add pressure on the Biden-successor administration — whichever party holds office — to articulate a more affirmative vision for AI cooperation with the Global South, rather than a purely restrictive one. Washington’s current posture, centred on export controls and ally-coordination, is effective at slowing China’s access to frontier chips but offers comparatively little to countries that primarily want AI capacity, not geopolitical alignment. The danger, as analysts studying China’s open-source AI advances have noted, is that the United States wins the semiconductor battle while losing the standards war.
There is also a question about how the technology industry itself responds. Companies operating globally — in cloud infrastructure, enterprise AI platforms and hardware — face an increasingly bifurcated regulatory and standards environment. The more that AI governance forks between a US-led bloc and a China-led bloc, the more costly it becomes to build products that work across both. That compliance burden, still modest today, could become a significant structural cost as both frameworks mature and diverge further.
For developers and data engineers tracking where AI norms are heading, the institutional signals coming out of Shanghai this week are as consequential as any product announcement — and considerably harder to parse than a benchmark score.
The Implications That Matter
- A parallel AI governance order now has institutional form. WAICO’s launch means the world has moved from rhetorical competition over AI norms to competing intergovernmental organisations — a structural shift that will be difficult to reverse regardless of WAICO’s operational effectiveness.
- The Global South is the decisive swing constituency. Both the US-led and China-led frameworks are competing for the loyalty of nations that have AI ambitions and limited capacity; whoever delivers working tools — not just principles — is likely to set the governance baseline for a large share of the world’s population.
- Infrastructure giveaways are governance instruments. China’s offer of meteorological AI tools and training slots is not purely altruistic; it seeds dependency on Chinese AI stacks, creating long-term alignment pressure on recipient countries that may matter more than any formal treaty commitment.
- US export controls risk strategic overreach. Restrictions designed to slow Chinese AI development may inadvertently accelerate the formation of a bloc that operates outside US-shaped standards altogether — a risk that policymakers in Washington will need to weigh against the genuine security rationale for controls.
- Technology companies face a widening standards gap. The longer AI governance diverges across geopolitical blocs, the more expensive it becomes to build globally interoperable AI products — a compliance and engineering cost that will ultimately affect innovation velocity everywhere.
