Chinese President Xi Jinping delivered his first in-person keynote at the World Artificial Intelligence Conference in Shanghai on July 17, an event China has hosted since 2018 but which the top leader had never personally headlined. The timing was not incidental. Xi’s address landed within hours of a prime-time broadcast in which US President Donald Trump attacked Beijing directly, and the two speeches read as a matched pair: one leader framing AI as a shared global project, the other treating it as a domain of national rivalry. The contrast was the point.
Xi’s message was built around a single organizing idea — that AI should not be the property of any one nation. He wrapped it in a Chinese aphorism about a single string being unable to make music and a single tree not making a forest, and called instead for what he described as a symphony of international cooperation. Underneath the rhetoric were three concrete moves: a doubling-down on open-source AI, the formal launch of a China-anchored governance body, and a five-year pledge to help developing countries build their own AI capacity.
What Xi Actually Committed To
The speech organized China’s position around a set of observations on AI development and governance. The first and most emphasized was openness: encouraging open-source models, cross-border collaboration, and shared research as the engine of both innovation and industrial application. The second was risk — Xi acknowledged the need to keep AI secure and controllable, a nod to the security concerns that have accompanied the open-weight model debate. He also reiterated Beijing’s standing objection to what he termed the overstretching of national security as a justification for restricting technology access, an unmistakable reference to US export controls.
The framing that drew the most attention was his characterization of unequal AI access as a historical injustice. Xi argued that closer international partnership could prevent the world from repeating patterns in which advanced technology concentrates in a handful of wealthy states. To back the claim with policy, he announced a five-year program of Chinese support for global AI capacity building, aimed squarely at the developing world.
The Strategic Logic of Open Source
China’s embrace of open-source AI is not charity, and the speech did not pretend otherwise. The commercial and strategic case is straightforward. The poster child is DeepSeek, whose models demonstrated that competitive performance could be reached at a fraction of the capital that leading US labs deploy. By positioning Chinese open-weight models as global public assets rather than proprietary products, Beijing accomplishes several things at once: it accelerates domestic development through external contribution, it seeds Chinese systems and standards into the infrastructure of other countries, and it offers a workaround to the US export-control regime that has restricted China’s access to the most advanced chips.
There is also a standard-setting dimension. Whoever supplies the base models that developers around the world build on gains influence over the norms, interfaces, and assumptions baked into the next generation of AI systems. Open source, in this reading, is less a philosophy than a distribution strategy for influence. Huawei’s presence at the conference underscored the hardware side of the same logic, with the company showcasing a high-end AI computing system pitched as an alternative to Western accelerators.
WAICO and the Global South Play
The concrete institutional output was the formal launch of the World Artificial Intelligence Cooperation Organization, an intergovernmental body to be headquartered in Shanghai. Ahead of the conference, 29 countries — including Brazil, Russia, Pakistan, and Kazakhstan — signed on. Xi described the organization as a milestone in the history of world AI development and cast it as an answer to Global South countries that have long wanted a larger voice in how the technology is governed.
The subtext is a contest over the architecture of AI governance itself. The United States has generally resisted efforts to regulate AI through international bodies, leaving an opening that Beijing is moving to fill. By building a coalition with its own headquarters, membership, and rule-making ambitions, China positions itself as the convening power for the parts of the world that feel shut out of the US-led AI economy. Analysts at the conference read the appearance as a signal that China intends to present itself as a reliable technology partner to the developing world — and as a deliberate refusal to concede AI leadership to Washington.
The Contradiction Xi Left Unaddressed
The openness narrative sits uneasily against a countervailing development. Beijing has reportedly been in discussions with domestic firms including Alibaba and ByteDance about potentially restricting overseas access to China’s most capable models. That tension — promoting open source abroad while weighing controls on the frontier at home — went unmentioned in the keynote. It matters because it exposes the limits of the public-good framing: openness is being offered at the tier where it builds market share and goodwill, while the highest-capability systems may be treated as strategic assets to be guarded, much as the US guards its chip technology. The rhetoric of shared abundance and the reality of national advantage are not fully reconciled.
The US Counter-Position
Washington’s stance is the mirror image. American-led restrictions have blocked China from acquiring some of the most advanced computing hardware, a policy Beijing frames as the very injustice the WAICO is meant to remedy. The US has also declined to route AI governance through multilateral institutions, preferring to retain freedom of action. Trump’s decision to deliver a combative address on the same day as Xi’s keynote reinforced the divergence: where Xi offered coalition-building and the language of cooperation, the US response emphasized confrontation and unilateral leverage.
Neither posture is purely principled. China’s openness serves its interests as much as US restriction serves Washington’s, and both governments are pursuing dominance through the tools available to them — Beijing through distribution and coalition, Washington through control of the supply chain’s chokepoints.
What to Watch
The near-term test is whether WAICO gains substance or remains a symbolic coalition. Membership numbers are easy to announce; functioning governance rules, shared standards, and real capacity transfers are harder to deliver, and the organization’s credibility will turn on the latter. The second test is the openness contradiction: if Beijing follows through on restricting access to its frontier models, the public-good framing will be difficult to sustain and the Global South pitch will look more transactional. The third is Western response — whether US allies treat the Shanghai body as a rival institution to be countered or a forum to engage.
What is already clear is that AI governance has become a full arena of great-power competition, with two incompatible visions on offer: one that routes global AI through open Chinese models and a Beijing-anchored institution, and one that keeps the frontier concentrated and governed on American terms. Xi’s Shanghai debut was the clearest articulation yet of the first. The contest over which vision the rest of the world adopts is now underway.