Letting Silicon Valley dictate US AI policy will benefit China

While humanoid robots sparred in the ring at China’s World AI Conference in Shanghai, a more consequential contest played out just steps away. Chinese officials unveiled plans for a new global governance body — the World AI Cooperation Organization (WAICO) — aimed at setting international standards and norms around artificial intelligence. Framed as a bid to prevent monopolies and foster inclusivity, the move was also a clear shot across Washington’s bow: a declaration that China intends not just to compete in AI, but to define its rules and global architecture.

That same week, the United States released its own high-level vision: Winning the Race: America’s AI Action Plan. Spearheaded by AI czar David Sacks, the plan outlines a sweeping, three-pillared strategy focused on accelerating innovation, building infrastructure, and leading in international AI diplomacy. It is a good start — a robust and ambitious document — but it still reflects a fragmented and uneven approach.

What is becoming clear is that the race for global AI leadership will be won not only by those who develop the most powerful models, but by those who can deploy them when and where they matter most. China’s practical, application-first approach — deploying open-source AI to improve agriculture, logistics, education, public health, and other public services — makes its model especially attractive to countries seeking early, visible gains. Its export of utility and digital infrastructure — including power grids and smart city platforms in bundles like Huawei’s compact “AI-in-a-box” systems — is enabling even low-resource nations such as Kenya, Thailand, and Egypt to scale AI applications without the need for expensive hyperscale data centers.

To be clear, China’s approach is not without risk to its users. Its AI systems often come bundled with surveillance tools, opaque financing, and long-term strategic dependencies. But for many global south countries, these are acceptable trade-offs for tangible benefits. Beijing is meeting rising demand for AI access at cost, at scale, and at speed. In doing so, it is turning AI access into a powerful instrument of soft power, one that fuses technological exports with influence over digital norms, governance preferences, and strategic alignment.

By contrast, the U.S. approach remains overly anchored in Silicon Valley’s commercial interests. American firms continue to lead in foundational models and generative AI, but the strategy often fails to connect technological excellence with global deployment or development impact. Without an accessible, adaptable model that serves diverse needs, the U.S. risks ceding the AI frontier to China by default.

With “Winning the Race AI Action Plan” now on the table, even with its imperfections, that may be starting to change. Its pillars — innovation, infrastructure, and diplomacy — signal Washington’s recognition that AI leadership demands more than cutting-edge research. It requires trusted systems, trusted rules, and trusted partnerships.

President Trump’s high-profile visit to the Middle East in May reflected this growing awareness. Gulf states pledged to import up to 500,000 Nvidia chips annually and build new AI data campuses in the UAE. While these deals remain largely commercial and defense-driven, they underscore rising U.S. acknowledgment of AI’s geopolitical heft, especially in regions where China has moved quickly to shape the terrain.

At the inaugural APEC Digital and AI Ministerial in Incheon, South Korea, U.S. officials took a further step, joining regional counterparts to discuss responsible AI standards, cross-border data governance, and inclusive digital development. Modest in scope but meaningful in signal, it showed a willingness to engage the multilateral arenas where China has been setting the pace.

But more engagement is needed. Once the domain of technical consensus, international standards bodies have quietly become battlegrounds, where norms on everything from algorithmic transparency to facial recognition are contested. China has moved swiftly to embed its preferred frameworks, language, and values into these technical settings. If Washington remains absent or reactive, it risks inheriting a global AI order built on someone else’s blueprint.

The private sector is beginning to respond as well. This week, OpenAI released its first truly open model, a departure from its proprietary stance and a subtle nod to global demand for AI…