Global AI Divide: US Pushes for Alignment, China Wins on Cost and Access

Key points

• Chinese AI models surged to 54-67% of global usage on platforms like OpenRouter and Vercel in 2026, up from under 15% a year ago, driven by lower costs and open-source accessibility.
• Xi Jinping launched the World AI Cooperation Organization in July, recruiting 29 countries including Russia and Indonesia, as a rival to Trump's 'Pax Silica' alliance, which has over 24 signatories.
• The Trump-Xi summit yielded limited progress on AI, establishing only a bilateral communication channel, as deep mistrust persists over US containment efforts and Chinese 'distillation' of US models.
• US lawmakers are pushing the 'Chip Security Act' to track advanced chips and prevent smuggling to China, despite industry warnings that such restrictions could push Southeast Asian nations closer to Beijing.
• Experts note that while US models lead in frontier capabilities, Chinese models are sufficient for many tasks, and the 'race' may be defined by economic diffusion and adoption rather than raw speed.

Background

Recent weeks have seen escalating tensions over AI governance, with China dismissing US calls to slow development as 'fearmongering' and 'Cold War tactics' (Archive, Sep 14). A minimal 'AI incident line' was proposed by Treasury Secretary Bessent to flag security issues, but China has not publicly agreed (Archive, Sep 22). Xi’s recent state visit to the US highlighted China’s resilient trade strategy despite domestic economic struggles (Archive, Sep 23).

How outlets are covering it

CNN and CNBC emphasize the rapid shift in global usage, with Chinese models dominating in the Global South due to cost and open access, while US firms like Airbnb and Shopify also adopt them for efficiency. The New Yorker challenges the 'race' narrative, arguing that 'diffusion' and adoption matter more than being first, and notes that US fears of Chinese 'superintelligence' may be overstated compared to practical diffusion. Politico highlights the physical reality of chip smuggling in Southeast Asia, where US investments in data centers inadvertently help China evade controls, creating a conflict between industry interests and congressional security demands. All sources agree that the Trump-Xi summit produced little substantive cooperation on AI due to entrenched mistrust.

Why it matters

The outcome of this AI competition will shape global technological infrastructure, economic influence, and security frameworks. If Chinese open models continue to dominate in emerging markets, the US may lose long-term economic leverage despite its frontier lead. Conversely, overly restrictive US chip controls could backfire, pushing key allies toward Chinese technology ecosystems. The definition of 'winning' AI—whether by capability, adoption, or profit—remains contested, with implications for how nations regulate and deploy AI in the coming decade.

What to watch

Watch for the passage of the 'Chip Security Act' in the US Senate, which could impose tracking mechanisms on advanced chips. Monitor whether Chinese open models remain unrestricted or if Beijing introduces licensing rules to limit their use in sensitive sectors. Observe the impact of the new US-China AI communication channel on future cooperation or conflict, and track adoption rates in Southeast Asia as a key indicator of global AI alignment.