AI Becomes the New US-China Negotiating Front
AI is emerging as a major front in US-China negotiations, influencing talks on trade, security and technology governance.
In a Nutshell:
- US and China will hold official AI talks in September, likely before Xi Jinping's September 24 visit to Washington, per five sources familiar with the matter.
- Treasury Secretary Scott Bessent will lead the US delegation; participants, agenda, and location remain undecided as preparations are at an early stage.
- China's foreign ministry said Beijing is willing to implement the AI consensus reached by Trump and Xi, calling AI a major global challenge.
- Trump confirmed on July 23 that AI will be discussed when Xi visits the US on September 24.
- The talks will be the first official US-China AI dialogue under Trump, building on a 2024 Biden-Xi understanding that humans, not AI, should control nuclear weapons decisions.
- The US has held off placing DeepSeek, CXMT, and 100+ other Chinese firms on a trade blacklist to avoid escalating tensions ahead of talks.
- China is reportedly considering curbs on overseas access to its most powerful AI models, including Anthropic's Mythos, over hacking concerns.
Context
The talks stem from the Trump-Xi summit in May 2026, where Bessent said discussions would aim at "halting proliferation of powerful AI models... to non-state actors". This follows a broader pattern of escalating US export controls on AI chips to China, including a January 2026 rule that permits limited exports of Nvidia H200 and AMD MI325X chips under strict volume caps and end-use certifications. Washington has simultaneously been building a global AI chip export regime that divides countries into allied, restricted, and conditional tiers, with China in the most restricted category.
Why it matters
These talks mark a pivotal moment in US-China technological rivalry, transforming AI from a purely commercial competition into a subject of formal intergovernmental diplomacy comparable to nuclear arms control negotiations. The stakes are high: both nations fear that the other's frontier models could enable cyberattacks on critical infrastructure, enhance military capabilities, or destabilise labour markets, while unresolved issues such as IP theft accusations and Chinese "watermarks" on US models threaten to derail cooperation before it starts. Strategically, the outcome will shape whether the world moves toward a bifurcated AI order, evidenced by Washington's push for allies to pick sides in the AI coalition versus Beijing's competing framework, or a more managed coexistence. Economically, chip export policy remains the central lever, with the US having already relaxed some restrictions on advanced chips to China even as it maintains a "presumption of denial" for Chinese-owned overseas data centres. Xi's September 24 visit adds political weight, making these talks a bellwether for broader US-China relations under Trump, and any agreement, even a narrow one on defining "frontier" AI models, would set a precedent that could influence global AI governance norms for years.
What we think
The emerging US-China AI dialogue is an early indicator of how global AI governance and chip-supply rules may crystallise, with Bangladesh relying on both American and Chinese technology ecosystems for its digital transformation ambitions. Washington’s reported pressure on partner nations to choose between competing AI coalitions could create alignment pressures for Bangladesh, potentially affecting access to advanced GPUs, cloud infrastructure and AI investment from either bloc. Bangladesh’s draft National AI Policy 2026–2030, which proposes duty-free imports of GPUs and AI hardware, makes developments in US export-control tiers and China’s model-access restrictions particularly relevant to the availability and affordability of critical hardware. Bangladesh’s semiconductor engagement with Chinese firms adds another layer of complexity, as deeper ties with China could draw greater scrutiny from US-led technology frameworks.