US and China trade leads in AI capabilities ahead of summit
Ahead of a Trump-Xi meeting, a comparison shows the US holds a dominant share of global AI computing power, while China leads in research output and is narrowing the gap on advanced models. Washington has proposed an AI notification mechanism with Beijing to manage risks. The analysis covers computing power, spending, models, and research.
The summit between Trump and Xi follows a U.S. proposal for an AI notification mechanism, a hotline to share alerts on national-security threats from AI incidents. Treasury Secretary Bessent confirmed the offer ahead of talks that also cover trade, Taiwan, and the U.S.-Israel conflict with Iran. Meanwhile, computing power remains a key U.S. advantage, with nearly 75 percent of global AI capacity, driven by Nvidia’s chip dominance and over 5,400 data centers. China holds about 14 percent, though Huawei’s share is growing.
On models, U.S. systems like GPT and Claude lead on Arena, but Chinese models from Alibaba, DeepSeek, and Tencent top OpenRouter by usage, partly due to lower costs and open-weight availability. A CSIS analysis places Chinese models “months, not years” behind U.S. frontier systems, while a U.S. government estimate suggested DeepSeek’s V4 Pro trailed by roughly eight months as of mid-2026.
The AI race’s outcome could shape global economic and security dynamics, affecting businesses, researchers, and governments that rely on advanced models. If the U.S. maintains computing dominance while China closes the research gap, competition may drive faster innovation but also raise risks of fragmented standards or accidental escalation. The proposed notification mechanism may reduce miscalculation, yet its effectiveness depends on trust. Society could see cheaper AI tools from open-weight models, but also heightened concerns about surveillance, job displacement, and geopolitical tension.