Full Breakdown
US-China AI Race: Capabilities, Spending and Talent Ahead of Washington Summit
By Drooid · · How we work
Summit Context and AI Focus
U.S. President Donald Trump and Chinese President Xi Jinping are slated to meet in Washington, DC this Thursday for a summit that will address trade, Taiwan, the war in Iran and artificial-intelligence (AI) policy. In advance of the talks, U.S. Treasury Secretary Scott Bessent said Washington has proposed an AI “notification mechanism” – a hotline to alert both sides when AI incidents could threaten national security. Leading AI firms have simultaneously warned about the technology’s broader risks.
Computing Power and Infrastructure
By the metric of total AI-related floating-point operations per second, the United States provides nearly three-quarters of the world’s AI computing capacity, while China accounts for just over 14 % (Epoch AI). The U.S. advantage stems largely from its access to the most advanced chips: Nvidia supplies more than 60 % of global AI computing capacity among major chip designers, whereas Huawei holds a much smaller, though growing, share (Stanford 2026 AI Index). The United States also operates more than 5,400 data centres—about ten times as many as any other nation—and leads with 84 dedicated AI data centres, outpacing the combined total of the next eight countries.
AI Models and Market Presence
Frontier models such as OpenAI’s GPT and Anthropic’s Claude remain U.S.-led, yet Chinese offerings are narrowing the gap. On the Arena leaderboard (March 2026) U.S. firms Anthropic, xAI, Google and OpenAI rank alongside China’s Alibaba and DeepSeek. On OpenRouter, Chinese models from DeepSeek, Z.ai and Tencent occupy the top three slots by tokens processed, helped by lower usage costs and open-weight availability. The Centre for Strategic and International Studies (CSIS) describes Chinese models as “months, not years, behind” U.S. frontier models, while the U.S. Center for AI Standards and Innovation (CAISI) estimates DeepSeek V4 Pro (released April) is about eight months behind leading U.S. systems.
Investment and Spending
U.S. hyperscalers—including Amazon, Microsoft, Google, Meta and Oracle—are projected to spend roughly $764 billion on AI infrastructure in 2026 (Goldman Sachs). Chinese counterparts Alibaba, Tencent, Baidu and ByteDance are expected to spend about $102 billion. Although the gap is large, TrendForce forecasts Chinese hyperscaler capital expenditure to rise by more than 80 % in 2026, compared with a 76 % increase for U.S. firms.
Research Output and Talent Landscape
China contributed over 27 % of global AI publications in 2024, versus 12 % from the United States (Center for Security and Emerging Technology). A MacroPolo study finds that 47 % of the world’s top 20 % of AI researchers earned their undergraduate degrees in China in 2022, up from 29 % in 2019, while 72 % of top AI researchers educated in China are currently employed in the United States.
These comparative metrics frame the strategic dialogue expected at the upcoming summit, highlighting where each superpower leads and where the competition remains tightly contested.
