AI race between US and China

The US leads the AI race worldwide. China is not close, yet not that far — and it is catching up, fast.

"Leading" can mean several things, though, and the answer depends on which part of the race you look at.

How close are the best models?

Stanford's 2026 AI Index measured this on the Arena leaderboard, where people compare answers from two AI models side by side and vote for the better one. In March 2026, the top American model, Anthropic's Claude Opus 4.6, scored 1,503. The top Chinese model, ByteDance's Dola-Seed-2.0, scored 1,464 — a gap of 2.7%.

In May 2023, the gap was between 17.5 and 31.6 percentage points. The US still has the top model, but the distance has narrowed sharply in under three years.

Who is spending more?

Private AI investment in 2025 was $285.9 billion in the US and $12.4 billion in China — about 23 to 1. Stanford notes that this likely understates China's spending, because much of it flows through government funds rather than private investors.

What about chips and computing power?

The Council on Foreign Relations estimates that the best American AI chips are about five times as powerful as Huawei's best, and that even under generous assumptions, Huawei produces around 5% of the AI computing power Nvidia does in 2026. Nvidia made about 4.5 million AI chips in 2025; estimates for Huawei range from 200,000 to 800,000.

Where does China lead?

In open models — AI that anyone can download, use and build on. Hugging Face's data for January to August 2026 shows that Alibaba's Qwen is the most-built-on open model family in the world, with 151,000 models built from it — 2.6 times as many as Meta's. And in almost every month of 2026, the largest and strongest open model released by anyone came from a Chinese lab.

Where the race stands

The two countries lead different parts of the race. The US leads by a wide margin on money and computing power, and narrowly on the best models. China is a close second on what its models can do, reached that point with about a twenty-third of the private investment, and leads on the open models that developers around the world build on.

And on the best models, the gap that was 17.5 to 31.6 percentage points in 2023 is now 2.7%.

These figures date from December 2025 to August 2026, in a field that changes month to month.