China May Be Only Six Months Behind in AI. Why That Number Matters

Industry experts put China's top AI labs as little as six months behind Anthropic. The estimate comes from a MarketWatch column published September 18, 2026. Its headline, “The World Isn’t Ready for an Anthropic IPO,” ties the lag estimate to a listing question. An IPO, or initial public offering, is the first time a company sells its shares to the public.

In my view, the gap is the number to watch for anyone pricing frontier AI companies, meaning the most advanced models at the leading edge of the field. It roughly sets how long a lead in that capability can last.

A separate comparison came from Foreign Policy, which published “How America and China Compare on AI” on October 8, 2026. The author, Matt Sheehan, lived and worked in China for six years and has helped host unofficial U.S.-China dialogues on AI.

The public record of Eric Schmidt's view on the race has moved in both directions. In May 2024, Business Insider reported that he said China could not catch up to the US in AI for four years, citing four factors including chip shortages. Chips are the specialized processors AI systems run on. The same report noted that China had approved over 40 AI models in a six-month period. By November 2024, the Harvard Crimson reported that Schmidt, speaking at a Harvard Institute of Politics forum, said the US was falling behind China in the race to develop more powerful AI. That later statement reverses the earlier four-year forecast.

Schmidt's earlier positions also cut against each other. In April 2023, Business Insider reported that he said a six-month pause on AI development in the West would simply benefit China. The pause proposal was about slowing development. Schmidt's objection was that slowing down would hand ground to Beijing. So the same six-month unit has now been used to describe a proposed pause, a current lag, and a lead that could be lost.

In my view, that recurrence is the thing to watch. A six-month gap measures elapsed time. It says nothing direct about capability, compute (the computing power used to train and run models), or revenue. Two labs can be six months apart on benchmarks, the standardized tests used to score models, and still differ sharply on what they can deploy at scale. Anyone building a valuation on the lead has to say what “behind” means: a benchmark score, a release cadence, or access to chips.

The risk framing has run alongside the competitive one for years. CNBC reported in March 2018 that Elon Musk said at SXSW that artificial intelligence is more dangerous than nuclear weapons. In May 2023, CNBC reported a statement that AI may lead to human extinction and that reducing its risks should be a global priority.

In my view, those warnings and the race narrative pull policy in different directions. A lead treated as a security asset gets a different regulatory response than one treated as a commercial edge. That split is where the six-month figure is most likely to surface next.

Taken together, the Schmidt record has him arguing that China was behind and, within about eighteen months, that it was pulling ahead. The gap is now being quoted as a single number. In my view, that number is doing a lot of work in arguments about money and policy, and it deserves the same skepticism as any forecast.