Stanford HAI: China has nearly caught up with the US in AI and is already ahead in patents and robots

The Stanford HAI AI Index shows that the technological gap between the USA and China in artificial intelligence has almost closed. Leading models are nearly on par, and by several research and industrial activity indicators, China is already pulling ahead.

The Gap Has Almost Closed

A couple of years ago, American leadership in AI seemed unconditional, but in 2025–2026 the picture has changed noticeably. According to Stanford HAI, American and Chinese models have repeatedly traded places in performance rankings since the start of 2025. In February 2025, DeepSeek R1 briefly matched the best American model, and in March 2026, the Chinese leader's lag from the top US model was only 2.7%. For the market, this is no longer a comfortable margin, but a statistical distance that can be quickly closed by a new model version or successful tuning.

"The performance gap between US and

Chinese AI models has essentially closed."

In practice, this changes the very logic of competition. If the conversation once centered on who was stronger in raw model quality, now factors like inference cost, reliability, update speed, and performance in narrow domains—from code and search to enterprise tasks—are becoming more important. In other words, the struggle is gradually shifting from flashy benchmarks to product and infrastructure. For American companies, this is bad news: technological advantage alone is no longer enough to feel unassailable.

Where China Is Ahead

By several systemic measures, China looks not like a catch-up player, but as a full-fledged leader. Stanford HAI notes that the country outpaces the USA in the volume of scientific publications, number of citations, and quantity of issued AI patents. Additionally, China dominates industrial robotization: in 2024, it accounted for 54% of all industrial robots installed worldwide. This is an important marker because it concerns not only research, but also the ability to quickly deploy AI in manufacturing, logistics, and equipment.

• China has more AI publications than the USA
• China leads in citations of AI research
• China also has the advantage in the number of AI patents
• China accounted for the majority of new industrial robots worldwide
• The USA maintains an advantage in the most prominent models and most influential patents

Another telling detail: China's share among the 100 most cited AI papers rose from 33 in 2021 to 41 in 2024. This means the Chinese ecosystem is strengthening not only in volume but also in the weight of results. At the same time, the USA still holds stronger positions in the most influential patents and the number of prominent models, while South Korea stands out separately as the global leader in AI patents per capita.

Money and Infrastructure

Where the USA remains noticeably stronger is capital and computational base. In 2025, private AI investments in the USA reached $285.9 billion versus $12.

4 billion in China. The gap looks enormous, but this figure comes with an important caveat: Chinese statistics poorly capture government financing mechanisms, so the real scale of Beijing's investments may be far higher than private estimates. Additionally, the USA remains the global center of AI infrastructure: the country has 5,427 data centers, more than ten times higher than any other country.

But here too, American leadership is becoming vulnerable. Stanford HAI writes that the influx of AI researchers and developers to the USA has dropped 89% since 2017, with an 80% decline in just the last year. This doesn't mean America has stopped being the main magnet for talent—it still concentrates more AI specialists than anyone else.

However, the trend matters: if top talent finds it increasingly difficult or less profitable to move to the USA, the advantage could gradually shift to those ecosystems that better retain their engineers, build local computational capacity, and scale their domestic market.

What This Means

The AI race no longer looks like a game with one obvious leader. The USA is still stronger in money, data centers, and commercial ecosystem, but China has almost closed the gap in model quality and already leads by several basic indicators of scientific and industrial power. For the market, this means one thing: the next phase of competition will be decided not by slogans, but by implementation speed, access to compute, and the ability to turn research into mass-market products.