AI Illustration: Chen Xia/GT

In recent days, there has been a lot of news regarding the development of AI. 

Faced with recent US remarks such as "whoever wins AI wins," and that the US is "leading China by a lot," commentators have since summarized the US' posture more bluntly: US is less worried about AI ending humanity than about America losing to China.

You have to wonder: Does someone genuinely ahead need to keep insisting on it?

I came across this news item in Chinese media. In Hebei Province, the Beijing-Tianjin-Hebei region just released an AI-for-manufacturing plan pairing 276 supply-side resources with 520 concrete factory-floor needs. At Shougang Jingtang company, a platform combining large language models, knowledge graphs and industrial AI agents now diagnoses equipment failures on its own and hands workers a repair plan - a case Hebei officially showcased as a model of "AI+ manufacturing." This means AI in China is moving into the factory, into the machinery, into specific, unglamorous problems.

That reminded me of a TIME piece published on September 2, 2026: "Why America's AI Boom Isn't an Industrial Boom." The authors lay out the numbers precisely: US hyperscalers' R&D and capital spending grew 50-fold over two decades, from $15 billion in 2005 to $750 billion in 2025, and could approach $1 trillion by the end of 2026. Staggering. But then comes the line that matters: "Productive investment as a share of US GDP has barely budged."

Money is pouring in, but the factories, equipment and infrastructure that actually make things haven't grown to match. To be fair, the article doesn't just diagnose - it proposes modular construction and AI-driven productivity gains as ways to close the gap. But even accounting for those fixes, the estimate still shows a remaining shortfall. That tells you the problem isn't a lack of ideas. It's how hard the problem actually is.

Even American media isn't fully convinced. The same TIME article notes that building semiconductors in the US costs roughly 40 percent more than in the world's most competitive locations; pharmaceuticals, 60 percent more. US construction costs are about double what they are in Asia, and construction times can be twice as long. The piece states it plainly: "One clear challenge to sparking a US industrial renaissance: It's expensive to make in the US."

That's the real divide. The US and China aren't running the same race.

The US is scaling upward - bigger models, more compute power, higher valuations, capital cycling through markets and boardroom narratives. That's real capability: chips, algorithms, foundational research. But has it translated into shop-floor productivity? 

The TIME piece notes the AI investment boom "has yet to reshape investment across the broader industrial economy." Factory-structure investment fell 6 percent by late 2025 after peaking in 2024. General industrial equipment spending was essentially flat. Reshoring that began in 2022 has "plateaued in the numbers." Capital is flying, but not landing.

China is taking root downward - not chasing the biggest model, but embedding AI into blast furnaces, compressor inspection lines, garment-sewing stations. Fault diagnosis, humanoid mass production - this is what that looks like. It's less glamorous than Artificial General Intelligence narratives or existential-risk debates, but it has something hard to replicate: scale of real-world scenarios. 

China's vast, diverse manufacturing base means AI never lacks work to do - every factory floor generates real problems, real data and a feedback loop that keeps improving the models. Once that loop is running, it's hard to stop.

Can the US cut China off? The US can restrict chips, control equipment and block certain technologies. But an industrial ecosystem willing to actually deploy AI on the factory floor isn't something export controls can choke off - it grows from its own soil.

The US' cloud-and-valuation-driven approach risks deepening an old problem: industrial hollowing-out. Capital flows into equities and a handful of giants; manufacturers can't get financing; costs stay too high to build; factories keep relocating abroad. The louder the AI boom gets, the wider that gap between the virtual and the real may grow. What happens to AI that only floats in the cloud, never touching ground? I don't know. However, I know that AI is now being deployed across Chinese factory systems at an unprecedented speed.

The author is a senior editor with the People's Daily. dinggang@globaltimes.com.cn. Follow him on X @dinggangchina