• Nvidia CEO Jensen Huang has warned about China’s big advantages in AI
• Those are the fast realization of the construction of data centers, and China’s robust energy infrastructure for fuelling the power demands of AI
• Meanwhile, a new study has found that China’s open source LLMs have secured almost a third of global AI usage

Nvidia‘s chief executive, Jensen Huang, has again warned about the swift progress that China is making with AI, and the advantages the country has in terms of the infrastructure for development therein.

Fortune reports that late last month, Huang spoke to John Hamre, the president of the Center for Strategic and International Studies (CSIS), noting that: “If you want to build a data center here in the United States from breaking ground to standing up [an] AI supercomputer is probably about three years. They [China] can build a hospital in a weekend.”

In other words, China is capable of realizing large construction projects at incredibly swift speeds, and also has a major advantage in terms of its energy infrastructure.

These are crucial elements for the development of AI in terms of building huge data centers quickly, to cope with processing needs, and having the energy to power all this.

Huang observed that China has “twice as much energy as we [the US] have as a nation, and our economy is larger than theirs” and that this “makes no sense to me”, and furthermore that growth in energy capacity is heading “straight up” in China, while remaining more or less flat in the US.

To balance out the worries aired, however, the CEO did make it clear that Nvidia is “generations ahead” of China when it comes to AI chip tech – there may be a tiny bit of bias in that assertion, mind – but Huang still said that this wasn’t a reason to rest on any laurels.

Huang has previously commented about China being “nanoseconds behind America” in the AI race, but we’re told the Nvidia CEO remains outwardly hopeful about the Trump administration’s push to boost AI investment and domestic manufacturing jobs.

Token efforts and a speedy ascent

Meanwhile, a separate article from the South China Morning Post (SCMP) claims that almost 30% of the global use of AI now comes from China’s open source models (LLMs).

That figure comes from a report compiled by OpenRouter, an independent AI model aggregator, along with venture capital firm Andreessen Horowitz. It’s based on a study of 100 trillion tokens, which are the units of data processed by LLMs (or in friendlier language, the building blocks behind how AI works).

The lion’s share remains with the closed source western world LLMs, such as ChatGPT, which hold the rest of the market (around 70%).

Remember, though, that just a year ago, Chinese open source LLMs only represented just over 1% of tokens, so reaching 30% now is quite a steep growth trajectory to say the least.

If you take just open source LLMs, we’re told that Chinese models average about 13% of weekly token usage, almost equalling the 13.7% which is drawn from the rest of the world. (This is open source usage, remember – the remaining majority are the closed source proprietary models like ChatGPT).

Another interesting point revealed here is that the open source LLMs from China are now equally pulling their weight, it’s not just all about DeepSeek (as was the case originally). Naturally, DeepSeek V3 is a major force in AI usage for China, but there’s also Alibaba’s Qwen models and Moonshot AI’s Kimi K2 which are big players.

The report contends that Chinese language prompts are second in token volume behind English now.

Tying all this together, the rise of China in the AI sphere is a rather dizzying ascent, then, and you can see where Huang’s concerns are coming from. Especially as it’s difficult to see this growth slowing in the near-term for China, and what Nvidia’s CEO observes about the country’s energy infrastructure is indeed a telling advantage over the US – again, one that’s difficult to see changing in the nearer future.

And then as we’ve recently seen with the release of DeepSeek’s new v3.2 models, there’s what China has to offer in terms of reducing the costs of using AI, to boot. It would seem that there’s a seriously competitive battle ahead for global AI dominance.

Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. Make sure to click the Follow button!

And of course you can also follow TechRadar on TikTok for news, reviews, unboxings in video form, and get regular updates from us on WhatsApp too.

Darren is a freelancer writing news and features for TechRadar (and occasionally T3) across a broad range of computing topics including CPUs, GPUs, various other hardware, VPNs, antivirus and more. He has written about tech for the best part of three decades, and writes books in his spare time (his debut novel – ‘I Know What You Did Last Supper’ – was published by Hachette UK in 2013).