One year ago, the world of artificial intelligence (AI) was shaken by the news that a Chinese startup company, DeepSeek, had developed a chatbot of comparable efficiency to those of the US with far less computing power and at a fraction of the cost.

The emergence of DeepSeek has been followed by the development of a number of Chinese AI companies which, like DeepSeek, have followed a different road from the US firms.

They have developed open-source facilities which means that their operations can be taken up and improved on. The US AI giants operate on a closed system in which the underlying technology is kept in-house in the belief that such practices will enable them to reap super profits to pay for the massive investment in data centres.

In an interview with the Financial Times (FT) last week, the president of Microsoft, Brad Smith, warned that US AI groups were being outpaced by Chinese companies outside the western powers, especially in so-called emerging markets.

“We have to recognise that right now, unlike a year ago, China has an open-source model, and increasingly more than one, that is competitive.”

He made the usual complaint of US firms facing competition from Chinese firms that state subsidies enabled its AI companies “to undercut American companies based on price.”

The FT reported that according to research by Microsoft the release of the DeepSeek model a year ago “helped accelerate the uptake of AI worldwide, particularly in the global south, due to its ‘accessibility and low cost.’”

Chinese capital in the form of technology groups such as Alibaba and Tencent as well as other investment groups is moving into AI.

Last Friday MiniMax, a Shanghai-based AI firm, raised $619 million in its initial public offering (IPO) on the Hong Kong stock exchange. Its stocks closed more than 100 percent higher than the offer price and took its market capitalisation to $13.5 billion.

Another Chinese startup, Knowledge Atlas Technology, also known as Zhipu, also debuted in Hong Kong and raised $558 million with its shares rising 37 per cent above their IPO price.

The possibility that China may catch up to the US in AI development and even surpass it is being widely canvassed.

A recent report by Leah Fahy, China economist at Capital Economics, said that China “will challenge the US as a global leader in AI.”

“One year on from DeepSeek’s breakthrough, Chinese AI is still hot on the heels of US tech. Despite hardware constraints, China looks set to remain close to the frontier of AI development,” the report said.

In comments reported by the FT she pointed to innovations developed in China which have enabled it to work around the problems caused by its lack of access to the most advanced chips because of bans imposed by the US.

“Models trained in China may still be competitive with the best models from the US if algorithmic efficiency, data quality and system-level design can continue to be leveraged,” she said.

Writing on the AI race FT columnist Tej Parikh said it should be viewed more as a marathon rather than a sprint and on that basis, China was in a better-placed position. There were two components to achieving tech dominance: innovation and diffusion.

“The US—with its high investment, quality chips and proprietary ecosystem—may be better placed to win the sprint to the best model. China is best positioned to integrate its good-enough models into physical application and proliferate them around the world,” he wrote.

He cited a recent study by the Massachusetts Institute of Technology which showed that China’s share of the global markets for “open” AI model downloads exceeded that of the US.

But Chinese AI developers acknowledge they face major problems caused by the bans on the most advanced chips.

At a conference earlier this month, Tang Jie, the founder of Zhipu, said the chances of catching up to the US were slim and the gap may even be widening.

“While we’re doing well in certain areas, we must still acknowledge the challenges and disparities we face,” he said.

But in a demonstration of the ultimate impossibility of preventing the spread of technology in a globalised production system, ways are being found to get around the impact of the bans imposed by the Trump administration.

Chinese firms are developing connections with companies running data centres in Southeast Asia and the Middle East to rent their computing power and so get access to the most advanced chips from the US AI chip giant Nvidia. While there are problems, it is legally compliant, at least for now.

After intense lobbying from Nvidia chief Jensen Huang, Trump has lifted the ban on its H200 chips to China. Huang has said that demand for the chips, which is below the standard of the top range made by the company, is very high and the chips were “flowing through the line.”

But it remains to be seen how much they will be used. Such is the rapidity of chip development that, according to a report in the Wall Street Journal, representatives of Chinese tech companies have said that the chip, which is two generations behind Nvidia top-of-the-range Rubin series, “has become insufficient for training state-of-the-art AI.”

And there is also the question of the extent to which the Xi Jinping government will allow the use of the H200s as it pushes for China to make its independent development in chipmaking.

Maintenance of its supremacy in the AI battle is becoming an existential question for the US. The AI boom is estimated to have contributed almost 1 percentage point to the US growth rate in the first three quarters of last year and the AI boom is the mainstay of Wall Street.

The massive investment running into trillions of dollars is dependent on sufficient returns being generated by application.

But if for any reason this is undermined, for example, by new and more advanced technologies being developed or by the US simply being outmuscled in markets by its China rivals—or even the perception that such events could take place—then the outcome could be a major financial and economic crisis.

Such developments are not on some distant horizon.

In presenting its latest economic outlook update, International Monetary Fund chief economist Pierre-Olivier Gourinchas warned that there was a risk of disruption in a “surprisingly resilient” world economy if there was a reversal of the AI boom.

“There is a risk of a correction, a market correction, if expectations about AI gains in productivity and profitability are not realised,” he said.

He warned that because the tech groups’ market capitalisation was much bigger than during the dot-com bubble at the start of a century, even a small reversal could have a “big impact” and the increased use of debt in financing the investment boom was a worry.