China’s AI strength suggests US curbs could backfire

AI has emerged as a global national security imperative.

US officials and tech leaders have spent much of this year celebrating the success of all-out efforts to stifle China’s AI ambitions.

But it’s too soon to count Beijing out of the race, and American attempts to limit China’s progress have pushed the nation to advance in creating a homegrown ecosystem at a time when developing sovereign AI has emerged as a global national security imperative. It would be unwise to underestimate the stakes for China, especially when the technology is still in its infancy.

At a developer conference last Friday, the chairman of Huawei Technologies Co’s consumer business arm, Richard Yu, said the tech giant — long kneecapped by US restrictions — has taken 10 years do what it took European and American counterparts 30 years. Yu touted Huawei’s latest processers as 1.1 times more effective in training AI models compared to other offerings on the market. (Notably, Yu stopped short of naming specifics here, while previous Bloomberg News reporting suggests costly and mixed results in the firm’s semiconductor push). But Huawei’s race to develop its own chips has also surprised industry watchers, and even caught the attention of Jensen Huang, the chief executive officer of AI’s golden child, Nvidia Corp.

Separately, Chinese quant firm High-Flyer Capital Management quietly released an open-source AI model earlier this month dubbed DeepSeek Coder V2 that impressed much of the global tech community with its ability to write code and do math. Its developers also claim it beat competitors at common benchmarks — at a fraction of the cost of other tools developed by US tech giants, and despite the intense restrictions on access to chips from Washington.

US tech curbs on semiconductors present the biggest obstacles to China’s ambitions. Companies have been cut off from accessing the most-advanced chips and manufacturing tools made by America and its allies. But this has also incentivized Beijing to double down on longer-term efforts to create a self-sufficient chips and AI ecosystem.

Dutch firm ASML Holding NV., which effectively holds a monopoly over the most-advanced machines required for developing cutting-edge chips, finds itself in the middle of the geopolitical hot mess. The Netherlands government, under US pressure, has banned the sale of its top-of-the-line equipment in China. But as Chief Executive Officer Christophe Fouquet

noted in the Wall Street Journal earlier this month, the more restrictions, the stronger the push for people to do it themselves.

Access to chips is a critical aspect of the global AI race, but access to talent is arguably the backbone of future development. And by this measure, China is making immense strides in both developing and attracting elite talent. Top-tier researchers originating from China jumped to 47% in 2022 from 29% in 2019, according to most-recent data tracked by think tank Macro Polo, while those coming from the US fell to 18% from 20%. Many of the brightest minds in the field are still choosing to work in the US, but China is rapidly closing the gap.

Indeed, by most measures, the US remains the global front-runner when it comes to AI, but the technology that promises to transform entire industries hasn’t even begun to peak.

America spent years trying to cripple Huawei’s tech dominance, and for a while it seemed to be working. But the company’s recent successes have since shown that these efforts may have backfired and made it stronger and more self-sufficient. The numbers don’t lie: Huawei’s profit surged a whopping 564% in its March quarter.

China has long resented foreign attempts to influence its economy, dating back to the 19th century when Britain tried to force China to buy opium. More recently, this patriotism has been cited as one factor that pushed domestic sales of Huawei smartphones and led to the company’s Harmony operating system overtaking Apple Inc.’s iOS in Chinese market share earlier this year.

Huawei’s surprise comeback shows it is unwise to underestimate how Chinese companies and consumers behave when they feel oppressed by a foreign power and how that’s galvanizing them in the AI race. Sheer market size alone and a collective national interest could tip the balance of power toward Beijing in the years ahead. Washington shouldn’t discount how imposing scarcity all too often breeds innovation.