China backseat driving risks sending AI off course

China's economy is a study in contradictions. Decades of centralised industrial planning in China have led to endemic overcapacity, which in turn has fuelled destructive price wars across an array of sectors. Yet Beijing's approach, for all its flaws, also helped create world-class corporate champions like electric-car maker BYD. Officials now want to curb what they call "disorderly competition" in the booming $140 billion artificial intelligence sector. There is some logic to intervening, but that will come at a cost.

Unlike with EVs or solar panels, it's not clear what exactly about AI is worrying China's economic planners. On the contrary, optimism is buoying up the sector, which combined with infrastructure and other related sectors will be worth $1.4 trillion by 2030, analysts at Morgan Stanley reckon.

Such optimism has pushed the Hang Seng Tech Index in Hong Kong up 30 percent this year, outperforming the mainland benchmarks. There is some justification for the hype. Despite US sanctions and export restrictions, DeepSeek, Alibaba and peers are still churning out open-source models that are competitive with, if not superior to, those from Meta, OpenAI and others in the West.

Chatbots and agents are proliferating across the country. Alibaba last week revealed that surging demand for AI services powered triple-digit year-on-year growth in related revenue. And chipmaker Cambricon Technologies revealed that its first-half revenue rocketed more than 4000 percent from a year ago, albeit from a low base.

But there's caution, too. Cambricon's Shanghai shares had doubled since the end of July. On Friday, hours before Beijing stated its willingness to intervene in the industry, the $88 billion company took the unusual step of warning investors that the stock price may have "deviated from the company's current fundamentals", triggering a selloff.

To be sure, pockets of wasteful investments are bubbling up. It's reminiscent of the mad scramble playing out among US tech giants, which analysts at research outfit IDC forecast, will spend a whopping $336 billion in 2028 on AI. In the People's Republic, local governments desperate to hit growth targets have embarked on a debt-fuelled spree to hoard chips and build data centres – many of which are sitting idle. And price wars are intensifying in what one tech executive dubbed "a war of a hundred models", even as business models remain hazy.

Yet the authorities now intent on coordinating the country's AI development and resources do not have a strong record of picking winners, particularly in sectors where the technology is rapidly evolving. Semiconductors are a prime example: the government has deployed billions of dollars into state-designated champions, only to see many go bust, including Wuhan Hongxin Semiconductor Manufacturing and Tsinghua Unigroup, whose former chairman was jailed for corruption and embezzlement.

Too much oversight and planning will also squeeze the private sector, where under-the-radar upstarts like DeepSeek have unexpectedly emerged. Beijing is prudent to want to avoid excess, but history suggests its approach risks sending progress off course.

An official from China's top economic planner, the National Development and Reform Commission, said on August 28 that the government will prevent excess and disorderly competition in artificial intelligence. "We will resolutely avoid disorderly competition or a 'follow-the-crowd' approach," Zhang Kailin told reporters at a briefing.

Separately, China's State Council on August 26 released an "AI Plus" action plan, which laid out targets for AI adoption across six key economic sectors by 2030.