China’s AI fever drives wider adoption across society, but is it mostly hype?

BEIJING – China has been gripped by a fresh wave of artificial intelligence (AI) fever, as home-grown sensation DeepSeek sparked a rush to adopt the technology that has seemingly spanned all corners of society.

That fervour permeated China’s recently concluded parliamentary meetings in Beijing, where many delegates, suited bureaucrats and corporate honchos alike, were keen to signal they were embracing the sector.

“Artificial intelligence is the buzzword of this year’s Two Sessions,” observed Mr Wu Qing, who heads the country’s securities regulatory body, referring to the meetings.

He was speaking at a press conference on March 6, at which officials said they would provide more funds and financing channels for tech and innovation, including a 1 trillion yuan (S$184.6 billion) venture capital guidance fund for cutting-edge sectors such as AI.

How much will China’s AI frenzy help the country in its push for new drivers of growth to boost a flagging economy?

“China’s generative AI adoption has undeniably hit full throttle, propelled by DeepSeek’s recent breakthrough,” said Ms Wei Sun, a principal analyst in AI at market research firm Counterpoint Research.

The start-up

stunned the world in January

with its open-source large language and reasoning models said to be comparable with the world’s best – but developed at a fraction of the cost and without the most advanced chips.

“To be sure, the current wave of AI rush carries its share of hype,” Ms Sun said. “But I think under this speculative frenzy lies genuine substance: the excitement reflects real, pent-up market demand.”

Earlier in March, investment bank Goldman Sachs revised upwards its forecast of the economic value that China can reap from generative AI adoption, after noting the rapid development of the sector of late.

Its researchers estimated that generative AI would start raising potential growth in China by 2026, and boost China’s gross domestic product by 0.2 to 0.3 percentage point by 2030.

At the Two Sessions, delegates spoke of how they were looking to increase the adoption of AI in their respective patches – a move that could raise productivity and thereby growth.

From the southern manufacturing hub of Guangdong, the boss of an industrial computing company called for more support to integrate AI solutions into manufacturing, and for industry leaders to showcase how this was done.

From Guizhou, one of China’s poorest provinces in the south-west, some delegates spoke proudly of its role as a hub for data centres, providing computing power for major products such as DeepSeek and animated movie Ne Zha 2.

A university president from Beijing said his school was preparing to set up an AI institute, and believed that students could combine traditional disciplines with the study of AI.

A cultivator of seeds in Hainan said that AI could be used for intelligent crop-breeding. And a medical practitioner from Sichuan said AI could be used to help doctors at grassroots hospitals with diagnoses and administering medicine.

Mr Qi Xiangdong, chairman of cyber-security firm Qi An Xin Technology, believes that the application of AI across industries will “definitely erupt this year”.

In the broader US-China tech race, which side has an edge when it comes to generative AI?

A global survey by AI software developer SAS Institute and market research firm Coleman Parkes in 2024 found that some 83 per cent of Chinese companies were already using generative AI in some form, significantly higher than the US’ 65 per cent.

But of this, only 19 per cent of Chinese companies said they had fully implemented the technology into their workflow, trailing the US’ 24 per cent.

Speaking to The Straits Times on the sidelines of the parliamentary sessions, Mr Qi said that even though generative AI models have been around for some time, smaller businesses faced difficulties using these as they were largely proprietary.

Now DeepSeek’s free-to-use and open-source models, he projected, would lead more small and medium-sized enterprises and traditional companies to use and integrate the technology in their processes.

Indeed, since DeepSeek made global headlines in January, a flurry of Chinese companies from telcos and cloud providers to electric carmakers and brokerages have announced they would adopt its models in some form.

Local governments, too, have been incorporating DeepSeek into their systems, including to deliver government services.

The AI frenzy further ratcheted up in March when a Chinese start-up

unveiled an AI agent named Manus,

which it said could perform by itself complex tasks such as comparing insurance policies and then generating a report with an analysis of how they differ and recommendations on which to opt for.

Its creators claim that Manus is the world’s first fully autonomous agent.

Manus is currently in a private testing stage, and only those who have an invitation code can use it.

But in an indication of public interest, these scarce invitation codes – for which interested users can join an online wait list – are reportedly being resold for as much as thousands of dollars.

Technology expert Paul Triolo, a partner at consultancy DGA-Albright Stonebridge Group, said the US still retains a lead over China in the most advanced AI models, and has access to better and more available advanced hardware.

The size of the gap between the two, he noted, will depend on two factors.

First, whether China’s domestic semiconductor industry can overcome US export controls to develop better chips and supporting software ecosystems.

The second is whether the US can successfully enforce its export controls and prevent workarounds by Chinese firms that will still strive to get access to the most advanced hardware outside China.

Meanwhile, Chinese companies could “lead in deriving economic benefits from AI deployment” as they develop more applications and use cases, Mr Triolo told ST.

This is even as not all the companies that claim to be using generative AI will necessarily be successful and derive benefits. Doing so across a complex business can be difficult, he noted, and requires expertise and resources to properly execute.