The sudden success of Chinese artificial intelligence (AI) start-up DeepSeek comes as a morale booster for the country’s researchers and tech companies, who have been tasked by Beijing to push for breakthroughs in the face of US-led curbs on the trade, industry participants and analysts said.
DeepSeek’s emergence has also changed perceptions among economists and investors towards China’s capacity for innovation, they added, vindicating a long-term push by the country to build up its home-grown talent and technologies.
“We had grown so used to the pervasive narrative that China can only play catch up and the US’ lead is nearly insurmountable, until DeepSeek emerged,” said Ye Kejiang, a doctoral researcher and head of the cloud computing programme at the Shenzhen Institute of Advanced Technology. “Its success has demonstrated that Chinese firms can conjure breakthroughs amid adversity. It’s a feat science and tech workers like me can relate to.”
Do you have questions about the biggest topics and trends from around the world? Get the answers with SCMP Knowledge, our new platform of curated content with explainers, FAQs, analyses and infographics brought to you by our award-winning team.
With DeepSeek’s R1 language model – unveiled in late January and stunning the tech world with performance rivalling industry leaders at a fraction of the development and use cost – redrawing the global AI landscape, the Zhejiang-based company has become synonymous with China’s efforts to overcome tech bottlenecks, largely in the form of import restrictions imposed by the United States.
Ye, whose institute is part of the Chinese Academy of Sciences, said the firm’s achievement has provided a powerful argument for the feasibility of staking out a claim in the complex tech sector, even without access to the most advanced components.
“DeepSeek shows … China is not behind the West in some aspects of the tech race, like software design,” he said. “China has no shortage of smart, capable talent, whose ingenuity can be on a par with those in the West and even outsmart them.”
The once little-known Chinese firm has also helped to upend assumptions about US dominance, suggesting that Washington’s efforts at containment have ultimately proved ineffective.
David Chao, Asia-Pacific strategist at American financial services company Invesco, said the implications are broad.
“The ramification of this recent development could be that Chinese tech companies are still able to innovate, especially on software, despite US export restrictions on hardware,” he said. “This episode also raises questions about the vast sums that are currently being invested in AI and whether it will turn out to be money well spent.”
Ye said DeepSeek’s open-source model, if popularised, could also save Beijing a bundle in research investment.
“DeepSeek represents a low-cost approach to extreme software optimisation, a departure from the West’s path of building up expensive computing power,” Ye said.
“Computing power and chips remain indispensable for outstanding AI performance. But given China’s circumstances, extreme optimisation is proving more feasible. DeepSeek’s cost-effective model is worth a closer look so we can replicate it in other realms.”
China continues to trail the US in aggregate computing muscle and – despite DeepSeek’s recent triumph – the embargo on shipments of advanced hardware is slowing Beijing’s march to parity as domestic alternatives have not yet been brought up to standard.
DeepSeek founder Liang Wenfeng, who joined figures from other industries to make policy recommendations to Premier Li Qiang in January, recently said his company’s biggest roadblock is access to advanced chips – something money can no longer buy.
State media has also admitted a lack of advanced equipment remains a challenge, citing a shortage of the top-tier lithography machines used to craft high-performance semiconductors – although domestic giants like Huawei Technologies have progressed their design capabilities to a degree.
But Ye said an open-source outlook can help overcome some geopolitical constraints, with other international partnerships employed to promote China’s indigenous tech fields.
“We remain confident. The trend of global cooperation and sharing is there,” he said. “With open source, DeepSeek made a formidable debut and had a huge impact. We can pool global intelligence to make DeepSeek better. The closed-source ChatGPT [model] from OpenAI was a historic mistake.”
In one notable concession, several US tech firms began offering the Chinese start-up’s R1 language model in their services this week – including industry titans Amazon and Microsoft.
This article originally appeared in the South China Morning Post (SCMP), the most authoritative voice reporting on China and Asia for…
Read More: DeepSeek’s success gives Chinese firms jolt as Beijing seeks sanction-beating