BEIJING – Chinese companies tap the newest artificial intelligence model of Deepseek to see how it can improve productivity.
The Chinese AI model conquered the world in recent weeks after presenting its reasoning process and claims to undermine Rival Rival OpenAi's Chatgpt about costs – despite American restrictions on Chinese access to the advanced semiconductors needed to develop technology.
Eight car manufacturers, including BydAt least nine financial securities companies, three state-owned telecommunications operators and smartphone brand Honor belong to the many who have rushed last week to integrate with Deepseek. Cloud Computing operators Alibaba, Huawei, Tencent and Baidu have all offered ways for customers to gain access to the latest Deepseek model.
“This is pretty unprecedented,” said Wei Sun, chief analyst of artificial intelligence at Counterpoint Research, in an e -mail Monday. She pointed to the speed of adoption, scale of business integration and width of specific industries.
“If we all have this, we know it has a major social and economic impact,” she said.
Optimism about artificial intelligence has spread to Chinese shares. UBS said on Wednesday that AI-related Chinese shares have risen by 15% since the beginning of the year, which performs better than the wider MSCI China index by 9%.
Timing is a big factor in the widespread interest. Deepseek released his latest R1 model on January 20 and the news about his cheap reasoning opportunities led to a global sale of technical shares on January 27 -net when millions of urban employees in China returned to their hometown to celebrate the eight -day moon – New year holiday.
As a result, fewer developed parts of China were given a better understanding of AI and its impact, a subject that was previously limited to conversations in China's largest cities, said Wenhao Zhang, CEO of the consumer marketing consultancy established in Beijing Doodod.
“It is an important education of the market. This will push the development of the entire ecosystem,” he said on Tuesday in Mandarin, translated by CNBC.
Zhang, who studied AI at Tsinghua University, founded Doodod in 2012 to build customer involvement through analysis of social media. He said that the company – that the bank of China Merchant and Toyota counts as customers – started looking at the Deepseek supply at the end of last year and started using the lake after the R1 release at the end of January.
Deepseek, founded in 2023 from a quantitative hedge fund, had released a basic version of R1 in November and a V3 model in December. It launched a smartphone chatbot app in January.
Open-source local implementation
Another attractive factor for companies is that the models of Deepseek are open-source, so that individuals and companies can download and adjust them.
Deepseek also advertised drastically lower prices for applications to use its technology versus that of OpenAi. Chatgpt is not officially available on mainland China and requires users to provide an overseas telephone number and payment method of a supported country such as the US.
Deepseek changed the perception that AI models only belong to large companies and have high implementation costs, said James Tong, CEO of Movitech, a business software company that says that its customers are Danone and China's state grid.
He said Movitech started integrating an earlier version of Deepseek in the fourth quarter of last year, which increased the sale by around 25% from the same period in 2023. The company is planning to have a new deep application at the end of March Launching to improve customers' ability to make decisions, he said.
Many recent videos on Chinese social media have shown how you can perform a local version of Deepseek Apple's Mac Mini.
Apple Mac Mini Online Sales in China climbed considerably from November to January, versus the same period the year before“ According to data from Consultancy WPIC. The JD.com site, aimed at electronics, registered the sale of unity of approximately 20,200 in January, against almost 19,400 in December and approximately 12,250 in November, showed the data.
The affordability of Deepseek is put under pressure from more expensive AI models to lower prices, so that more companies can take over the technology, said Chim Lee, Senior Asia analyst at the Economist Intelligence Unit. He added that open-source models allow financial, banking and health care companies that Subject to strict data protection rules in China – to develop AI applications locally.
“It is still very early to point out concrete business applications, but an important collection meal is that Deepseek will speed up the commoditization of AI,” Lee said.
Beijing also increases support. The National Supercomputing network of China announced on Tuesday that eligible companies and private individuals can obtain three free months from Deepseek access, together with subsidized computing power.
The network is comparable to the Trump-stundled Stargate project of OpenAI in the US for building AI infrastructure with the potential for “even faster scale,” said Winston Ma, deputy professor at the NYU School of Law. He is also the author of “The Digital War: How China's Tech Power is the future of AI, Blockchain and Cyberspace.”
Not centered on Deepseek
The hurry to try Deepseek does not mean that it will be the only AI provider for Chinese companies. Developers in the US and China regularly publish new models.
Movitech also uses the QWen AI model of Alibaba, said Tong, and noted that the market wants the technology that can lower the costs and produce the most results, whether it is open or deep chat.
Hanghang AI, which has invested several hundred millions of Yuan to develop AI solutions for companies in 20 industries, uses a series of models, said partner and COO Shu Weibing.
Many people first used “Baidu, then realized that it wasn't as good as Kimi, then it was not as good as [ByteDance’s] Doubao, who also reduced the prices, “said Shu in Mandarin, translated by CNBC.” Now it is Deepseek. “
It is still to be seen how many generative AI the productivity and profit can increase.
Shu predicts that small companies and companies that integrate AI with hardware will benefit more than large, consumer -oriented internet platforms, whose AI so far, he said, has focused more on stimulating efficiency rather than creating new consumer services.
Despite the falling prices of AI models, “small and medium-sized companies may still be in a period of waiting, Senior Director Analyst at Gartner, said on Wednesday in Mandarin Translated by CNBC.
But the consultancy predicts that by 2027 the average price for access to a generative AI model will be less than 1% of what it costs now – and that by 2029, 60% of Chinese companies have included AI in their primary products and services, Forming the best factors of income growth.