China's obsession with generative AI has led to the emergence of products almost daily using the technology, whether from startups or tech giants, but investors warn of this state of affairs and see change as cost and profit pressures mount.

The uproar in China over big language models, sparked by the success of OpenAI's Chat GPT program nearly a year ago, has led to what a senior executive at Chinese tech firm Tencent described this month as a "100-model war."

The reason for this name is that China now has at least 130 linguistic models, accounting for 40% of the global total, and slightly less than the US share of 50%.

A "large language model" is a type of machine learning model that can perform a variety of natural language processing tasks, including creating and classifying text, answering questions in a conversational manner, and translating text from one language to another.

In addition to the major core models, Chinese companies have also announced dozens of "language models for a number of industries," derived from 100 basic models.

Tensions between Beijing and Washington have affected the learning models sector (Getty Images)

However, most investors and analysts say they have not yet found viable business models, as most existing models are very similar and costly.

Tensions between Beijing and Washington have weighed on the sector, with U.S. investment funds investing less in projects that manufacture artificial intelligence chips outside the United States.

Asmi Bao, head of internet and digital assets research in China at Macquarie Financial Services Group, said: "Only those with the strongest capabilities will survive."

She added that a number of leading companies have indicated that they will offer competitive prices to gain market share, just as they have done before in the field of cloud services.

"Between the next 6 and 12 months, many large, low-capacity language models will be phased out due to chip restrictions, high costs and stiff competition," Bao said.

Founders and Giants

Opinions differ on which companies are candidates to continue, and Z&Y Capital, head of venture capital firm Yuan Hongwei, said she believes only 2 or 3 large general-purpose language models will dominate the market.

That's why her company looks for founders with experience in this field when deciding which startups to invest in.

Z&Y, whose previous investments include drone maker DJI and self-driving startup Bonnie AI, eventually decided to back Baixuan Intelligence, a 5-month-old company looking to build an open-source AI model to compete with Meta 2.

Baishwan Intelligence looks to build an open-source AI model to rival Meta's model (Reuters)

Baishuan was founded by Wang Xiaochuan, the founder of China's second Suzhou search engine, and Baichuan became one of the first five companies to receive Beijing's approval to launch the chat software. Wang said the company was on track to close the second round of funding, which would be valued at one billion dollars.

"We see an opportunity here, Wang himself is leading this project. Given his understanding of digital business, his success with Sogo and how he is getting industry-wide attention, we think this is our best bet."

Many other entrepreneurs and tech executives are behind new Chinese AI startups, such as former Google China CEO Kai Fu Lee and Yan Guiji, former vice president of artificial intelligence firm Sensei Time.

Observers believe that the largest technology companies in China, such as Alibaba, Tencent and Baidu, have the greatest chances of success, due to the large numbers of users they own and the wide range of services they provide.

Tony Tong, managing director at Gobi Partners, said tech giants have gained an added advantage thanks to the business environment they have developed, noting that some investors regret investing prematurely in AI companies earlier this year, as a number of these startups are currently struggling to build strong business models and are now looking to partner with tech giants to find a larger user base that these companies own or potentially sell to them.