The Disenchantment of Chinese Investors with ai Language Model (LLM) Startups
In recent times, there has been a noticeable shift in the attitudes of Chinese investors towards ai startups specializing in large language models (LLMs). Two prominent figures, Allen Zhu Xiaohu and Baidu CEO Robin Li Yanhong, have publicly expressed their concerns about the viability of investing in these startups.
Allen Zhu Xiaohu: No Business Sense in Investing in LLM Startups
Allen Zhu Xiaohu, a well-known early investor in Didi Chuxing, China’s ride-hailing giant, recently voiced his skepticism about investing in LLM startups. In an interview with Tencent News, Zhu stated that it makes no sense to invest in such companies as they lack the necessary application scenarios and data required for a successful business model.
Zhu had previously invested in two ai startups: an ai job interviewer and an ai advertising producer, before the world witnessed the release of OpenAI’s ChatGPT in November 2022. This event fueled the global craze around ai, but Zhu did not invest in any LLM startups on the Chinese mainland last year and has no plans to do so.
“I believe in applications,” Zhu emphasized, adding that those who continue investing in LLM startups are doing so out of the fear of missing out on potential opportunities.
Baidu CEO: LLM Startups a “Huge Waste of Resources”
Robin Li Yanhong, the CEO of Baidu, China’s leading internet search engine, echoed similar sentiments about LLM startups. In an interview in October 2023, Li stated that these startups represent a “huge waste of resources.”
According to Li, as of October 2023, there were 238 ai models launched in China, but there was a conspicuous lack of successful ai applications based on these models. Li urged the government to initiate policies that would encourage the development of more ai-native applications, leading to a thriving ai ecosystem and driving a new wave of economic growth.
“There are too many big models in China,” Li acknowledged, “but there are too few ai-native applications based on these models.” He concluded by stating that if the industry could focus more on creating ai-native applications, it would lead to a prosperous ai ecosystem.
