Can Start-ups Profit from ChatGPT-Style Services in China?
A debate has sparked among entrepreneurs and investors in China about the business potential of large language models (LLMs) like ChatGPT, following the launch of OpenAI’s chatbot. In China, where the availability of ChatGPT and Google’s Bard is unlikely due to strict content regulations, local technology firms have been striving to develop similar services.
According to a report by the Institute of Scientific and Technical Information of China, the country had at least 79 large AI models with over 1 billion parameters as of May. However, there is uncertainty about whether start-ups have a good chance of profiting from this new AI technology.
Allen Zhu Xiaohu, managing partner at GSR Ventures, expressed concerns that ChatGPT could be a curse for AI start-ups since the service is so powerful that it can outperform many of their existing functions. On the other hand, Fu Sheng, CEO of internet company Cheetah Mobile, disagreed with Zhu’s perspective, believing that ChatGPT holds significant business potential.
Industry experts have weighed in on the matter. Chen Yiran, an electrical and computer engineering professor at Duke University, suggested that while it may be expensive for start-ups to build LLMs from scratch, there are numerous opportunities to develop applications based on LLMs. Hou Zhenyu, a vice-president at Baidu, concurred, stating that it is not feasible for start-ups to develop LLMs due to the high costs of resources like money, graphics processing units, and data.
Consolidation in China’s generative AI industry is already visible, with Meituan, a leading food delivery giant, announcing the acquisition of Light Year, a four-month-old generative AI start-up. ChatGPT, developed by OpenAI, has triggered a global AI arms race, with Chinese firms, including tech giants Tencent and Alibaba, entering the race to catch up with the San Francisco-based start-up.
Li Dahai, Chief Technology Officer at Zhihu, a Chinese Q&A platform, and CEO of Modelbest, a Beijing-based start-up, pointed out that building a successful LLM requires significant resources and involves uncertainties. However, OpenAI remains committed to developing more GPT models. Wang Sheng, an investor with InnoAngel Fund, stated that while there are various layers of AI investment opportunities in China, including infrastructure, large AI models, and applications, it may be too late to invest in LLMs now.
Experts such as Huang Tiejun from Peking University’s Institute for Artificial Intelligence and Qu Kai, founder of 42 Capital, agree that the world likely only needs three large AI models. They believe that the majority of future investments will flow into AI applications.
The debate on whether start-ups can profit from ChatGPT-style services in China continues. While building LLMs may be costly for start-ups, opportunities lie in developing applications based on LLMs. However, it may be too late to invest in LLMs specifically, considering the industry’s current state. Nonetheless, the AI landscape in China remains dynamic, with technology firms vying for a competitive edge in the AI market.
