As OpenAI’s inaugural developer conference displayed the astronomical pace of product development in U.S. high-tech firms, Chinese counterparts in the generative AI space are engaged in a competitive ‘arms race’ to secure funding and catch up with ChatGPT’s advancements.
Chinese generative AI startups are rapidly gaining ground in the industry. Notably, three companies – Baichuan AI, Minimax, and Zhipu AI – have emerged as frontrunners, having secured significant funding. These firms are actively competing to develop and launch their own versions of ChatGPT.
Baichuan AI is reportedly gearing up for its latest funding round. The company, valued at US$1.6 billion pre-funding, plans to raise another US$200 million to US$300 million. Founded in March this year, Baichuan AI had previously completed a funding round of US$300 million, with a pre-funding valuation of US$1 billion and a post-funding valuation of US$1.3 billion. Its investors include major players like Alibaba, Tencent, Xiaomi, Joy Capital, and Shenzhen Venture Capital.
Zhipu AI, on the other hand, announced it had completed a total funding of RMB2.5 billion yuan this year. Investors include Meituan, Ant Group, Alibaba, Tencent, Xiaomi, Kingsoft, Sequoia Capital, and Hillhouse Capital, among others. Sources close to Zhipu AI said the last funding round valued it at approximately RMB12 billion yuan (US$1.65 billion), and it is currently trying to raise another round.
Minimax, valued at around US$2 billion, has completed three funding rounds with investors including Tencent, Hillhouse Capital, miHoYo, Yunqi Partners, and IDG Capital.
Additionally, other generative AI startups such as Moonshot AI is finalizing a new funding round, valued at about US$800 million, with Meituan Longzhu leading the investment. Chinese venture capitalist, Kai-Fu Lee, founded generative AI startup 01.AI and it has completed its funding round at over US$1 billion, led by Alibaba Cloud.
To date, a total of 26 generative AI startups in China received funding this year, primarily in seed or angel rounds. Over 130 companies have launched generative AI-related products this year in China. Chinese AI sector has raised a total of RMB21.4 billion yuan (US$2.9 billion) in funding, with the top three generative AI startups accounting for over 30% of this amount.
Currently, most Chinese generative AI companies are still in the stages of accumulating computing power and running training programs, making financial resources a core competitive imperative. This led to massive financing rounds that investors hope would lead to the emergence of China’s own ChatGPT and the subsequence success.
Chinese generative AI startups lagged their American counterparts not only in product development, but also in valuations. OpenAI reportedly has a valuation of around $29 billion, while Anthropic is reportedly raising at a US$20 billion valuation.
Since OpenAI released ChatGPT less than a year ago, China’s AI startups have witnessed a new wave of investor enthusiasm. The last time the Chinese AI sector was this active was between 2017 and 2018. During those two years, SenseTime, founded in 2014, completed six funding rounds, reaching a valuation of US$6 billion. At that time, SenseTime, Megvii, Yitu Technology, and CloudWalk, known as the ‘AI Four Little Dragons,’ all received multiple rounds of funding.
These four companies didn’t fulfill investor expectations, however. Only SenseTime and CloudWalk successfully went public, but their stock performance has been underwhelming. SenseTime’s current market value is HK$47.2 billion, an 85% decrease from its peak. CloudWalk’s latest market value is RMB13.6 billion yuan, a 70% decrease from its peak.
Both companies have returned to their pre-2018 valuation levels. They remain unprofitable, and according to the financial reports of the two listed companies, their revenue growth has significantly slowed or even declined sharply. SenseTime’s revenue in the first half of 2023 grew by only 1.3% year-over-year, while CloudWalk’s revenue decreased by 58%.
For investors, the ‘AI Four Little Dragons’ are far from a showcase of success. Investors have largely felt "burnt" by their AI investments until their enthusiasm was revived by the recent surge in generative AI.
It looks like the current generative AI investment surge is repeating history. When a new concept becomes hot, investors flock in, driving up startup valuations. Consequently, investor demands for revenue, profit and scale force startups to primarily focus on revenue growth, rather than research and innovation.
In China, there’s a widespread expectation that the broader application market for generative AI would be in the B2B sector. This year, several tech companies have launched vertical generative AI products, attempting to solve industry-specific problems that were previously challenging.
The core competitive edge of vertical generative AI lies in their deeper industry understanding compared to basic models. This implies that the threshold for vertical models is higher than for basic ones. If a basic large model is likened to a generalist, a vertical large model is akin to an industry expert.
This approach, however, has been tested and didn’t prove to be effective. The previous AI wave during 2017 to 2018 showed that serving 2B clients is a difficult path for AI startups. The demand for 2B clients varies greatly, and it’s challenging to scale such a business. In addition, most enterprise clients have also preferred to run their own models for consideration of preserving their sensitive business data.
Will generative AI be different? Investors are betting and hoping for a different outcome this time.