Risk averse: China’s private capital unmotivated to fund disruptive innovation applications

There are two major challenges facing China’s current wave of AI startups. I have already mentioned in my previous series of articles the chip industry — which is the foundation of AI — being restricted by the US; the other is the oft-overlooked issue of the lack of venture capital.

From a financing perspective, the last wave of technological advancements in China and the US was actually dominated by venture capital funds from the US. Be it relatively older enterprises like Baidu and Alibaba or emerging giants such as ByteDance, Pinduoduo and Didi, their growth and listing involved US financial capital.

Statistics from Wind show that between 2010 and 2020, 200 Chinese startups were listed in the US, with China absorbing tens of billions of dollars in venture capital each year. Based on Crunchbase statistics, the figure hit a record of more than US$87 billion in 2021. During this period, the hottest stocks on Wall Street were various Chinese concept stocks.

Since the China-US “Cold War” began, the US’s containment strategy has not only included restricting the flow of critical technologies such as chips into China but also limiting the flow of venture capital into China.

The effect of such policies has been immediate — statistics from PitchBook, which tracks startups, show that deals for Chinese startups that included US investors fell 88% between 2021 and 2023, from US$47 billion to US$5.6 billion. That is to say, dollar-dominated venture capital in China is less than one-tenth of what it was at its peak.

However, based on the total amount of capital, the size of China’s private and state-owned capital cannot be underestimated. Unfortunately, for various reasons, this portion of capital rarely flows to disruptive innovation applications in AI, but only to large models with relatively clear applications.

China’s state-owned capital is policy-oriented and currently rides on two fundamental principles: one, to focus on overcoming the US’s chokehold on critical technologies, which obviously does not include innovative applications in large AI models; and two, to avoid high-risk projects with low success rates.

The more disruptive the innovative applications of large AI models, the higher the risk and lower the success rate. Thus, China’s existing state-owned capital almost never dabbles in disruptive innovation applications and only invests in large AI models with clearer applications.

China’s private capital is also not motivated to invest in disruptive innovations like large models. In the field of tech enterprise investments, from the perspective of company stock price or valuation increase, investment in the target company is generally divided into three stages.

In the first stage, the commercial value of the enterprise’s innovative technology has been demonstrated, or there is at least a realisable opportunity for application. If the technology develops smoothly, its market penetration will begin rising steadily from less than 1% at this stage, to around 10% to 25%. Thus, in terms of ultimate benefit, the stock price usually corresponds to a tenfold increase or more. Several investors jokingly refer to this as the “dream-building” stage.

In the second stage, the market share of the tech enterprise or this category of enterprise increases from 10-25% to 90% or more, and the enterprise becomes mainstream and the entire sector’s process of technology renewal is complete. In this stage, the stock price only increases by a few times, and the return on investment for investors begins to drop rapidly.

In the third stage, since the tech enterprises in this category have completed their industrial renewal and replacement, theoretically speaking, the company’s dividend now replaces the company’s growth as an investment rationale.

Disruptive innovation applications such as large AI models are clearly at the primary phase of the first stage at the moment.

For various reasons, listed companies in China’s stock market are subject to numerous profitability requirements, many of which are still reasonable today, such as protecting people’s interests.

However, it is virtually impossible for companies with disruptive innovation applications such as large AI models to be profitable in the first stage, when even the basic profit model is still unclear. Even OpenAI, the global leader in large AI models, has no profit model yet. Thus, such enterprises are unable to get listed in China in the short term.

But in the American stock market, such enterprises can get listed, which is why most of those tech enterprises got listed in the US during China’s internet boom, and why tech elites and experts in China and the world, regardless of whether they have lived in the US, spend most of their time focusing on US stocks.

At the same time, because they are unable to get listed during the first stage, investments in the secondary market are unable to yield dozens of times of “dream-building” returns. Thus, China’s stock market also lacks capital risk appetite, which further reduces the risk appetite of private capital in the country. For these reasons, China’s private capital lacks motivation to support disruptive innovation applications such as large AI models.

The lack of venture capital has thus made Chinese AI teams that choose to embark on the path of disruptive innovation prioritise going overseas to grow their business.

Related: With Sora, is China’s AI falling further behind? | China’s AI race is not about expanding users