Li Feng, Founding Partner of FreeS Fund: Analyzing the Current Cycle Stage of the AI Industry from the Perspective of Capital and Liquidity

"In 2026, the waves in the venture capital circle surged once again: AI moved from a technological concept into the deep waters of the industry, and hard - tech entrepreneurship changed from a 'niche track' to a'mainstream consensus'. Young entrepreneurs are redefining the future coordinates of Chinese innovation with codes and their hands.

Every year, the WAVES Conference hosted by 36Kr · AnYong is the annual vane of the Chinese venture capital circle. This year's WAVES 2026, themed 'This Summer', was held at Liangcang Xinzao Creative Park in Panyu, Guangzhou. Over the course of two days, we gathered top - tier investors, industry leaders, and emerging entrepreneurs. Through 14 in - depth round - tables and dozens of independent speeches, we dissected the underlying logic of core tracks such as AI, hard - tech, going global, and healthcare, and witnessed how the perseverance of those 'few' converged into a wave that changed the industry."

Li Feng | Founding Partner of Fengrui Capital

The following is the content of the speech, edited by 36Kr:

Li Feng: First of all, I'm very grateful to 36Kr for inviting me to give the opening speech. We're just throwing out a brick to attract jade. As you all know, in the past two days, much of the discussion has been about the stock price fluctuations of newly - listed US and Chinese companies and the changes in the capital markets on both sides. So, the 'brick' for today's opening speech is about AI. However, apart from it being a new technological cycle, we won't mainly talk about technology today. Instead, we'll talk about AI from the perspective of'money'.

I don't know how you judge and view the stage of this round of AI and the capital market cycle. But there's a very simple sentence in a book or from historical experience for your reference. All the following content is my personal opinion, just for your reference. The historical experience summary is: In the first half of the first capital cycle of technological innovation, people only talk about technology. In the second half of the first capital cycle, which means before the first capital cycle reaches its peak and starts to decline (not a collapse), people stop talking about technology and only talk about money. In other words, in 2024, people were mainly talking about whose model had better capabilities and whose new version was better. In the second half, people generally only talk about whose market value is how much, how much money someone has raised, and how much money someone has burned recently. At this time point, we'll simply talk about this round of AI from the perspective of money.

First, we need to answer the question of why this round of AI is unprecedented. As we just said, we won't talk about technology today, only money. Here are a few simple economic concepts that I'll briefly explain. In 2020, to cope with the economic impact of the pandemic, all countries around the world adopted a measure. In layman's terms, it's called 'printing money', and in economic terms, it's called over - issuing the base currency. The major central banks around the world expanded their balance sheets by 1.2 trillion US dollars, which is the base currency. What is the base currency? After the central bank prints the money, it gives it to entities such as commercial banks or the fiscal department. Those who receive the money will conduct a recycling process. For example, when a bank receives the money, it will make loans. Those who receive the loans will deposit the unused part back into the bank, and the bank will use the deposited cash for further loans. This is called the circulation in the economy. So, the base currency generally corresponds to a much larger total amount of broad - money.

There's a concept called the money multiplier, which means how many times the base currency is multiplied to get the money we talk about in daily life. The money multipliers of different countries around the world are different. China's financial system is bank - based, so the money multiplier is higher, around 6, 7, or 8. The money multiplier in the US is lower, around 3 or 4. Taking the global unified data and a conservative value of about 4, the 1.2 trillion US dollars of central bank money corresponds to a global money supply of about 40 - 50 trillion US dollars. An increase of nearly 50 trillion US dollars in global liquidity in one year is unprecedented in human financial history.

So much money led to the prosperity of the global capital market in 2021 because there was too much money. In 2022, Europe faced some challenges. The Russia - Ukraine conflict brought huge uncertainties to the underlying energy, the intermediate industrial production, and the upper - level national security or potential national risks. So, Europe became very troublesome. At that time in China, due to reasons such as pandemic prevention and control, it was inconvenient for people to travel, and there was no such convenience of multi - country visa - free travel as there is now. So, like Europe, China also became a place with relatively high uncertainties. If you are in charge of a large amount of money, you usually don't want to put it in a place with extremely high uncertainties. So, at that time, the willingness to allocate money to Europe and China was very low, and money even started to leave these two places.

So, here are three points. First, the appearance of so much money in 2020 was extremely rare in global human history. Second, so much money led to the increase in capital market prices or the prosperity of the global stock market in 2021. Third, after the sudden factors such as the Russia - Ukraine conflict in 2022, so much money around the world happened to be unable to go to Europe or China and basically could only go to the US. So, although the US experienced a sharp interest - rate hike from January to July, the money still flocked to the US after August and September.

When a large amount of money flowed to the US, going back to the first point, no one in the world had ever seen so much money in one year. When so much money went to the US, everything became more expensive, including inflation factors, and the capital market also boomed. When everything is getting more expensive, the capital market needs to find a reason and support for the price increase. Coincidentally, when the global money went to the US, at the end of 2022, a concept and logic suitable for the increase in capital market prices emerged, which is the well - known ChatGPT. So, this round of the AI cycle started around the concept of large - language models. Apart from it being a new technology and a new technological cycle, the reason why the AI boom or the market value of AI - related technology companies has increased so much is probably due to the money or liquidity.

Due to time constraints, we'll just state the conclusion. Currently, the total market value of the US capital market is less than 80 trillion US dollars, and its GDP is about 30 trillion. Its capital market value is less than 2.5 times its GDP. Of course, its capital market value accounts for more than 60% of the global market value, which means one country accounts for about two - thirds. There's a well - known concept called the Buffett Indicator, which is an indicator to measure the overall valuation level of a country's or region's capital market. If the ratio of a country's or an economy's capital market value to its GDP is between 0.8 and 1.2, it's considered a reasonable range. If it exceeds 1.2, it means there is a bubble, and if it's below this range, it means the capital market is not fully developed. Currently, China's Buffett Indicator for the capital market is less than 1, while the US is at a level of about 2.3 - 2.4. We won't discuss the rationality as different people have different views.

Let's talk about a few issues related to today's capital market. In the past two months, a special situation has been observed. In April and May this year, the global liquidity or money was a bit like in the second half of 2023. For different reasons, money started to flow back to US - dollar assets, resulting in some fluctuations. However, the biggest difference between today's situation and the back - flow in 2022 is that there is no large - scale money - printing by any country around the world. So, when the global money is being re - allocated, since the total amount of money remains the same, when A rises, B, C, and D will fall. If B rises, A, C, and D will fall. The total amount of money is fixed. So, in fact, since the second half of last year, the global assets have basically been like a seesaw. If A on the seesaw rises, B, C, and D will fall; if A and B rise, C and D will fall; if B and C rise, A and D will fall. This situation should last for some time because it's a concept of total - amount balance. As for how the money will flow next, we'll leave it to time to answer.

Next, let's list a few figures. Among the US giants, the more conservative ones with the longest - lasting and largest - scale profits are probably Microsoft and Google. These two companies were established early, started making profits early, and have high gross margins. Google recently issued new shares worth 80 billion US dollars. In 2025, Google's net cash reached 164.713 billion US dollars, and its capital expenditure was 91.447 billion US dollars. This year, Google's expected capital expenditure is about 175 - 185 billion US dollars. Let me briefly explain. At Google's current spending rate, mainly on data - center investment and capital expenditure, excluding the 80 - billion - dollar supplementary capital, Google's net cash flow this year may turn negative. By the end of last year, for Google, Microsoft, Amazon, and Meta, when comparing the cash and cash equivalents in the hands of these four giants with their long - term debts (the debts issued for building data centers), Google and Microsoft were in a better situation, with their debts accounting for about 50 - 60% of their cash equivalents. For the other two, it was 70 - 80%, and for other small and medium - sized Internet companies, it was close to or more than 100%. If the current situation continues and excluding Google's 80 - billion - dollar share issuance this year, by the end of this year, the ratio of its debt to cash and cash equivalents will probably be close to or exceed 100%. This is the situation of the capital expenditure of US giants. Last year, the four giants spent about 400 billion US dollars, and this year it is expected to be 750 billion US dollars. Including small and medium - sized Internet companies, it will probably exceed 1 trillion US dollars.

Currently, what the global capital market is concerned about is how long this kind of capital expenditure can last. Let me give an inappropriate example for reference only. During the Internet bubble cycle in 2000 - 2001, which company had the highest market value? Maybe it's a bit unexpected, but the answer is Cisco, with a market value of 555 billion US dollars, which was a world record at that time. Why did Cisco have the highest market value at that time? Because people generally thought that even in the second half of the Internet cycle, even if the Internet model had a bubble, you still had to buy switches and routers, and you still had to buy infrastructure. Moreover, Cisco had tens of billions of US dollars in outstanding orders. So, no matter how high the degree of bubble was, Cisco would definitely be okay, and thus its market value reached 550 billion US dollars.

Today, the company with the highest market value is NVIDIA, exactly ten times the market value of Cisco at that time. The US money supply has theoretically increased by more than four times compared to that year. Many people's logic today is similar. No matter if there is a bubble in AI, you still have to buy GPUs and chips.

The good news is that after the Internet bubble burst, Cisco didn't collapse. Its market value only dropped by nearly 80%, mainly because of those tens of billions of US dollars in outstanding orders. After the bubble burst in 2001, some companies disappeared, and some companies stopped capital expenditure even if they had to lose some deposits. This is what happened at that time.

In today's sharing, the last important small topic is about the base models. People say that this round is different from the previous one. The previous Internet round was about burning money, while this round is about improving productivity. Maybe these statements are all correct. But in the previous round, since no one had used the Internet before, users needed to be educated to use something they had never touched. So, almost all the money burned in the Internet round was used to attract users and let them try the Internet, resulting in negative gross margins on the front - end. Today's situation is different. Base - model companies and large - model companies are currently not making money. Most of the people using AI today have already used the Internet, so in principle, they don't need to be educated on how to use AI. So, today's money - burning is almost not on the front - end but mainly on the back - end. The market may sooner or later require large - model companies to make money. Although the capital investment is large, such as the amortization from large - scale infrastructure, the current data - procurement cost, the large personnel expenditure, and the cost of running the large model itself, after considering all these, if large - model companies want to make money, the cost of tokens and the token - charging price need to be balanced. Once large - model companies stop burning money, how will the token price change? Everyone can make a judgment. And if the token price changes, what changes will it bring to the industry? This is a question that we need to think about individually.

It sounds a bit difficult. What should we do? Generally, investments in all technological cycles are divided into three stages. In the first stage, when an innovative technology emerges, people only care about how different the technology itself is because the technology is still in a stage of rapid iteration. After the technology passes the stage of rapid iteration, it enters the second stage. At this time, people care about what industries an emerging 'technology' is most likely to disrupt and change, or what new things that can use this new technology will emerge. Of course, the most imaginative applications of a technology may be different in China and the US. In these two stages, the value of a company is mainly supported by imagination. The drawback of a large amount of imagination is that it's not particularly easy to quickly implement or make money. Every time a technological cycle reaches the stage where people need to consider making money, after the peak of imagination, people start to say 'calm down'. Regardless of how powerful someone's technology is, we should first focus on who can really make money using this technology. This is probably the third stage of technological evolution.

We don't need to be overly worried. Whether it's big data, face recognition, or autonomous driving, China has an advantage. We've proven that once we pass the stage of original technological innovation and enter the stage where applications drive technological progress, we'll become quite powerful. This is mainly because China has a large number of industrial chains, complete industrial chains, good digital infrastructure, and policy support. In short, when AI + various industries or AI applications start to be implemented, China has an advantage. However, we need to wait until the capital cycle or the imagination cycle reaches its peak and starts to decline before everyone starts to pay attention to the implementation of AI applications. As for when this node will arrive and what stage we are at today, you can make your own judgment.

So, today we've only talked about AI from the perspective of'money', not about technological changes, technological differences, or technological trends. The main conclusion is that in the first half of the first capital cycle of technological innovation, people talk about technology, whether they understand it or not. But in the second half of the capital cycle, people stop talking about technology and almost only talk about'money', whether they are in the technology field or the investment field. These are all my personal views, just for your reference. Thank you for your time."