What is a bubble?
To identify a bubble, we first need to understand what a bubble is.
Bubbles often originate from the emergence of new technologies. The market becomes overly optimistic about the future development of the technology, leading to excessive investment and blind following. This causes its value to exceed what the real economy can bear, followed by a sharp decline, and finally bursting like a soap bubble.
Combining Hyman Minsky's "Financial Instability Hypothesis," Jordi Gali's "Monetary Policy and Rational Asset Price Bubbles," and other classic papers on economic bubbles, we summarize the core conditions for bubble formation.
These mainly include: favorable economic fundamentals for investment, the emergence of information asymmetry, and the inflationary effects of psychological and behavioral factors. In simple terms: the market has money and investors invest irrationally.
First, the market must have money, which means there must be ample liquidity. A basic economic situation of credit expansion and excess liquidity in a low-interest-rate environment can trigger a bubble.
For example, in 2022, we experienced a period called the "everything bubble." To respond to the economic downturn caused by the pandemic, the Federal Reserve implemented near-zero interest rates and quantitative easing (QE) from 2020 to 2021. This measure attracted investors to make riskier investments and allowed unsustainable business models to develop based on low-interest loans. Almost all stock market assets appreciated rapidly, repeatedly breaking U.S. historical records. Until 2022, when the Federal Reserve raised interest rates again to curb inflation, the stock market plummeted, with Google falling 40% and Tesla and Meta stock prices dropping 60% within a year.
Second, investors invest irrationally. New technologies allow investors to obtain quite high returns through early investment. The monopolistic nature of certain tracks makes their potential future returns even higher. Sufficiently high profit margins lead to blind optimism in the market, causing investors to underestimate risks and overestimate returns.
For example, the internet bubble that burst in 2000. In 1995, a large amount of venture capital poured into internet-related fields such as e-commerce, telecommunications, and software services, with investment returns far exceeding other industries like chemicals, energy, and finance. When speculators noticed the rapid growth in stock prices, they expected further rises and bought in. In 1999, the investment amount in U.S. internet-related industries reached $28.7 billion, nearly 10 times that of 1995.
What's the ceiling for AI investment?
Remember the two prerequisites for bubble formation we mentioned above? The first is that the market must have money.
However, the current liquidity in the U.S. financial market is not optimistic, which means the ceiling for the AI bubble can't be too high.
Regarding this, Weiming Xiong, partner at Huachuang Capital, points out: "The extent of this bubble is actually far less than the internet bubble 20 years ago, or even the cryptocurrency bubble in 2017, or the NFT bubble in 2021. These bubbles were characterized by valuations far exceeding the investment return cycle that actual products and services could obtain.
If measured in proportion, I think the extent of this bubble might only be 20% to 30% of the dotcom or NFT bubbles. The degree of this bubble is absolutely not comparable to the previous ones."
The financing environment in the past two years has been relatively poor. To curb the highest inflation in 40 years brought by monetary easing during the pandemic, the Federal Reserve has raised interest rates 11 times from March 2022 to July 2023.
Meanwhile, the Federal Reserve also began large-scale balance sheet reduction. Starting from June 2022, the Fed reduced its holdings of Treasury bonds by $60 billion and mortgage-backed securities (MBS) by $35 billion per month.
In one sentence, during the AI boom, the Federal Reserve is implementing the most aggressive monetary tightening policy since the 1980s.
The market has no money, and even though almost all VCs are caught in FOMO, the overall trend of U.S. stock venture capital is still declining rather than increasing. According to Crunchbase data, global financing in the first half of this year decreased by 5% year-on-year.
Of course, AI startups among them stood against the trend, increasing by 24% year-on-year, even receiving the largest quarterly investment of $24 billion in the second quarter of this year, but the total value is still only 70% of that in 2021.
This is because the easing during the everything bubble period in 2021 brought massive liquidity, and its aftereffects have not yet dissipated. The market is not as wealthy as in 2021, but it's still quite wealthy.
Weiming Xiong compares: "In the past two years, AI may have reached its peak from a capitalization perspective. In 2021, the U.S. issued $6 trillion in half a year, which is unprecedented in human history. This kind of capital ripening effect is unprecedented."
However, VCs are holding their money much tighter than in 2021.
From the data provided by COATUE, although this round of AI investment is lively, VCs have not gone all out. Private equity firms still have $1 trillion in uninvested funds, at a historical high level.
This is mainly for two reasons.
One is that exit paths are not smooth, making VCs hesitant to invest. After the last round of the "everything bubble," the number of unicorn companies surged from 67 in 2016 to 580 in 2021. However, their rate of obtaining refinancing has been declining sharply. From 2016 to 2022, the proportion of unicorns obtaining refinancing in the same period dropped from 50% to below 20%.
What about IPOs? It's even more dismal, with basically only single-digit numbers since 2022.
"In fact, there were 970 IPOs in the U.S. stock market in 2021, which dropped to 162 in 2022, and only about 44 in the first half of this year. This indicates a clear trend of contraction in the global capital market."
In this situation, the only exit method left is through mergers and acquisitions. This path is too narrow.
Another reason is that the current stage of AI development has a higher investment threshold, limiting many VCs from entering the field.
"The early internet industry needed to build its own servers and infrastructure, similar to today's AI field. The cost of running a large model ranges from tens of thousands to hundreds of millions of dollars, at the early stage of new infrastructure construction."
We find that most of the money entering the artificial intelligence field flows to foundational layer companies, that is, the large model companies we are familiar with, such as OpenAI, Anthropic, Gemini, etc.
They then use this part of the funds to purchase chips from computing layer companies like NVIDIA to train their large models.