<p>As the year-end approaches, it's time to review and backtest. A month ago, I wrote an article titled &#34;AI Bubble, Bubble AI,&#34; and the scenario seems to be accelerating, raising concerns that it may soon become a reality. According to foreign media reports, Microsoft founder Bill Gates warned that some overvalued AI companies will suffer losses in the fiercely competitive AI industry, as investors are assessing whether the bubble is about to burst.</p>
<h2>Identifying a Bubble is Easy, Judging When it Will Burst is Difficult</h2>
<p>Currently, the market is once again showing similar concerns. AI concept stocks have become the focus, with Palantir's valuation exceeding expected earnings by 200 times; the valuation of the S&amp;P 500 index relative to real earnings over the past decade is second only to the highs of 1999 to 2000, with the price-to-sales ratio even exceeding that period by more than 60%. A slight pullback in tech stocks (like in November) has triggered a surge in volatility, and cryptocurrency enthusiasts overlap with AI investors, showing a high correlation in risk appetite. How to judge when the bubble is about to burst?</p>
<p>An asset bubble refers to asset prices significantly deviating from their fundamental values, primarily driven by speculative behavior, excessive optimism, and an influx of capital. Once a bubble bursts, prices often plummet, causing widespread losses. Historically, the tech stock bubble of 2000 is a classic case. Identifying a bubble is easy, but judging when it will burst is difficult, which is why in recent years, phrases like &#34;the bubble is forming&#34; and &#34;embrace the early bubble&#34; have become somewhat cliché; even professionals may exit too early and miss out on substantial gains.</p>
<p>High valuations are a common characteristic of bubbles and even a condition for their occurrence, but they cannot accurately predict the timing of a burst. Indicators like the cyclically adjusted price-to-earnings ratio (CAPE) are more effective in predicting ten-year returns, showing that the higher the starting valuation, the lower the long-term returns. There is almost no correlation between valuation and returns within a year. Data shows that from 1990 to 2024, stock price performance in the year following high CAPE years shows no significant pattern. The current valuation of the S&amp;P 500 has reached extremes, even more so for AI stocks, but this only indicates low long-term returns and does not mean an immediate market crash. If investors rely solely on valuations, they may exit too early like the bears of the late 1990s.</p>
<h2>Experts Who Predict Correctly Often Cannot Hold On</h2>
<p>Another strong signal is that professional investors are forced to close their funds due to counter-market operations. Back in 2000, on the eve of the Nasdaq crash, legendary investor and Tiger Fund founder Julian Robertson faced withdrawals of up to $21 billion from clients for avoiding tech stocks, ultimately leading to his fund's closure. He had generated an average annual return of 25% for clients since 1980 but was forced to exit entirely at the peak of the bubble. Ironically, those experts who correctly predict bubbles often cannot hold on until the end.</p>
<p>The movie &#34;The Big Short&#34; tells the story of legendary hedge fund manager Michael Burry, who successfully predicted the 2008 mortgage crisis. This year, he also publicly shorted AI stocks, including Palantir and Nvidia. However, by the end of October, he announced the closure of his fund, citing a severe disconnect between his value estimates and the market. This is a typical example of how short sellers are ruthlessly eliminated in a bubble environment. The increasing occurrence of such &#34;intolerance for dissent&#34; events suggests that the market may be nearing its peak</p>
<h2>Beyond Valuation: More Reliable Market Signals</h2>
<p>This article aims to explore more reliable short-term market signals beyond valuation, including public attention peaks. The Economist describes it as the &#34;taxi driver indicator&#34; (the modern version of casual discussions about stocks) that relies on monitoring Google search volumes. When internet users search for specific investment fads (such as NFTs, Bitcoin, SPACs, ARKK, cannabis stocks, wearables, space stocks) reach a high point, it often indicates that prices are about to adjust. As search volumes peak, asset prices tend to drop significantly in the following 12 months, with price movements occurring almost simultaneously.</p>
<p>The annual word for 2025 is &#34;slop,&#34; referring not to a &#34;kill pig plate&#34; but to a &#34;feed pig plate&#34; that actually represents AI algorithms. These algorithms are not designed for human observation; their true target audience is the algorithms of online platforms and search engines, meaning they are meant to cater to machines rather than us humans. Similar to the dark traffic generation tactics of the Yuen Long Da Bao pan-fried meat pie rice lurking within Threads, in the financial world, this can distort or amplify traditional quantitative models, causing asset prices to choke deeper and faster.</p>
<p>The reason lies in the fact that retail investors often rush in at the last moment, fulfilling the self-fulfilling prophecy of &#34;buying the last bite,&#34; followed by panic selling that triggers a crash. Although not all peak search volumes necessarily lead to asset price declines (for example, AI stocks continued to rise for several weeks after peaking in mid-August), when search volumes drop sharply while asset prices are volatile, it inevitably causes unease.</p>
<h2>AI Stocks and Bitcoin: Similar Investor Backgrounds</h2>
<p>The search volume for AI stocks has significantly declined from its peak, with Nvidia's stock price dropping by 13%. Bitcoin has fallen over 25% since early October; although thematically not directly related to AI, the investor backgrounds are similar and closely linked to overall risk appetite.</p>
<p>Bitcoin is often seen as a risk appetite indicator, with its bubble bursting accompanied by a chain reaction. Data shows that Strategy Company (formerly MicroStrategy) is a typical case. The company has been heavily in debt to buy Bitcoin since 2020, holding about 650,000 coins (accounting for 3% of the total), almost becoming a &#34;Bitcoin reactor.&#34; By the end of last year, this strategy seemed successful: the company's stock price rose nearly 600%, far exceeding Bitcoin's own nearly 200% increase.</p>
<p>However, since early October, as Bitcoin's price dropped by a quarter, the company's stock price fell over 40% during the same period, with its market value at one point falling below the value of its Bitcoin holdings. The company needs to pay $800 million annually in dividends and interest, with almost no other sources of income. Although it raised $1.4 billion by issuing stock in early December (equivalent to 21 months of dividends) as a buffer and announced a potential sale of its Bitcoin holdings (contradicting its &#34;never sell&#34; promise), the convertible bonds maturing in 2027 will bring greater pressure.</p>
<p>Research data indicates that the total amount of all trading orders currently is less than $600 million, and the liquidity in the Bitcoin market is becoming thinner, making large sell-offs more likely to trigger sharp price drops (for example, on October 10, $19 billion in leverage was liquidated). Moreover, the details that make speculative behavior more dangerous are usually not noticed during a bull market and only come to light after a crash
Such traffic-based, machine-generated false information may lead investors to make wrong buying and selling decisions.</p>
<p>You play AI, AI plays you.</p>
<p><strong>JiuLiSheng</strong></p>
<p>Facebook page</p>
<p>Related articles: Key points for selecting two major Hong Kong financial stocks | JiuLiSheng</p>

PLTR

MSFT

<p>Microsoft founder Gates warns that competition in the AI industry is fierce, and some overvalued AI companies may suffer losses. AI concept stocks like Palantir are overvalued, and the S&P 500 index is close to historical highs. Identifying bubbles is easy, but predicting their burst is difficult, and investors need to be cautious</p>

- The article discusses concerns about a potential bubble in the AI sector, highlighted by Bill Gates' warning about overvalued AI companies facing losses amid fierce competition.  
- It notes that AI stocks, such as Palantir, are experiencing extreme valuations, reminiscent of the dot-com bubble, with indicators suggesting a possible market correction.  
- The author emphasizes the importance of monitoring public interest and search trends as reliable signals for market movements, particularly in relation to AI and cryptocurrency investments.

You play AI, AI plays you | Jiu Li Sheng