Howard Marks' Annual "Bubble" Observation

On December 9, Howard Marks, the founder of Oaktree Capital, released the latest issue of "Is It a Bubble?", directly addressing the core concern of the current market—"Is there a bubble in the AI sector?" Interestingly, this is his second public statement on the matter this year. As early as January 7 at the beginning of the year, he had already started his risk assessment of the stock market with an article titled "On Bubbles." By comparing these two reports, one can not only see his consistent investment philosophy but also gain insight into his valuation perspective on market cognition.

[1. The Anatomy of Bubbles: Psychology Over Valuation]
In these two reports, Marks adhered to his classic analytical framework: the essence of a bubble lies in market psychology, not just quantitative indicators. He outlined a clear path of irrational reasoning:
Spark Ignition: New and revolutionary developments generate extreme excitement;
Idolization: The market blindly worships the asset, believing it "can do no wrong";
FOMO Spread: Widespread "fear of missing out" drives blind capital inflows;
Logic Collapse: Eventually, a market consensus forms that "any price is reasonable," leading to a crash.

[2. A Shift in Perspective: From Macro to AI-Specific]
The key difference between the two reports lies in the granularity of observation:
Early Year (Macro Perspective): Focused more on the overall stock market dominated by the "Magnificent Seven" (M7), emphasizing valuation pressures at the financial market level.
Year-End (Final Focus): The spotlight shifted to AI itself. Marks not only analyzed Nvidia's sudden market cap drop, OpenAI's circular trading, and the dilemma of "high valuation without products" among some companies but also deeply dissected the uniqueness of AI—no longer just a tool but a potential replacement for human cognition. He also sharply pointed out the ambiguity in business models (e.g., ChatGPT's single-query decline) and the uncertainty in profit distribution.

[3. The Taxonomy of Bubbles: Mean Reversion vs. Inflection Points]
By year-end, Marks proposed a classification of bubbles into two types:
Mean-Reversion Bubbles: Such as the 2008 subprime crisis. Based on ideological promises of risk-free high returns, they do not represent human progress but are purely short-term frenzies that reset to zero after bursting.
Inflection Bubbles: Such as railroads, the internet, and today's AI. Rooted in genuine technological advancements, they lay new foundations for human society. The argument: While technology improves the world, the wealth destruction during bubble bursts is real, and the world does not revert to its original state.

[4. Strategic Conclusion: Insight Without Exit]
Rather than saying Marks became more pessimistic by year-end, it's more accurate to say his risk perception became more nuanced and dialectical. As a master well-versed in financial history, he opposes the binary choice of "all-in" or "all-out." He advocates for "rational judgment, selective participation, and prudent choices," leaving behind the classic adage:
"No asset is so good that it can't be overvalued, and no asset is so bad that it can't be undervalued."