The Korean Stock Market Paradox: Soaring Profits Yet Valuations Hit Bottom

Event Background

In the first half of 2026, South Korea's Composite Stock Price Index posted cumulative gains of approximately 80%, repeatedly setting new record highs. Two memory chip giants—Samsung Electronics and SK Hynix—saw profits surge dramatically, driving the index's forward P/E ratio down to just 6.4x—a level even lower than during the 2008 global financial crisis.

This figure seems counterintuitive: stock prices are rising, profits are growing, yet valuations (P/E ratios) are hitting historic lows. Recent market skepticism about artificial intelligence stocks has triggered fresh sell-offs, further depressing valuations.

The Nature Behind Surface Contradictions

This isn't a contradiction but rather a mismatch between profit cycles and valuation cycles.

Imagine two curves:

• Profit curve: In 2024-2026, memory chips experienced explosive demand from AI, with supply far short of demand. Samsung and SK Hynix's profits surged 150%-200% from their troughs.
• Valuation curve: The market measures this with the P/E ratio. P/E = Stock Price ÷ Earnings Per Share. When the denominator (earnings) grows faster than the numerator (stock price), the P/E ratio falls.

In this case: stock prices rose 80%, but earnings may have grown 150% or more. As a result, the P/E ratio collapsed from historical normal levels (15-20x) to 6.4x.

Why Does This Reversal Occur?

The market is pricing in fear of a cycle turning point.

The implicit assumption of investors is that this round of memory chip prosperity (driving earnings growth) is temporary.

• AI chip demand may have peaked (or is extremely overexpected)
• Memory production capacity is being rapidly deployed, competition is about to intensify
• Consumer electronics recovery momentum is fading
• Once demand softens, these super-high profits will collapse quickly

So the market's logic is: "I don't deny that current earnings are very high, but I'm applying a deep discount to these super-high earnings—because I believe they won't last long."

Two Interpretations

Interpretation A (Bull Case):

The market is excessively pessimistic. Memory chips are experiencing structural rather than cyclical prosperity in the AI era. Earnings could persist for 5-10 years. Buying a company with long-term high earnings at a 6.4x P/E ratio is actually gaining an implicit yield of 15%+ (reciprocal relationship). This is undervalued.

Francis Tan, Chief Strategist for Amundi Wealth Management Asia, represents this camp. He emphasizes that "corporate profit fundamentals are sound, and the market expects continued strong earnings going forward," meaning this elevated profit level won't be fleeting.

Interpretation B (Bear Case):

The market's fear is not unfounded. Memory chips have never historically experienced a super-cycle lasting 10 years. Every boom has been accompanied by capacity deployment, price competition, and ultimately collapsing profits. The P/E ratio compression now precisely signals that smart money is gradually reducing positions—because they see the other side of the cycle approaching.

Deep Principle: The Valuation Trap of Cyclical Businesses

The core of this phenomenon stems from an ancient investment dilemma: how to value cyclical businesses?

Consumer staple companies (Coca-Cola) can maintain stable P/E ratios of 25-30x because their earnings are relatively stable and predictable.

But memory chips are different. They are classic cyclical commodities: supply is constrained by physical cycles (factory construction takes 3-5 years), while demand is influenced by macro conditions. The result is prices swinging from astronomical levels to below cost, then bouncing back. A complete cycle may span 7-10 years.

The reason investors apply discounts to memory chips is:

1. Non-normalization of profits: This year's 150% net profit growth cannot be used to extrapolate next year's earnings 2. Subjectivity of cycle position: No one can precisely predict whether "high prosperity will last 1 year or 5 years" 3. Historical lessons: The 2000-2001 chip depression, the 2015 memory chip collapse—each arrived suddenly

Three Meanings of Low Valuations

Meaning 1: Warning of Value Traps

If you had seen this combination of "low P/E + high earnings" at memory chip stock bottoms in 2000, 2008, or 2015, you would have excitedly bought in. Then you would have been trapped for 3-5 years. Because earnings would collapse with cycle contraction—P/E ratios would spike from 6x to 100x (as both numerator and denominator deteriorate).

Meaning 2: Opportunity for Excess Returns

If memory chip earnings can truly sustain for 5-10 years (this time is different), then the current 6.4x P/E is the historical lowest margin of safety. You can buy a long-term high cash flow business at a bargain price, capturing implicit returns of 15%+.

Meaning 3: Market Perception Rift

The 80% stock price gain shows the market acknowledges "short-term demand is strong." The 6.4x P/E shows the market doubts "this won't last." Both could be right—near-term is indeed strong, long-term won't persist. The question is how long "long-term" is: 1 year? 3 years? Or 10 years?

Lessons for Investors

This case teaches us that we cannot invest in cyclical commodities using the simple rule of "low P/E = cheap." We must ask:

1. Are current high earnings at a cycle peak, or a structural new high? 2. If they revert to normal levels, how much will profits fall and how quickly? 3. Can the company soften cycle impacts through cost control and differentiation? 4. How long is my holding period—can I endure 3 years of decline?

For memory chips, the critical variables are: the long-term demand trajectory for AI chips, the pace of capacity deployment, and companies' cost-side competitive advantages. Yet none of these can be read from the current P/E ratio figure.