SEOUL
Samsung Electronics forecast a third consecutive record operating profit for the second quarter at 89.4 trillion South Korean won, or $58.7 billion—a nearly 19-fold increase versus the year-earlier period and above consensus. Revenue is expected to more than double to 171 trillion won.
The stock fell 7 percent during Tuesday trading, with intraday losses reaching 10 percent. SK Hynix declined 6 percent. The Kospi index dropped nearly 5 percent.
The disconnect between earnings acceleration and stock performance reveals a structural shift in how markets price semiconductor beneficiaries of AI buildout. Samsung's operating profit surge stems from sustained demand for specialized memory chips used in AI training and inference—segments where supply remains tight. Analysts expect this demand-supply imbalance to persist.
Yet consensus expectations have inflated alongside earnings. Investors are no longer satisfied by strong revenue or profit growth; they now demand three conditions: durable pricing power, guidance that rules out demand destruction, and proof the AI cycle has years of runway.
Charu Chanana, chief investment strategist at Saxo, identified the pivot: "The memory cycle is still strong, but the market is starting to ask whether the easy part of the trade is already behind us. The question is no longer whether memory demand is strong. It is whether today's shortage could eventually become tomorrow's overcapacity problem if supply comes back too aggressively."
This dynamic extends across AI semiconductors. Nvidia, the sector bellwether, has faced muted reactions to strong earnings in recent quarters as Wall Street raised the bar for "surprise." Nvidia CEO Jensen Huang captured the paradox in November: "If we delivered a bad quarter, it is evidence there's an AI bubble. If we delivered a great quarter, we are fueling the AI bubble."
James Thorne, chief market strategist at Wellington-Altus, characterized the crowded trade as self-defeating: "That is what happens when a bottleneck trade gets crowded: fundamentals stay strong, but earnings stop impressing because perfection was already priced in."
Thorne signals a rotation. The next phase of AI infrastructure deployment will likely reward power generation, grid capacity, cooling systems and physical infrastructure—enablers rather than direct semiconductor plays. "Artificial intelligence is not over," he said. "But easy trade in it is."
