SK海力士計畫在2028年前投資約1300億美元擴充產能，是過去三年投資額的四倍以上。然而，晶片產業向來以繁榮與蕭條的劇烈循環著稱，紐約州已率先限制資料中心建設，AI大廠如Anthropic或Google也可能放緩擴張步伐，這些因素都可能導致供過於求的局面出現。該公司雖嘗試與關鍵客戶簽訂長期合約以穩定營收，但在紐約發行270億美元新股僅三週後即未達盈利預期，顯示其預測能力令人存疑。
從估值角度看，SK海力士目前股價不到預期兩年後盈利的3倍，表面上似乎便宜。但若2028年盈利下降五分之一，本益比約為3倍；若利潤腰斬，本益比則升至5倍，接近其十年均值近7倍。這並不代表SK海力士是糟糕的投資標的，但市場已認清它不再是只漲不跌的穩賺選擇，投資人需審慎評估週期性風險。
SK Hynix shares plunged 10% on Wednesday and have lost 50% in five weeks, as booming sales and operating profit still fell short of overheated market expectations. The broader memory-chip sector, including Samsung Electronics, has similarly retreated. After a year of surging AI-driven demand that propelled earnings estimates and stock prices to extraordinary heights, investor sentiment has shifted decisively toward caution and a reckoning with cyclical risks.
The company plans to invest roughly $130 billion through 2028, more than four times its spending over the prior three years, while the chip industry's notorious boom-bust cycles loom large. New York State has already moved to restrict data centre construction over energy and water concerns, and major AI players like Anthropic or Google could slow their buildout. Although SK Hynix has begun signing long-term contracts with key customers—some including upfront deposits—its failure to meet earnings forecasts just three weeks after a $27 billion share offering in New York raises serious questions about visibility into future demand.
At less than 3 times forecast earnings two years out, SK Hynix may appear inexpensive, but that valuation hinges on sustaining its pricing power and demand outstripping supply. If 2028 profits were cut by one-fifth, the multiple stays near 3 times; if profits halved—a scenario well within historical precedent given that operating profit dropped 90% in 2011—the stock would trade at about 5 times earnings, closer to its 10-year average of nearly 7. The market now recognises that SK Hynix is not a one-way bet, and investors must weigh meaningful downside scenarios alongside the AI-fuelled upside.