An analyst formerly with MantaNetwork stated that a report released by Korea Investment Securities (KIS) predicts SK Hynix's second-quarter operating profit to be 60.4 trillion won, 8% lower than the market expectation of 65 trillion won. The fundamentals are sound; SK Hynix's profitability remains strong. The impact of long-term HBM contracts on profit elasticity is due to the lack of short-term spot price increases seen in conventional memory. In fact, there is significant disagreement in the market regarding HBM4 pricing. HBM4, with its logic cells, is highly customized. Its actual price depends on many variables, such as: Nvidia's procurement contract price; HBM4 performance, power consumption, and yield; who bears the advanced packaging costs; whether Nvidia is willing to pay a premium for stable supply; and the final order share SK Hynix will receive. These factors are not entirely transparent. Therefore, the agency itself acknowledges that HBM4 pricing depends on the contract with Nvidia, making it difficult to predict. The agency may have used relatively conservative assumptions about HBM prices and profit margins, resulting in a second-quarter profit forecast at the lower end of market expectations. However, HBM4 pricing is not a simple cost-plus model. The statement that "the average selling price of HBM4 will return to the market average after mass production" is more accurately interpreted as not "the price of HBM4 itself returning to the average price of regular DRAM." That's impossible. A more reasonable interpretation is that as HBM4 production ramps up, the overall average selling price growth rate of SK Hynix's DRAM may gradually approach the market average. In other words, the discussion concerns the company's overall average DRAM selling price growth rate, not HBM4 itself becoming the price of regular DRAM.
In short, the fundamentals of memory remain strong; the problem lies in the macro environment.
