As the two diametrically opposed labels of a "557% surge" and "missing expectations" are pinned to SK Hynix's single earnings report, the anxiety and excitement of global semiconductor investors are ignited simultaneously.
On July 29, 2026, SK Hynix delivered a quarterly earnings report worthy of making history: revenue of 79.32 trillion KRW, operating profit of 60.54 trillion KRW, and net profit of 93.92 trillion KRW, all reaching record highs. Yet, every single metric in this dataset fell short of the consensus expectations of South Korean securities analysts by approximately 4 trillion to 5 trillion KRW. This "perfect imperfection" has triggered a tug-of-war between bulls and bears in the capital market over the memory cycle and the AI narrative, driving SK Hynix's after-hours stock price in the U.S. market to swing from -9% to +1.7% at one point.
Precise Mismatch in HBM Profitability
The script for this expectation gap was written as early as July 13. On that day, domestic South Korean brokerage KIS accurately forecasted Q2 operating profit at 60.4 trillion KRW, directly triggering a 15% plunge in the stock price. This precision was not prophetic; rather, it stemmed from a deep insight into SK Hynix's unique business structure: an excessively high proportion of HBM, which paradoxically became a "drag" during the price hike cycle.
The market generally considers HBM a highly lucrative product, yet HBM prices and volumes are typically locked in through long-term supply agreements spanning 3 to 5 years. This means that while conventional DRAM and NAND ride the spot market wave with soaring prices, SK Hynix's massive HBM shipment prices remain "locked in" at relatively low levels by these long-term contracts. Consequently, the increase in its overall average selling price (ASP) actually lagged behind Samsung and Micron, which are more heavily exposed to the spot market.
This explains two core puzzles:
Why was the forecast so precise? South Korean brokerages, based on a re-evaluation of long-term supply agreement prices, "normalized" SK Hynix's profit forecasts by lowering expectations for this year and next, thereby aligning closely with the actual financial performance.
After digesting the panic over "missing expectations" at the market open, investors quickly realized: this is not a deterioration in fundamentals, but a structural feature of the long-term contract era. While long-term contracts smooth out short-term price spikes, they deliver exceptionally high earnings certainty. When SK Hynix announced the mass shipment of HBM4 in Q3 and the signing of long-term supply agreements with 10 major customers, the market recognized the long-term value of this "profit stabilizer."
How Long Can the AI Infrastructure Narrative Last?
SK Hynix CEO Kwak Noh-jung previously warned that 2027 will face the most severe memory shortage in history. This statement is both a factual projection and a form of "narrative marketing."
The sustainability of the AI infrastructure narrative hinges on a fundamental question: Can cloud giants recoup the hundreds of billions of dollars they are pouring into GPUs and HBM? As long as AI fails to generate commercial returns at scale, the sustainability of these capital expenditures will be called into question.
In the short term, however, the narrative is far from over. The mass production of HBM4 and the widespread adoption of long-term supply agreements are changing the rules of the game in the memory industry—shifting from a highly cyclical model "at the mercy of market cycles" to an "order-driven" growth logic. As long as the AI arms race continues unabated, SK Hynix's position as the "picks and shovels" provider will remain rock-solid.
Conclusion
SK Hynix's earnings report marks a watershed moment for the memory industry's transition from "cyclicality" to "growth." Although profits missed expectations, long-term contracts have secured the future; the deep V-shaped reversal in the stock price exposes the market's hunger for certainty. In the second half of the AI infrastructure boom, the competition is no longer about who expands capacity most aggressively, but rather who has locked in more orders and whose technology can better weather the cycles.