🤖 AI Curated
Samsung Electronics reported a Q2 2026 operating profit of 89.4 trillion won (up 1,810% year-over-year), surpassing Nvidia to become the world's top tech company in a single quarter. However, on the day of the announcement, its stock price plummeted 6.92% due to a "sell-on" phenomenon. This highlights the two diverging faces of the AI memory supercycle, where strong performance didn't align with stock market reaction.
Samsung Electronics announced preliminary results for Q2 2026 (April-June), with sales of 171 trillion won and an operating profit of 89.4 trillion won. Compared to the same period a year ago, operating profit soared by over 1,810%. Analysis suggests that this single quarter's operating profit more than doubled last year's full-year operating profit (approximately 43.6 trillion won) and even exceeded the combined profits of the past three years. Some estimates suggest that if performance-based bonus provisions (around 20 trillion won) are excluded, the company's actual financial strength was in the 100 trillion won range.
What particularly caught attention was that Samsung surpassed Nvidia, often called the "AI leader," in operating profit. Nvidia's previous highest quarterly operating profit was around $53.5 billion (approximately 82 trillion won), but Samsung exceeded this, becoming the tech company with the highest operating profit globally in a single quarter. It was also reported to have surpassed Apple's highest quarterly operating profit.
Most of this performance stemmed from semiconductors, specifically memory. It's estimated that well over 90% of the total operating profit came from the memory business division. The boom in AI data center investments led to an explosion in demand for server DRAM and HBM (High Bandwidth Memory). This, coupled with supply shortages, significantly drove up prices for general-purpose DRAM and NAND flash. As of Q2, DRAM prices reportedly rose by over 40%, and NAND by around 50%. Samsung had announced the start of HBM4 mass production in February, further boosting the effect of expanding high-value products.
However, on the day of the announcement, the stock price moved in the opposite direction. Samsung Electronics closed on July 7th at 296,000 won, down 6.92%, and at one point during trading, it dropped by about 10% to 286,000 won. SK Hynix also fell by around 6%, and the entire KOSPI index plummeted 4.91%, leading it to be dubbed "Black Tuesday." In essence, the market screamed in pain on the very day the company reported its highest-ever performance.
Securities firms interpret this as a "sell-on" phenomenon. The expectation of strong earnings had already been sufficiently reflected in the stock price, so when the results were actually announced, profit-taking sales flooded the market. This was compounded by "peak-out" concerns, a worry that if big tech's AI investments eventually slow down, the memory supercycle could also falter. There were also criticisms that labor-related costs, such as performance-based bonus provisions, constrained profit distribution.
However, the industry outlook isn't entirely bleak. Some observations suggest that memory supply shortages could persist until the end of 2027, and the period of greatest performance improvement might be from Q4 2026 to H1 2027. Ultimately, this "two-faced" situation illustrates a facet of the AI memory supercycle: that performance (fundamentals) and stock price (market sentiment) don't always move in the same direction.
This case demonstrates that the common wisdom "good performance always leads to higher stock prices" isn't always true. It's an industry news piece that helps general readers understand the "sell-on" phenomenon, which occurs when expectations are already priced into a stock, and the "peak-out" debate within the semiconductor cycle.
🤖 AI-curated from multiple sources. Verify accuracy with the originals (sources).