Can we still buy memory stocks after the crash? Samsung Securities analyst: It's a correction within the cycle, not a peak in the economy.

Samsung Securities semiconductor analyst Jong-wook Lee published a research report on March 30, stating that although recent DRAM spot prices have weakened and memory stock prices have clearly fallen, this downturn is a normal adjustment within the cycle rather than a sign of a turnaround in business conditions. He suggests that investors get through the correction and seize this risk-reward opportunity.

Samsung Electronics and SK hynix shares plunge nearly 10% over five days

Over the past five trading days, Samsung Electronics (KRX: 005930) shares have fallen 9.82%, while SK hynix (KRX: 000660) has seen an even steeper decline of 12.08%. The two major memory leaders continue to slide today, closing down 1.69% and 5.31%, respectively, drawing market attention.

The trigger for this wave of selling pressure comes from a pullback in prices in the DRAM spot market. After DDR5 16GB spot prices peaked on March 19, they continued to correct. The premium of spot prices versus contract prices has narrowed from an average of about 30% in the first quarter to roughly 20% recently. Investors are worried whether this change signals that memory stocks are entering a downward cycle.

Analyst interpretation: This is a correction, not a peak-out

In response, Lee clearly said that this selloff is a “correction” rather than a “peak-out.”

He noted that the market narrative is gradually shifting from “how much DRAM prices will rise” to “whether DRAM prices can keep rising.” During the transition between these two themes, the market finds it difficult to accurately catch the turning point, making volatility unavoidable. For that reason, he believes that holding back during the correction to wait for clearer signals can deliver a more favorable risk-reward ratio, whereas exiting too easily may mean missing out on the subsequent rally.

Supply reaction lag creates a potential “plateau”-type earnings trend

His argument is based on structural changes on the supply side of this memory cycle. Lee said that, compared with prior memory upturns and cycles, the capacity expansion response on the supply side in this round has been delayed by about one full year. This implies that the imbalance between supply and demand will persist for a longer time.

Therefore, he expects the earnings curve of memory manufacturers will not form the past “spike-and-plunge” pattern, but instead move into a comparatively steady “plateau” trend, helping investors build stronger confidence in the continuity of the cycle, and also making valuation support more solid.

The real signal that the business cycle has peaked is tied to AI demand

Lee also reminded that DRAM is ultimately a cyclical industry, so investors should not ignore cyclical risks. However, they also should not rush to exit due to any vague fears.

He emphasized that the true signal of the peak in memory business conditions must come from a structural shift in demand from downstream industries. Specifically, investors need to closely watch generated AI leading companies such as OpenAI and Anthropic, along with business developments and capital expenditure plans from major internet service providers (ISPs).

As of now, AI infrastructure investment is still accelerating, and there is still room for further upward movement on the demand side. There has not been any sign of a peak in business conditions.

(Google’s new technology scares the market—AI memory demand is down six times! SK hynix and Micron both plunge)

Concerns about natural gas supply are a cost issue, not a production crisis

Another concern in the recent market is that tight natural gas supply may impact memory production. In response, Lee believes the issue is being overinterpreted by the market. He said that the natural gas supply problem is, at most, pressure from rising costs—not the risk of production disruption. The nature of the issue is fundamentally different.

For the memory industry, its cost structure has long been mainly driven by process technology evolution, improvements in wafer yields, and the scale of capital expenditures. The impact of raw material price fluctuations has been relatively limited. In addition, in a strong-demand market environment, increases in raw material costs are typically able to be passed through by adjusting selling prices, so the erosion of overall profitability is limited.

Selling pressure led by sentiment; fundamentals have not deteriorated yet

Based on the analysis above, Lee defines this stock-price decline as a “sentiment problem,” not a “fundamentals problem.” He said that after DRAM prices and memory stock prices both surged significantly, investors developed intense anxiety about whether business conditions can keep improving. They are looking for concrete data to resolve uncertainty, and the temporary weakness in spot prices has become an outlet for this anxious sentiment.

He expects that once the market direction after the third quarter shows verifiable and specific supporting evidence—including AI server procurement trends, the inventory levels of major customers, and the outcomes of contract price negotiations—memory stock prices will likely stabilize and begin reflecting companies’ true earnings power again.

This article Memory stocks collapse—can they still be bought? Samsung Securities analyst: Correction within the cycle, not a peak in business conditions is first published on Chain News ABMedia.