This sector's correction: "Short when appropriate, go long on rebounds."

Investing.com - Memory chip stocks have fallen 20-30% over the past two weeks, but one analyst believes this sell-off reflects overly exaggerated concerns rather than any deterioration in fundamentals, and expects the “pullback to be temporary, while the rebound will be long-term.”

Learn more about the outlook for the chip sector on InvestingPro

During this period, the shares of Micron Technology, SK hynix, and Samsung Electronics fell by 30%, 24%, and 20%, respectively, under pressure from Middle East geopolitical risk and concerns about Google’s TurboQuant. TurboQuant is an artificial intelligence efficiency technology; some investors worry that this could dampen chip demand.

“Despite concerns about geopolitical uncertainty and interest-rate volatility, 2026 artificial intelligence data center capital expenditures are still continuing to grow, and chip orders for 2Q 2026 are above expectations,” said Jeff Kim, head of research at KB Securities.

“In other words, we believe the recent pullback was driven by profit-taking caused by overblown worries, rather than weakness in fundamentals,” he added.

On the earnings outlook, Kim expects Samsung Electronics and SK hynix’s combined 2026 operating profit to be 397 trillion won, a 337% year-over-year increase. Samsung is expected to generate 220 trillion won in operating profit, with DRAM margins of 74% and NAND margins of 58%. SK hynix is expected to generate 177 trillion won in operating profit, with DRAM margins of 78% and NAND margins of 56%.

This sell-off has pushed valuations to near historical lows. The three chipmakers’ 12-month forward P/E ratios are currently in the 3-5x range—Samsung at 5.3x, SK hynix at 3.7x, and Micron Technology at 3.7x.

“With valuations nearing historical lows, this indicates limited downside potential,” Kim wrote.

Regarding concerns specifically about TurboQuant, the analyst takes a view opposite to the market, saying the technology that improves artificial intelligence efficiency “should lower the entry barrier, while also boosting long-term compute demand, chip content, and ultimately driving demand for artificial intelligence.”

Drawing on historical parallels, he emphasized that the spread of personal computers in 1995 and the launch of the iPhone in 2007 both triggered more than a decade of structural growth. Kim believes the artificial intelligence cycle that began in 2023 may still be in its early stage.

“Given the industry direction that memory chips are being positioned as strategic assets for artificial intelligence infrastructure, we believe the current pullback will be temporary,” he concluded.

This article was translated with the assistance of AI. For more information, please see our Terms of Use.