Written by: Rita

Trend Guide

Since June 2025, the memory sector has outperformed semiconductor equipment by a cumulative 661 percentage points. If memory prices correct, will equipment follow suit? This is the most anxiety-inducing question for investors at the moment. Bernstein responded to this question today using historical data, concluding that equipment has its own rhythm, and history has provided answers. To validate this conclusion, Bernstein selected the leading companies in memory and three equipment manufacturers as benchmarks, reviewing over a decade of stock price data and identifying two key periods of divergence.

The historical correlation between equipment and memory is not as high as imagined

Bernstein selected the three largest memory companies (Samsung, SK Hynix, Micron) and the five largest equipment companies (Applied Materials, Lam Research, ASML, KLA, Tokyo Electron) as benchmarks. From 2012 to 2018, the rolling correlation average between the two was only 0.4. Although it rose after 2019, it barely reached 0.6. In comparison, the correlation between equipment stocks and the Philadelphia Semiconductor Index was as high as 0.8 to 0.9 during the same period. Equipment stocks generally follow the entire semiconductor sector, while memory is often the one that deviates.

Historically, there have been multiple instances of equipment rising while memory falls

Bernstein highlighted two typical periods of divergence. From January 2015 to December 2016, the stock prices of the top five equipment companies rose by 21.9%, while memory fell by 16.2%, resulting in a relative outperformance of 38.2%. From January 2021 to December 2022, equipment rose by 15.3%, while memory fell by 34%, showing a relative outperformance of 49%. These two periods of divergence lasted for two years, indicating they were not short-term deviations. Equipment and memory can move in different directions, creating a diversification effect in investment portfolios rather than being passively bound together.

After a 661% excess return, mean reversion may occur

From 2011 to 2019, the long-term returns of equipment and memory were roughly equal. From 2019 to June 2025, memory continuously underperformed equipment, only catching up in February 2026, with both accumulatively rising 36 times over the 15 years. Thereafter, memory accelerated its rise, significantly outperforming equipment by 661% to date. Bernstein's core logic is that if mean reversion occurs, the relative returns in the next phase may tilt toward equipment.

The long-term logic of equipment has not changed

SK Hynix recently announced an additional investment of 100 trillion won (approximately $67 billion) in a new wafer fab in Cheongju, while the South Korean government is considering supporting Samsung and SK Hynix in building wafer fabs in the southwestern region. Even if memory prices normalize by 2027, the combination of logic and foundry investment driving memory expansion will adequately support continued growth on the equipment side. Bernstein expects that the equipment market and company EPS still have room for upward adjustments before 2028.

Trend Perspective

The true judgment is that even if memory remains stagnant, the equipment has its own rising logic, not merely predicting a decline in memory. The $67 billion investment from SK Hynix is already a confirmed project, while equipment orders are certain revenue that do not necessarily correlate with short-term fluctuations in the memory secondary market prices. The market bundles equipment and memory pricing together, but historical data indicates that this bundling is overstated. When certain capital expenditures meet the pressure of 661% mean reversion, the odds for equipment are becoming asymmetric.

Disclaimer

This article is a summary and interpretation of a third-party brokerage research report (Bernstein, July 13, 2026) by Trend Research. The ratings, target prices, earnings forecasts, and related judgments cited within are solely the views of the analysts from the brokerage and do not represent the views of Trend Research, nor do they constitute any investment advice.

The market is risky, and decisions should be made independently. This article should not be used as a basis for buying or selling any securities.