Is the Memory Stock Pullback a 'Fake Fall'? Morgan Stanley: Now is a 'Golden Pit' Entry Opportunity

U.S. memory stocks have recently experienced a sharp sell-off, sparking market concerns about the semiconductor cycle peaking. However, Morgan Stanley (MS) stated in its latest research report that this pullback actually presents an excellent entry opportunity for investors, which it has labeled a 'golden pit'.

Morgan Stanley argues that the market has overreacted to perceived 'negative factors' while overlooking the ongoing and intensifying supply shortages in the data center segment.

Strong Data Center Demand

After visiting data center procurement channels, Morgan Stanley's analyst team confirmed that the intensity of memory shortages in data centers shows no signs of weakening. While signals from consumer electronics markets such as PCs and smartphones are mixed, AI infrastructure demand remains the core driver of this cycle.

Morgan Stanley forecasts that data center memory prices in the third quarter will increase by at least 25% quarter-on-quarter compared to the second quarter—a projection significantly higher than its previous estimates and those of third-party research firms.

Market Overreacts to Growth Slowdown

Regarding recent factors pressuring stock prices—such as slowing growth momentum and rising capital expenditures—Morgan Stanley points out that these were foreseeable 'old news' a month ago. As the memory industry's quarterly revenue climbs from $460 billion to over $2000 billion, a slowdown in growth rate is inevitable and does not indicate a deterioration in fundamentals.

Morgan Stanley emphasizes that the debate should now shift from 'peak earnings' to 'how long high profitability can last,' and expects memory supply shortages to persist through 2027 to 2028.

Memory as a Performance Bottleneck

The report notes that AI compute spending is growing at an annual rate of over 50%, far exceeding the 3% to 5% growth in traditional PC markets. As HBM4 manufacturing complexity increases and future NVIDIA (NVDA-US) Rubin Ultra platforms double memory capacity requirements, memory has become a critical bottleneck in AI infrastructure.

Morgan Stanley believes this structural constraint will support a long-term profit cycle for the industry, providing stronger valuation support than single-year explosive growth.

Focus on Computing and Memory Leaders

In terms of specific investment targets, Morgan Stanley highlights that while memory stocks are rapidly catching up, computing stocks such as NVIDIA and Broadcom (AVGO-US) still offer excellent investment value. Broadcom, in particular, recently secured a multi-year agreement with Apple worth over $30 billion and partnered with OpenAI to launch an AI inference chip designed specifically for large language models.

Most market analysts currently maintain a 'Strong Buy' rating on Broadcom and are optimistic about the prospects of its in-house ASIC chips.