In a notable display of resilience, Morgan Stanley has reaffirmed its bullish stance on memory chip giants Micron Technology, Inc. (MU) and Sandisk, despite a sharp selloff that has sent semiconductor stocks into a downward spiral over the past month.
Last month, the iShares Semiconductor ETF experienced a troubling decline of nearly 10%, fueled by concerns over valuations, changing demand trends, and the introduction of new artificial intelligence technologies. On March 24, Google announced the launch of TurboQuant, a compression tool that is reported to decrease the memory requirements of AI models by up to six times, igniting fears of decreased demand in the memory market.
Following this news, shares of Micron and Sandisk both suffered significant losses, plummeting over 10%. Micron’s stock closed at $357 on March 27, although it still registered a notable gain of 25% year-to-date, indicating the company’s strong performance prior to this downturn.
Morgan Stanley analyst Joseph Moore responded promptly by releasing a research note on March 26, reinforcing his Overweight ratings for both Micron and Sandisk. He set ambitious price targets of $520 for Micron and $690 for Sandisk, suggesting that the recent selloff may not signify underlying issues in the memory market.
Moore emphasized that the decline constitutes “a healthy pricing in of durability concerns,” rather than a reflection of diminished demand. He contended that the strength of these memory firms is “more durable than the market thinks,” providing a cautious yet optimistic view of their future.
Memory Takes Center Stage as AI Bottleneck
While Nvidia’s GPUs have historically dominated the conversation around AI infrastructure spending, Morgan Stanley suggests that memory has emerged as a critical bottleneck in the growth of AI technologies. The firm pointed out that “Memory is a bottleneck, increasingly the bottleneck, to AI builds,” highlighting the fact that customers are now prepaying for larger volume deals indicative of high demand and limited supply.
Moore noted that the available slack in DRAM has evaporated. “Everywhere we look we see indications that it is a true bottleneck,” he commented, adding that AI’s contribution to semiconductor spending could climb significantly, potentially surpassing 50%. This escalating demand presents a challenge, as supply struggles to keep pace with the extraordinary needs of AI.
Interpreting Google’s TurboQuant Impact
Addressing the implications of Google’s TurboQuant directly, Morgan Stanley asserted that the market has misjudged its significance. The tool’s effects are limited to KV Cache memory rather than affecting overall memory utilization. “They are just talking about KV Cache memory, not memory overall,” the firm clarified.
KV Cache, which is usually contained in high-bandwidth memory, represents a specific segment of the memory landscape. Morgan Stanley characterized TurboQuant as a “normal course productivity improvement,” rather than a game-changing innovation that could obliterate memory demand.
Although Morgan Stanley anticipates that gross margins hovering near 81% may not last indefinitely, the firm sees little reason for them to decline in the near term. Additionally, they highlighted the potential for robust free cash flow generation from memory companies, ultimately concluding that “duration is all that matters,” with current indicators suggesting positive trends ahead.
As of March 26, 2026, both Micron and Sandisk hold their Overweight ratings, signaling that even amidst market turbulence, the future remains bright for these key players in the semiconductor industry.
