TLDR
- Memory chip stocks declined approximately 10% in the past month, with Micron and Sandisk experiencing double-digit losses following Google’s TurboQuant announcement
- Google’s TurboQuant technology promises to cut AI memory requirements by 6x, triggering investor anxiety over future demand
- Morgan Stanley views the recent decline as a normal market correction rather than evidence of fundamental weakness
- The investment bank argues memory has replaced GPUs as the primary constraint limiting AI infrastructure expansion
- Morgan Stanley maintains Overweight recommendations on both Micron and Sandisk with $520 and $690 price targets respectively
Morgan Stanley continues to express confidence in memory semiconductor manufacturers despite a significant market downturn that shook investor sentiment in late March.
The iShares Semiconductor ETF experienced approximately a 10% decline during the past month. Multiple factors contributed to the retreat, including valuation concerns, demand uncertainty, and emerging AI technologies.
On March 24, Google introduced a compression technology named TurboQuant. The innovation promises to slash memory requirements for operating AI models by as much as six times. The announcement triggered widespread investor anxiety.
Both Micron and Sandisk experienced losses exceeding 10% in the sessions immediately after the disclosure. Micron settled at $357 on March 27, though the stock maintained a 25% gain for the year-to-date period.
Morgan Stanley analyst Joseph Moore challenged the bearish sentiment in a research communication distributed on March 26.
Moore confirmed Overweight recommendations for both Micron and Sandisk. The firm’s price objectives remain unchanged at $520 and $690, respectively.
According to Moore, the market decline represents “a healthy pricing in of durability concerns” instead of a fundamental transformation in underlying demand patterns. The investment bank believes memory manufacturers possess “more durable” competitive advantages than current market pricing suggests.
Memory Emerges as Primary Constraint in AI Infrastructure Development
Throughout the previous two years, Nvidia’s graphics processing units dominated conversations as the critical component enabling AI infrastructure investment. While that narrative maintains validity, Morgan Stanley contends memory has emerged as the true limiting factor.
“Memory is a bottleneck, increasingly the bottleneck, to AI builds,” the research team explained. They highlighted that clients are now committing to prepayment arrangements for substantial volume agreements, indicating how constrained supply has become.
According to Moore, available DRAM capacity has completely evaporated. “Everywhere we look we see indications that it is a true bottleneck,” he stated.
The bank projects AI’s portion of semiconductor expenditure could reach “well north of 50%.” Expanding supply chains appear unlikely to satisfy demand at these elevated levels.
Morgan Stanley’s Assessment of Google’s TurboQuant Technology
Morgan Stanley specifically examined Google’s TurboQuant announcement, arguing the market fundamentally misinterpreted its implications.
The compression technology exclusively targets KV Cache memory, not total memory consumption. “They are just talking about KV Cache memory, not memory overall,” the research team clarified.
KV Cache typically resides in high-bandwidth memory configurations, representing a specialized and constrained category. Morgan Stanley characterized TurboQuant as “normal course productivity improvement” rather than a demand-eliminating breakthrough.
The investment bank acknowledges gross margins approaching 81% probably won’t persist indefinitely. However, analysts identify minimal catalysts for near-term margin compression.
Morgan Stanley additionally emphasized substantial free cash flow generation potential within memory manufacturers. The firm determined that “duration is all that matters,” and by that standard, market indicators “all appear positive.”
Micron and Sandisk retained their Overweight designations as of March 26, 2026.
