Shares of memory chip giants Micron Technology and Sandisk Corporation fell sharply last week after Google revealed details of its TurboQuant algorithm, a compression technology that reduces the memory footprint of AI models by at least six times while boosting inference speeds up to eightfold. The market interpreted this efficiency breakthrough as a potential threat to future demand for memory chips, sending both stocks down more than 15% from recent all-time highs.
Mizuho Securities analyst Vijay Rakesh countered the market's negative reaction, reiterating Outperform ratings on both stocks with price targets of $530 for Micron and $710 for Sandisk. In a client note, Rakesh argued the selloff was "overdone" and advised investors to "buy the TurboQuant memory pullback." His central thesis hinges on Jevons’ Paradox—the economic principle that improvements in efficiency often lead to increased, not decreased, total consumption of a resource.
Rakesh pointed to historical precedents where efficiency gains spurred greater spending: the adoption of virtualized machines, the launch of DeepSeek AI models in 2025, and the shift to optical networking. He contends that TurboQuant will enable larger AI models and faster inference, thereby accelerating AI adoption and increasing overall memory demand. Supporting this view, he noted that NAND memory content in AI servers has doubled in the past year and spot pricing continues to rise quarterly.
Separately, Micron completed a $4.3 billion debt tender offer this week, targeting senior notes maturing between 2031 and 2035. While the move demonstrates financial strength, with the company holding $16.7 billion in cash and equivalents, it also signals an aggressive capital expenditure cycle to capture AI-driven demand. Micron is ramping up investment in DRAM and high-bandwidth memory (HBM) production, including expanding its facility in Taiwan.
Despite record revenue of $23.86 billion and optimistic forward guidance around $33.5 billion, Micron's stock edged lower about 1.8% following the debt tender announcement. Analysts remain split, balancing a bullish long-term AI demand outlook against concerns over memory pricing cycles and the costs of heavy investment.
