Key Takeaways
- Shares of Micron and SanDisk experienced significant declines following Google’s disclosure of TurboQuant, an algorithm that reduces AI memory requirements by a minimum of 6x
- Market participants worried that TurboQuant would diminish memory chip demand, causing both equities to retreat more than 15% from their recent peak levels
- Vijay Rakesh from Mizuho Securities maintains that the market reaction is excessive and reaffirmed Outperform recommendations for both companies
- According to Rakesh, efficiency gains such as TurboQuant typically drive greater AI deployment, ultimately increasing memory consumption—a phenomenon called Jevons’ Paradox
- NAND content in AI server configurations has expanded twofold over the last twelve months, with spot market prices climbing consistently each quarter
Shares of Micron and SanDisk experienced sharp declines in the previous week after Google disclosed technical specifications for TurboQuant, an advanced compression algorithm that reduces the memory requirements of artificial intelligence models by a factor of at least six. The technology also delivers inference speed improvements of up to eight times while maintaining model precision.
Investors interpreted this development negatively for semiconductor memory manufacturers. The implication that AI systems would require less memory per model suggested potential weakness in chip demand from industry players such as Micron and SanDisk.
Both equities have declined by no less than 15% from all-time peak valuations reached in the final days of the prior month. Trading activity on Thursday showed SanDisk down 5.9% at $652, with Micron falling 5.5% to reach $347.78.
Compounding market pressure, President Trump’s Wednesday evening remarks created uncertainty regarding the timeframe for resolving the Iran situation, which dampened overall market confidence entering Thursday’s trading.
TurboQuant research initially emerged in 2025, with updated findings on AI inference capabilities recently published by Google’s research team.
Mizuho’s Case Against the Market Panic
Vijay Rakesh, an analyst with Mizuho Securities, challenged the market’s interpretation in a client communication. He maintained Outperform ratings for both Micron and SanDisk while preserving price objectives of $530 and $710 respectively.
Rakesh advised clients to “buy the TurboQuant memory pullback,” characterizing concerns about peak memory consumption as “overblown.”
His central thesis centers on historical patterns showing that AI efficiency improvements have consistently resulted in increased spending rather than reduction. This concept is recognized in the technology sector as Jevons’ Paradox—when a resource becomes more economical or efficient to utilize, aggregate demand typically increases rather than contracts.
Rakesh referenced three historical precedents. Server virtualization technology was anticipated to decrease server requirements but produced the opposite outcome. The DeepSeek introduction in 2025 generated concerns about GPU demand deceleration, yet AI infrastructure investment continued expanding. The transition from copper to optical networking infrastructure, despite offering 10x bandwidth improvements, was expected to lower costs but instead stimulated higher capital expenditure for AI servers.
NAND Market Fundamentals Show Continued Strength
Rakesh observed that NAND memory capacity in AI server builds has increased by 100% during the past twelve months. Spot market valuations have maintained an upward trajectory each quarter.
He contended that compression technologies like TurboQuant would “enable larger large-language models, faster inference and better tokenomics, spurring more spending” throughout the AI ecosystem.
Given both pricing momentum and robust fundamental demand characteristics, Mizuho expects Micron and SanDisk could surpass already-elevated earnings projections in upcoming periods.
SanDisk currently trades near $652, compared with Mizuho’s $710 valuation target. Micron hovers around $347, versus the firm’s $530 objective price.





