Mizuho recommended buying Micron and Sandisk shares on the downturn. He doesn't see Google as a threat
A bank analyst says investors should ignore the information noise around the memory market

Mizuho advises to buy Micron and Sandisk despite collapse due to Google / Photo: Rokas Tenys / Shutterstock.com
Mizuho Securities analyst Vijay Rakesh reiterated his recommendation to buy shares of memory chip makers Micron Technology and Sandisk after their sharp decline last week, Barron's writes. The fall came after a Google unit unveiled its TurboQuant algorithm, which the market fears could reduce demand for these chipmakers' products.
Details
Mizuho believes that investors should ignore the information noise related to concerns about the impact of the TurboQuant algorithm on memory demand. The bank's analyst confirmed the "bullish" position on the shares of both manufacturers. The target prices set by him suggest that the quotations of Micron may grow by 1.3% over a horizon of 12 months, and Sandisk securities will soar by almost 45%.
Sharply increased demand for memory amid expanding AI infrastructure and limited supply have already led to shortages and explosive price increases for these chips.Major cloud providers, including Amazon and Google, plan to spend about $650 billion on data centers and the purchase of processors, including chips for data storage, Bloomberg reported on March 26. As a result, Micron's capitalization jumped by 466% over the year, and Sandisk increased its market value by 2230%.
However, after the publication of Micron's quarterly report on March 19, the company's quotes fell sharply in price. Despite the fact that it presented strong results - for example, its revenue almost tripled - investors were spooked by rising capital expenditures. By the end of March, Micron's shares entered the "bear market" zone and were the cheapest in the S&P 500 index, as calculated by MarketWatch.
However, it was the release of TurboQuant that became the key industry pressure on chipmakers, Barron's believes. Since then, Sandisk has lost more than 10% of its value and Micron has lost more than 26%.
The TurboQuant compression algorithm, according to Google researchers, reduces the amount of "key memory" in artificial intelligence models by at least six times and can speed up their operation by eight times without loss of accuracy. The new development has led investors to fear that such technologies could alleviate shortages in the memory market, reduce the need for chips and consequently lead to lower chip prices, Bloomberg explains.
Why Mizuho doesn't see Google as a threat
Mizuho believes that the introduction of TurboQuant will ultimately play into the hands of memory manufacturers. According to the analyst, the increased efficiency will accelerate the spread of AI technologies and, conversely, increase demand for components, including memory chips. This effect is called the Jevons paradox, Barron's writes - after the English economist who described the use of coal by steam engines. Applied to the AI market - the expected data compression will enable larger language models, speed up inference (the process of processing queries and generating responses), and lower the cost of processing a single query, which will drive further cost increases, the bank analyst predicts.
The key driver of growth in the sector, according to its assessment, remains the dynamics of the flash memory market for data storage (NAND): the amount of memory in AI servers has doubled over the past year, and its prices continue to rise from quarter to quarter. Against this backdrop, Micron and Sandisk may exceed even the inflated expectations of analysts, according to Mizuho.
Wall Street consensus on Sandisk shares is mostly positive: the majority of analysts - 18 out of 24 - recommend buying its securities. Six take a neutral stance. There are no recommendations to sell. On Ma securities there is almost complete unanimity: 45 analysts out of 48 assigned "bullish" ratings to the company. The rest suggest to keep the shares in the portfolio.
This article was AI-translated and verified by a human editor
