Micron's stock falls on fears about Nvidia's new chips, but investors may be overreacting

Dow Jones01:23

MW Micron's stock falls on fears about Nvidia's new chips, but investors may be overreacting

By Britney Nguyen

Some investors have been worried about Micron's role serving Nvidia's forthcoming chip lineup, but one analyst says he still likes the company's positioning

Micron's stock was down more than 2% on Monday morning.

Micron Technology's stock has been one of the market's hottest this year, but it's cooling down on Monday as investors assess the competitive landscape.

One issue potentially weighing on Micron's stock $(MU)$ on Monday is that rival Samsung Electronics (KR:005930) will start mass production of its sixth-generation high-bandwidth memory chips, HBM4, later this month for use in Nvidia's (NVDA) upcoming Vera Rubin graphics processing units, South Korea's Yonhap News Agency reported on Sunday.

That news followed a report from research firm SemiAnalysis on Friday that said Micron's HBM4 is not expected to be part of the supply in Rubin's first 12 months, as the chip maker has been unable to meet Nvidia's pin-speed requirements.

Mizuho trade-desk analyst Jordan Klein said in a Monday note that he thinks Micron's stock drop is related to a "moronic press" indicating that Micron's HBM4 offerings aren't fast enough for Nvidia and thus will get shut out of the Rubin GPUs.

Micron's stock was down more than 2% on Monday morning.

See more: Tech stocks have been shaky, but these 20 companies could still see rocketing sales growth

Most of the HBM4 supply for Nvidia's Rubin GPUs will likely come from South Korea's SK Hynix (KR:000660), SemiAnalysis said. Korea Economic Daily also reported on Monday that SK Hynix will likely make up a mid-50% share of HBM4 supply for Nvidia, while Samsung takes a mid-20% share. But that report signaled there still may be a place for Micron, at 20% of HBM4 supply for Rubin.

Klein noted that Micron executives pushed back on reports of performance issues with its HBM4 during the company's earnings calls in September and December. The company could confirm its HBM4 status at the Morgan Stanley technology conference in March, Klein said.

Still, Klein said he expects yields, or the number of usable chips from a single wafer, to drive share of the market for HBM4. So far, Micron's and SK Hynix's yields are better when compared with Samsung's more advanced approach, he added.

Meanwhile, the latest round of tech earnings reports saw some cloud providers lift their capital-expenditure forecasts, and RBC Capital Markets analyst Srini Pajjuri said rising prices in the memory market are having a major impact there.

According to Pajjuri, the price surge itself could be behind a third of the overall increase in AI spending this year at the top 10 hyperscaler and neocloud companies that RBC tracks, including Alphabet $(GOOGL)$ $(GOOG)$, Amazon (AMZN) and CoreWeave (CRWV).

In its earnings report last week, Alphabet raised its capex forecast for this year to between $175 billion and $185 billion, almost double what it spent in 2025. The following afternoon, Amazon announced that it plans to spend $200 billion in 2026.

Don't miss: 'It's existential': How Big Tech found itself in a $650 billion spending spiral

Prices for memory and storage components have jumped in recent quarters as AI-driven demand leads to supply shortages. At the same time, companies such as Micron, Western Digital $(WDC)$ and Seagate Technology $(STX)$ have been reluctant to add capacity out of fear that an eventual slowdown in demand would lead to oversupply - a common occurrence in the typically cyclical industry.

Pajjuri is modeling for memory chips to make up about 45% of capex dollar growth this year and for three quarters of that to be from higher prices. Dynamic random-access memory prices are expected to more than double this year, while NAND prices could rise more than 85%, Pajjuri said, citing TrendForce data. Both types of memory have been in high demand to support larger, more powerful AI models with longer context windows, or the amount of data being processed at one time.

He estimates that the market for data-center memory will be $237 billion in 2026 and make up about 30% of the aggregated capex for the top 10 cloud providers.

Pajjuri estimates that in 2025, spending on data-center memory and storage components such as DRAM and enterprise solid-state drives made up about 20% of the aggregated capex for the top 10 hyperscalers.

He is modeling for memory prices to moderate "but not severely correct" in 2027.

However, changes in memory pricing "are difficult to predict and remain possibly the biggest wildcard for capex growth (or potential decline) heading into 2027," Pajjuri said.

Read on: This Micron stock chart is sending an ominous signal, if history is any guide

-Britney Nguyen

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.

 

(END) Dow Jones Newswires

February 09, 2026 12:23 ET (17:23 GMT)

Copyright (c) 2026 Dow Jones & Company, Inc.

At the request of the copyright holder, you need to log in to view this content

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Comments

We need your insight to fill this gap
Leave a comment