The artificial intelligence industry has entered a new phase: the focus is shifting from the chips that process data to the hardware needed to store it.
“Right now, we’re in the early stages of the memory cycle,” DA Davidson analyst Gil Luria told Yahoo Finance’s Opening Bid. “The advances we’re making in AI models make memory the next frontier. We need more memory in chips, devices, servers and data centers.”
“Companies like Micron, SK Hynix and Samsung making memory are becoming more and more important now,” he added.
The comments come as Micron Technology (MU) continues its vertical rise, with the stock up about 240% over the past year. Despite the gains, the valuation remains odd to some observers, trading at just 9.9 times forward earnings, a steep discount compared to 22 times for the S&P 500 (^GSPC) and 25 times for Nvidia (NVDA).
As the memory supercycle approaches, analysts have identified three key names for the next phase of the revolution.
Idaho-based Micron Technology has transformed from a cyclical laggard to a cornerstone of the artificial intelligence server stack. The main driver is high-bandwidth memory (HBM), a variant of specialized DRAM critical for AI training. Micron recently predicted that the total potential market for HBM will reach US$100 billion by 2028, with a compound annual growth rate of an astonishing 40%.
“It’s like getting a signed Mickey Mantle card at a garage sale,” Wedbush analyst Dan Ives said of Micron’s current price. Because HBM production is so complex, it is consuming capacity that would otherwise be used for traditional products such as smartphones and flash memory, allowing Micron to achieve richer margins and unprecedented pricing power.
Read more: How to protect your portfolio if you’re worried about the AI ​​bubble
While Micron is popular at home, many on Wall Street view South Korea’s SK Hynix (000660.KS) as the real center of the memory boom. SK Hynix is ​​Nvidia’s main supplier of HBM, maintaining a market share of approximately 60% by the end of 2025.
However, SK Hynix’s bull run is also its biggest risk, as its lead is so clear that it faces severe capacity constraints. If SK Hynix cannot meet the growing demand for HBM4 (next-generation AI memory), it risks losing out to competitors in 2026. However, UBS recently predicted that SK Hynix’s HBM4 market share could reach 70% by 2026 as the company plays a key role in Nvidia’s next-generation Rubin platform.
Sandisk (SNDK) is a surprising standout. Last year, its shares soared more than 800% after it was spun off from Western Digital (WDC). While most AI discussions focus on DRAM, or short-term memory, Sandisk is a leader in NAND flash memory, or long-term storage, which is becoming increasingly important for what Luria describes as “edge AI.” Such innovations include devices such as robots and self-driving cars, which rely on technology to process and store data locally.
