How to capitalize on the memory semiconductor race in the AI era

The artificial intelligence sector is undergoing a fundamental reshuffle. While recently the focus was on massive computing processors, attention is now centered on a less visible but equally critical component: memory chips. For investors looking to capitalize on this paradigm shift, three companies stand out as key players in this technological and economic transformation.

Gil Luria, an analyst at DA Davidson, explains that “we are still in the early stages of an unprecedented memory cycle. The sophistication of AI models has turned data storage into the decisive bottleneck. The system requires greater memory capacity in chips, infrastructure, servers, and global data centers.”

From Processing to Storage: The Market’s Strategic Shift

For years, the tech narrative revolved around raw processing power. Nvidia cemented its dominance with specialized computing processors. However, the architecture of AI systems has revealed a vulnerability: no matter how much computational power you have, if you don’t have efficient places to store data in transit.

High-bandwidth memory (HBM) represents the solution to this dilemma. It is a specialized version of DRAM that enables extraordinary data transfer speeds, essential for training next-generation AI models. Industry estimates suggest that the HBM market could reach $100 billion by 2028, growing at an annual rate of 40%.

This shift in focus has created investment opportunities in companies traditionally considered commodity producers. The market is revaluing assets that for years remained under the radar of institutional investors.

Three Companies Positioned to Capitalize on Semiconductor Growth

Micron: From Cyclical Player to Critical Infrastructure

Micron has experienced a spectacular transformation. Its shares have appreciated approximately 240% over the past twelve months, marking a change in market perception. Most notably, this appreciation occurs while the company’s relative valuation remains contained: trading at 9.9 times projected earnings, well below the S&P 500 (22 times) and much lower than Nvidia (25 times).

Headquartered in Idaho, the company has transitioned from a cyclical producer to a vital component of AI server infrastructure. Its specialization in HBM positions it directly in the sector’s fastest-growing segment.

SK Hynix: The Heart of the Global Boom

While Micron is the favorite among North American investors, SK Hynix represents the central hub of the worldwide memory revolution. The South Korean company is Nvidia’s main HBM supplier, controlling approximately 60% of the market by the end of 2025.

However, success brings operational pressures. SK Hynix faces capacity constraints that could threaten its position if it cannot meet demand for HBM4, the next generation of memory. UBS experts project its market share could expand to 70% in 2026, especially if it secures the specifications for Nvidia’s Rubin platform.

SanDisk: The Surprising Leader in Flash Storage

SanDisk has emerged as a surprise player in this scene. After spinning off from Western Digital, the company’s shares have surged over 800% in the past year. Its strength lies in NAND flash storage, the technology responsible for long-term data storage.

While most analysts focus on DRAM, SanDisk is increasingly occupying a space: “edge AI.” This term refers to innovations like autonomous vehicles and robotic systems that require on-site processing and storage, without relying on connections to central data centers.

Hidden Risks Behind Memory Speculation

However, investors should consider structural risks. Unlike Nvidia, which owns a proprietary software ecosystem that creates dependency, memory chips are fundamentally interchangeable. This commodity characteristic means pricing power is weak.

“Nvidia can increase orders to SK Hynix for a year and switch to Micron the next,” says Luria. “This flexible supply model offers less long-term stability for memory suppliers.” Once supply constraints normalize, profit margins could be significantly compressed.

For now, investors are prioritizing immediate supply shortages over structural concerns. “In critical bottleneck contexts, short-term investment decisions are not influenced by risks expected in the medium term,” adds Luria. This limited time horizon maintains upward momentum, though constant monitoring of demand-supply dynamics in the semiconductor sector is necessary.