This AI Stock Is Nearly a 17-Bagger for Investors Who Bought in 2021 (Hint: Not Nvidia). 1 Wall Street Analyst Sees 78% Upside.

The artificial intelligence (AI) trade has created significant wealth for investors who bought in five years ago.

There are now 14 publicly traded companies with market caps over $1 trillion, many of which have become multibaggers many times over.

Remarkably, one company is now nearly a 17-bagger since 2021. No, it's not Nvidia, and one Wall Street analyst still sees 78% upside.

Image source: Getty Images.

Memory stocks have crushed it

As many investors now know, demand for compute memory has skyrocketed due to the rise of artificial intelligence. Specifically, there has been intense demand for dynamic random access memory (DRAM).

DRAM sits atop the graphics processing units (GPUs) and essentially feeds the chips data. The even more advanced memory is called high-bandwidth memory (HBM), which is simply multiple DRAM chips stacked together, allowing more data to be processed and reducing the distance to the GPUs.

As data centers have scaled clusters of GPUs due to increased consumption of large language models (LLMs) and LLMs have also grown larger, there is significantly greater demand for DRAM and HBM.

This means that if you're one of the big memory companies, you've done well. One of the leaders in the DRAM and HBM space is SK Hynix (KOSE: A000660), based in South Korea. The company was early to invest in HBM and quickly gobbled up market share.

According to the market research firm Counterpoint, at the end of 2025, SK Hynix held 32% of the global DRAM market share by revenue, second only to Samsung. However, in the global HBM market by revenue, the company is the undisputed leader with 57% market share. Samsung ranked second with 22% market share.

The upside case for SK Hynix

Now, as mentioned, the stock has been a tremendous winner if you bought it five years ago, and even if you bought it one year ago. SK Hynix is up roughly 978% over the past year and 1,694% over the past five (as of May 27).

The stock currently trades at 2,243,000 South Korean Won (SKW) per share, equivalent to roughly $1,493 per share. According to 38 Wall Street analysts who have issued research reports and price targets on the stock over the past three months, the stock is slightly overvalued, with the average price target of 2,076,603 (SKW).

However, some of these price targets are a month or two old.

On May 17, Nomura analyst CW Chung raised his price target on SK Hynix from 2,340,000 to 4,000,000 SKW, implying roughly 78% upside from current levels.

Memory companies are often viewed as cyclical because demand can come and go, while it can take the industry time to match supply.

However, analysts like Chung think the AI supercycle will lead to demand far outstripping the supply of memory. As a result, memory giants like SK Hynix, which trades around 6 times forward earnings, should trade more like Taiwan Semiconductor, which trades at around 20 times forward earnings, Chung wrote in his research note.

SK Hynix plans to list American depositary receipts (ADRs), which Chung believes will help close the discount, in part by expanding its shareholder base.

The main driver here is clearly demand for DRAM, and analysts at UBS believe DRAM supply will be constrained until at least midway through 2028, suggesting the memory run can last for at least a few more years.

While some of this is undoubtedly priced in, if there's a memory stock to buy, I believe SK Hynix is a good pick, given it trades at a discount compared to competitors such as Micron, which currently trades around 16 times forward earnings.

Micron has made some gains, to be sure, and has another little advantage by being based in the U.S. However, as mentioned, SK Hynix is the undisputed leader in HBM, which is specifically tailored for AI chips.

Bram Berkowitz has no position in any of the stocks mentioned. The Motley Fool has positions in and recommends Micron Technology, Nvidia, and Taiwan Semiconductor Manufacturing. The Motley Fool has a disclosure policy.