SK Hynix Supplies More Than Half the World's HBM Memory. Its Stock Costs Under 4 Times Next Year's Earnings.

SK Hynix (NASDAQ: SKHY) sells more than half of the world's high-bandwidth memory (HBM) -- the specialized chips that feed data to the processors running artificial intelligence (AI) models. The company carried a 58% share of that market in the first quarter of 2026, according to Counterpoint Research, and demand for its products is so intense that its operating margin reached 76% last quarter.

Yet the stock costs less than 4 times forward earnings estimates, even after jumping 8% on Tuesday. As of this writing, shares sit near $154, about 21% below their 52-week high of $194.80.

A forward multiple that low is usually the market's way of saying earnings are about to peak. So, is the market right this time? I'm not so sure it is.

Image source: Getty Images.

The earnings the market doubts

The memory maker's second quarter, reported at the end of July, was extraordinary. Revenue came in at 79.3 trillion Korean won (about $55 billion), up 257% year over year, and operating profit soared 557% to 60.5 trillion won (about $42 billion).

The trajectory is steep even quarter to quarter. Compared with the first quarter, revenue grew 51% and operating profit climbed 61%. The company's operating margin widened to 76%, up from about 72% a quarter earlier and about 41% a year before that. And first-half revenue crossed 100 trillion won for the first time in the company's history.

Memory prices are still moving up, too: The company said both DRAM and NAND flash prices rose from the first quarter, with high-value products like HBM and DRAM for AI servers leading the increases.

Some of the growth is contracted, not just hoped for. SK Hynix said it has finalized long-term agreements with about 10 customers. And it began mass shipments of HBM4, its newest product generation, in the second quarter, with production ramping in the second half. Sample shipments of the following generation, HBM4E, went out during the first half.

The NAND side of the business is moving upmarket. The company's newest 321-layer chips already account for the largest share of its NAND production, and it's targeting about half of its domestic capacity by the end of the year.

The company also counts Nvidia as a key customer. And in late July, Nvidia and SK Hynix's parent, SK Group, announced a partnership worth more than $500 billion, with a long-term AI memory supply agreement for SK Hynix inside it.

What a multiple under 4 is pricing in

SK Hynix carries a market value of about $805 billion. Divide that by less than 4, and analysts' estimates imply somewhere around $210 billion of net income next year -- roughly double what the company earned over the past 12 months.

In other words, the market believes next year will be enormous and still won't pay up for it. That is how investors treat a cyclical business at what they suspect is the top of its cycle.

The suspicion isn't baseless. Memory has been a boom-and-bust industry for decades, and the busts can be ugly.

Supply is rising, too. SK Hynix itself plans capital expenditures in the high 40 trillion won range this year (more than $32 billion), and rival Samsung Electronics began shipping its first HBM4 chips earlier this year. Rising supply is exactly how past memory booms have ended.

To be fair, this cycle has features the old ones lacked. The long-term agreements commit customers to capacity years out, and HBM sells into an AI build-out that keeps getting bigger. The company also holds a net cash position of 69.4 trillion won (about $48 billion), so it could spend through a downturn without strain.

Still, a 76% operating margin invites competition and new capacity. Investors probably shouldn't count on profitability like that persisting deep into the decade, and I don't.

The stock doesn't need it to persist, though. At less than 4 times the earnings analysts expect next year, even a steep profit decline in 2028 and beyond could leave the stock looking reasonably priced in hindsight. But if those estimates start falling instead, the stock may fall hard with them. With memory, that risk is never small.

So, this is one I'd actually consider buying. Ultimately, I think the market's pessimism about the cycle is overdone, given how much of the demand is now under long-term contract. I'd buy in moderation, though, and go in expecting the rough stretch that every memory cycle eventually brings.

Daniel Sparks and his clients do not have positions in any of the stocks mentioned. The Motley Fool has positions in and recommends Nvidia. The Motley Fool has a disclosure policy.