Stock prices hit new highs, yet storage stocks still trade at a valuation trough

Original author: Jialiu

Today, Micron delivered a historic earnings report that has significantly boosted confidence across the semiconductor sector.

FY2026 Q3 revenue reached $414.6 billion, exceeding market expectations by nearly $60 billion. A memory company, long tagged as a "low-margin commodity," delivered gross margin guidance comparable to a software company. Its stock price surged 13% to 14% in after-hours trading, pushing its market capitalization to $1.16 trillion.

Micron's gains this year have already been substantial. Closing at $1,211.38 on June 22, it has more than tripled year-to-date and surged over 850% in the past 12 months, making it the third best-performing stock in the S&P 500 for 2026, after SanDisk and Western Digital – also memory companies. The entire sector is moving up by this magnitude. SK Hynix has risen over 800% in the past 52 weeks, and Samsung over 400%.

With such rises, many people's first reaction is, of course, "it's too expensive." But in reality, a high stock price doesn't necessarily mean expensive valuations. From many perspectives, memory remains a very "cheap" hot sector.

Stock Price Up 9x, Yet PE Remains Stagnant

One of the most common indicators to judge whether a company's stock is expensive or cheap is PE, the price-to-earnings ratio.

Simply put, PE measures how much the market is willing to pay for every $1 of profit a company earns. A PE of 10 means investors are willing to pay $10 for $1 of annual profit. A high PE usually indicates strong future growth expectations; a low PE might mean the stock is cheap, or that the market believes current profits are just a cyclical peak that will soon decline.

This is the most counterintuitive aspect of memory stocks right now: stock prices have soared, but PEs remain low.

According to FactSet data cited by MarketWatch in mid-June, Micron's forward 12-month PE is around 9 times, while SK Hynix and Samsung are around 6.5 times. Barron's estimates Micron's forward PE at about 9.74 times, compared to the NASDAQ Composite's ~25.5 times and the S&P 500's ~20.3 times. GuruFocus data from June 21 shows Micron's forward PE at 9.90 times, SK Hynix at 5.92 times (late May data), and Samsung around 5.45 times.

This means most data sources suggest the forward PEs of the memory big three are all in the single digits to low teens.

Within the entire AI industry chain, these figures are almost the lowest tier.

Nvidia's forward PE is around 23 times, Broadcom around 30 times, AMD around 25 times, TSMC around 20 times, and the overall semiconductor industry median is around 36 times. This means the valuation level of the memory big three is roughly one-third of Nvidia's and one-quarter of the semiconductor industry median.

Ironically, however, the money in the AI industry is increasingly flowing to the memory sector.

AI servers aren't just about GPUs. Every high-end AI accelerator card needs HBM, every inference server needs large-capacity DRAM, and KV cache, model weights, local caching, and data throughput all depend on SSDs. Without HBM, there are no GPU training clusters; without server DRAM, there are no inference clusters; without high-capacity NAND, the storage and caching costs for AI applications can't be reduced.

Memory is no longer just a standard component in the AI industry chain; it's a physical bottleneck that all AI capital expenditure must overcome. A number from Micron's recent earnings illustrates this: single-quarter data center revenue was $250 billion, with enterprise SSD revenue at $50 billion, accounting for 20% of data center revenue.

It's evident that this bottleneck is now even beginning to affect consumer electronics.

AI data centers have driven up the capacity and prices of HBM, DRAM, and NAND, eventually forcing even highly price-powerful terminal companies like Apple to face cost pressures and pass some of the increases onto consumers. In the past, discussions about who profits from AI immediately focused on Nvidia; but now it's increasingly clear that a large portion of the AI bill is flowing to memory manufacturers.

Memory stocks have risen a lot, but profits have risen even faster.

Micron just reported Q3 EPS of $25.11, compared to $1.91 in the same period last year – a more than tenfold increase in one year. SK Hynix's Q1 2026 operating profit was 37.61 trillion KRW, up 405% year-over-year. Samsung's semiconductor division saw Q1 operating profit surge over eight times year-over-year. Stock prices have multiplied, but profits have multiplied even more, preventing PEs from being stretched.

AI money is flowing in real, tangible terms into the income statements of memory manufacturers.

Sector Catalysts Unfolding in Unison; Micron Just the First Shot

Micron's earnings report is the starting gun for this round of memory earnings season.

Next, the memory sector enters a month with a high density of information: TSMC on July 16, Samsung on July 23, SK Hynix and Western Digital on July 29.

The force of Micron's report has already set the tone for the others. Its most crucial information isn't the single-quarter beat, but the guidance for Q4: $500 billion in revenue and an 86% gross margin.

This guidance essentially tells the market that price increases haven't peaked; they are accelerating. The subsequent four companies will, essentially, test or confirm the same trend revealed by Micron's guidance, albeit in different markets with different product mixes.

First, look at TSMC, reporting on July 16.

TSMC doesn't make memory, but it's the foundation of the entire AI chip supply chain. Nvidia's GPUs, Broadcom's custom accelerators, and AMD's data center chips all come from its fabs. TSMC addresses a more fundamental question than memory: has the capacity bottleneck for AI chips been resolved? Q1 revenue was $35.9 billion, up 40.6% year-over-year, with gross margin at 66.2%. Advanced nodes accounted for 74% of wafer revenue. Q2 guidance is $39.0 to $40.2 billion.

There's a multiplier relationship between TSMC and memory. Every additional advanced-node wafer TSMC sells translates into more AI accelerators downstream, and each additional accelerator requires more HBM stacks. The Nvidia Vera Rubin platform's single GPU will use several times the HBM capacity of the previous generation. The more TSMC ships, the tighter the memory capacity becomes.

July 23 is Samsung's earnings report.

Fifteen brokerages expect Samsung's Q2 operating profit to be around 88.3 trillion KRW, with an operating margin flat or higher than Q1's 66%. For a conglomerate that also makes mobile phones, panels, and home appliances, its margin has been pulled to this level by the memory division alone.

But the most important aspect of Samsung's report isn't the profit figure; it's HBM4. Samsung holds only about 17% share in the HBM market, far behind SK Hynix's 62% and Micron's 21%. The HBM4 generation shift is Samsung's only window to close the gap. In its Q1 conference call, it made several specific statements: HBM sales in 2026 will grow more than three times year-over-year, and starting from Q3, HBM4 will account for over 50% of its HBM sales. Micron just disclosed that its HBM4 36GB 12-Hi has begun volume shipments and delivery. The positioning battle among the three on HBM4 will be the most crucial contest worth watching in the second half of the year.

On July 29, SK Hynix and Western Digital report on the same day.

SK Hynix's Q1 was textbook: quarterly revenue of 52.6 trillion KRW, up 198% year-over-year; operating margin of 72%; net profit margin of 77%. A hardware manufacturer achieving a 77% net profit margin – Apple is around 25%, Nvidia around 58%. Some brokerages predict Q2 operating margin could approach 80%. Micron has already achieved an operating margin of 81.2%, surpassing TSMC. As the leader in HBM market share, SK Hynix will likely deliver Q2 results comparable to Micron's. The combined Q2 operating profit of Samsung and SK Hynix is expected to exceed 150 trillion KRW. Together with Micron, the single-quarter combined profit of the big three will set a new record.

Western Digital releases its Q4 on the same day. It has no DRAM or HBM business, focusing purely on NAND and SSDs. It provides another dimension of AI storage demand: inference KV cache requires large-capacity SSDs. Q3 Cloud revenue grew 48% year-over-year, and gross margin hit a record 50.5%. Notably, Western Digital and its spun-off SanDisk are the two best-performing stocks in the S&P 500 for 2026, ahead of Micron. NAND line growth isn't as explosive as DRAM, but the direction is entirely consistent.

AI Transforms Memory from Commodity to Luxury Good

Stock prices at all-time highs, yet PEs are low. Earnings reports are increasingly spectacular.

Reading this, some might still doubt whether this is sustainable or just another cyclical frenzy destined to collapse.

Let's consider the analysis from Jukan, a semiconductor analyst at Citrini Research.

Back in Q1 2024, when SK Hynix and Micron were still mired in post-pandemic DRAM inventory gluts and low stock prices, Citrini's team called for these two to outperform. Subsequently, these stocks rose several times, nearing 10x. They have been largely correct throughout this memory cycle. A detail from early June 2026 illustrates his market influence: he retweeted a SemiAnalysis report regarding changes to Nvidia's Rubin server memory configuration, which immediately put visible pressure on Micron and SK Hynix during the trading day.

Jukan's core bullish thesis for memory isn't a short-term call like "prices will rise." Instead, he argues: AI has transformed memory from a commodity into a luxury good.

First, HBM has broken a sixty-year trend curve. From 1957 to 2020, the cost per Gb of DRAM decreased by roughly an order of magnitude every five years. Prices were always falling. This was the underlying law of the memory industry, and the entire sector's competitive model and valuation framework were built upon this line. Jukan points out that the AI-driven demand for HBM has decisively broken this law. Manufacturers shifted capacity to HBM, which is more complex to produce and consumes more silicon area, squeezing traditional DRAM supply.

Currently, no manufacturer plans to convert HBM production lines back to traditional DRAM. The reason is simple: HBM profit margins are far higher than standard DRAM. Rational manufacturers won't swap high-margin lines for low-margin products. This transforms the supply tightness from a cyclical phenomenon into a structural one that won't reverse as long as AI demand persists.

Therefore, the sustained price increase for HBM memory will be long-term.

HBM's full-year volume and pricing are largely agreed upon at the beginning of the year, providing manufacturers with strong earnings visibility. TrendForce data confirms this: in Q1 2026, traditional DRAM contract prices rose 90% to 95% quarter-over-quarter, the largest single-quarter increase on record, with Q2 expected to rise further. In a normal DRAM cycle, price increases typically peak within 4 to 6 quarters. This current cycle has been rising for nearly 8 quarters without stopping. JP Morgan even predicts DRAM prices could rise for four consecutive years, something never seen in industry history.

So, it's almost safe to say that memory has transitioned from a commodity to a luxury good.

The biggest difference between luxury goods and commodities lies in pricing. Commodity prices are determined by marginal cost; anyone can expand production, and competition ultimately erodes profits, hence a low valuation. Luxury good prices are determined by scarcity and pricing power; supply is controlled, allowing profits to remain high for extended periods, hence a premium valuation. The old rule that "low PE equals peak" presupposes that earnings will revert to that long-term declining trend line. But if the trend line itself has turned, then where earnings revert to becomes an open question.

Returning to the initial paradox: stock prices are at historical highs, valuations are at historical lows. This anomaly exists because the market is still using the old framework for commodities to price an industry that has become a luxury. Micron just dealt a heavy blow to this old framework with its 84.9% gross margin earnings report and 86% gross margin guidance. If things don't revert, the current P/E of 5 to 10 times is wrong.

Therefore, we believe that despite new highs, memory stocks are still not expensive.

After HBM, is NAND the Real Main Course?

Midway through every major market cycle, the same question arises: the leaders have surged, who is next to take the baton?

HBM and DRAM have been the absolute protagonists of this memory cycle, with the meteoric rises of the big three largely attributed to them. NAND has consistently been treated as a supporting actor.

But if you look closely at the supply-demand structure, you find a counterintuitive truth: NAND, often treated like a side dish (perilla leaf), might actually be the main course itself, and its scarcity is, in some ways, more severe than HBM's.

First, why is HBM so hot? HBM is standard equipment for AI accelerator cards – high unit price, high margins, high entry barriers. SK Hynix leveraged it to achieve 62% market share and 77% net profit margin. These are facts. However, HBM has one characteristic: while supply is tight, the expansion path is clear. The big three are heavily investing capital to expand HBM capacity. Samsung and Micron are chasing SK Hynix, climbing generations from HBM4 to HBM4E. Supply is increasing at a visible pace, just temporarily lagging behind demand.

In contrast, NAND manufacturers haven't meaningfully expanded capacity for years.

The reason is that the NAND price crash of 2022-2023 left all players deeply scarred. Capital expenditure on NAND by Kioxia, Western Digital, Samsung, and SK Hynix was slashed to very low levels, and new production lines have been repeatedly delayed, with the earliest now expected in 2027.

The big three prioritize wafer capacity and capital expenditure for HBM and high-end DRAM, leaving even fewer resources for NAND. Micron even went so far as to shut down its consumer-grade Crucial business entirely, reallocating all capacity to enterprise and GPU-grade storage.

On the supply side, NAND output is missing. On the demand side, needs are enormous.

Large model inference requires massive KV cache and data throughput, directly fueling explosive demand for enterprise SSDs (eSSD). In Q1 2026, global eSSD revenue grew 86% quarter-over-quarter. Another factor is HDD shortages. Mechanical hard drive supply is also tight, forcing data centers to substitute HDDs with high-capacity SSDs, shifting some demand originally belonging to HDDs onto NAND.

The CEO of Phison Electronics stated, "Every NAND manufacturer has told us that 2026 is sold out." Kioxia also confirmed that its entire 2026 NAND capacity is fully booked. The price of a 1Tb TLC NAND chip rose from around $4.8 in July 2025 to around $10.7 by the end of 2025, more than doubling in a few months.

HBM is tight, but supply is increasing deterministically. NAND is tight, but there is almost no incremental supply on the horizon. HBM's tightness has a cure, just slow-acting. NAND's tightness currently has no cure because no one is making the medicine. From this perspective, NAND's supply-demand gap is more rigid than HBM's, and the sustainability of its price increases might be stronger.

This explains why the two best-performing stocks in the S&P 500 for 2026 are not the HBM leader SK Hynix, nor Micron, but the pure-play NAND and SSD company Western Digital, and its spun-off entity SanDisk. The market has already voted with its feet, quietly moving NAND from the supporting cast to the lead role, though most haven't noticed yet.

Of course, NAND has its risks. It doesn't have the rigid demand binding of AI accelerator cards like