Micron (MU): The 7× P/E Wasn’t a Typo

Micron just delivered 87% gross margins, another beat-and-raise, and guidance well above Wall Street expectations. The stock barely reacted. The market’s problem with Micron is no longer the earnings. It is how long those earnings can last.

Event Re-underwrite

Micron entered Wednesday’s earnings report with one of the stranger setups in the market.

The stock had risen roughly 550% over the prior year. Memory prices had exploded. AI data centers were consuming high-bandwidth memory and conventional DRAM faster than the industry could comfortably supply them. Yet immediately before earnings, Micron was still trading at only about 6.5× estimated forward earnings.

Usually, after a stock rises fivefold, the argument becomes that investors are paying too much for the future.

With Micron, the argument is almost the opposite.

The earnings have risen so quickly that the stock still looks cheap.

Wednesday’s results made that paradox harder to dismiss.

Another Beat. Then Another Raise.

Micron reported fiscal fourth-quarter revenue of $54.23 billion, ahead of the roughly $51.3 billion Wall Street expected. Adjusted earnings reached $33.42 per share, versus consensus near $31.72.

Non-GAAP gross margin reached a record 87%, up from 84.9% only one quarter earlier.

Those numbers are extraordinary by historical Micron standards, but another strong quarter was largely expected.

The guidance was more interesting.

For fiscal Q1 2027, Micron expects approximately $61.5 billion of revenue and $38.15 of adjusted EPS at the midpoint. Wall Street had been around $57.4 billion and $35.47 respectively.

Micron didn't merely clear the quarter.

It raised the earnings curve again.

And yet the initial after-hours reaction was remarkably subdued.

That may tell us more about the Micron debate than another earnings beat ever could.

This Is Still A Pricing Story

One of the most important things to understand about Micron’s earnings explosion is what isn’t happening.

Micron isn't suddenly shipping five times as many memory chips.

Much of the financial acceleration has come from scarcity-driven pricing.

In the previous quarter, DRAM average selling prices increased in the low-60% range sequentially while bit shipments increased only in the low-single digits. NAND pricing rose even faster. Those economics pushed Micron’s gross margin to 84.9%, more than double the level from a year earlier.

That was the clue.

AI demand is colliding with a memory industry that cannot manufacture incremental supply overnight. New fabs take years to build, equipment takes time to install and qualify, and each generation of high-bandwidth memory consumes substantial manufacturing capacity.

The result is unusual pricing power for an industry investors have historically treated as a commodity business.

That distinction matters because the investment case now depends increasingly on duration.

Another quarter of extraordinary pricing is good.

Several years of structurally tighter memory supply would be something much bigger.

The 7× P/E Isn't The Thesis

Before earnings, Micron's valuation had fallen to roughly 6.5× estimated future earnings, despite the enormous stock-price appreciation.

At first glance, that looks absurdly cheap.

But the market knows how to divide price by earnings.

The low multiple is not an arithmetic mistake. It is the market expressing skepticism about the denominator.

Memory has traditionally been one of the semiconductor industry's great boom-and-bust businesses. Demand strengthens, prices rise, manufacturers invest, new supply eventually arrives, prices fall, profits collapse, investment stops—and the cycle begins again.

That creates a dangerous feature in cyclical investing:

A stock can look cheapest precisely when earnings are closest to their peak.

That is why simply calling Micron a “7× P/E stock” misses the question.

The real question is:

What does normalized Micron earnings power look like after the current shortage ends?

Consider the difference.

If Micron can sustainably earn $200 per share, a $1,075 stock is only about 5.4× earnings.

At $150 of normalized EPS, it is roughly 7.2×.

At $100, the multiple becomes 10.8×.

At $75, it becomes more than 14×.

Same stock price. Completely different investment.

The durability of the E matters far more than the apparent cheapness of the P/E.

The Contracts May Be Changing The Cycle

This is where the Micron story becomes more interesting than a normal memory shortage.

Micron now has 26 Strategic Customer Agreements, up from 16 at its prior earnings report. Management says those agreements represent more than 35% of projected revenue through 2030.

These are not the loose annual purchasing arrangements historically common in memory.

Micron has described its strategic agreements as multi-year, binding take-or-pay commitments designed to give customers guaranteed supply while giving Micron greater demand and pricing visibility.

That changes the economics in both directions.

During a downturn, price floors and committed purchases can protect Micron from some of the devastation associated with previous memory cycles.

During an extreme upcycle, price ceilings can prevent Micron from capturing every last dollar available in the spot market.

Micron is effectively exchanging a portion of unlimited upside for something memory investors historically almost never received:

visibility.

If that works, the implication reaches beyond the next few quarters.

A company with materially more predictable multi-year earnings deserves to be valued differently from a company whose profits disappear every time the memory cycle turns.

That possibility is one reason we are moving Micron into Hayabusa Living Research.

AI Is Changing More Than HBM

HBM gets most of the attention because it sits directly alongside AI accelerators.

That makes sense.

But focusing exclusively on HBM may understate the broader memory demand created by AI infrastructure.

AI servers need enormous pools of conventional DRAM. Inference workloads increasingly require high-capacity memory. Fast storage becomes more valuable as context windows expand, data sets grow and applications need to move information rapidly between different layers of the computing stack.

The architecture increasingly resembles a hierarchy:

HBM near the accelerator → server DRAM → fast NAND storage → larger storage pools.

More compute creates more demand at several layers.

Micron participates across much of that hierarchy.

That makes the deeper thesis less dependent on one product category and more dependent on something broader:

AI systems are becoming increasingly memory-intensive.

If compute continues scaling while memory per system rises at the same time, the demand curve facing Micron may look very different from the PC- and smartphone-dominated cycles investors remember.

The Supply Response Is Coming—Slowly

None of this means scarcity lasts forever.

Micron and its competitors are spending heavily to increase capacity. The economics practically demand it.

But semiconductor manufacturing has a frustrating characteristic for anyone waiting for supply:

Money can be committed quickly.

Fabs cannot.

Micron has repeatedly emphasized that greenfield fabrication facilities are extremely complex projects requiring years of construction, equipment installation, qualification and production ramping. In prior commentary, management said it still lacked clear visibility into when memory supply would fully catch the demand curve.

This creates a lag.

The industry can simultaneously be spending enormous amounts of capital and remain supply constrained.

That's probably where Micron sits today.

Eventually those investments become wafers.

The question is whether AI demand keeps moving faster than that new supply.

Old Micron Versus New Micron

There are now two reasonable ways to underwrite the company.

Old Micron

The first says nothing fundamental has changed.

Today's 87% gross margin is what peak-cycle economics look like. Extraordinary scarcity creates extraordinary prices. Extraordinary prices encourage extraordinary investment. Supply eventually catches demand, pricing weakens and earnings fall sharply.

Under this interpretation, a 6× or 7× multiple isn't strange.

It's exactly what a cyclical company should trade for near peak earnings.

New Micron

The second interpretation says AI has altered both sides of the equation.

Demand is increasing faster because AI systems consume enormous amounts of memory. HBM absorbs manufacturing capacity. Advanced nodes are increasingly difficult to produce. New fabs take years to bring online. And customers are now willing to commit billions of dollars and sign multi-year agreements simply to guarantee access to supply.

Under that scenario, earnings eventually normalize—but they normalize at a level dramatically above Micron's historical cycle.

Wednesday's earnings added evidence to the second interpretation.

They did not prove it.

That distinction matters.

What Would Change Our Mind?

Living Research is most useful when the thesis includes the evidence that can break it.

For Micron, pricing comes first.

If DRAM or NAND pricing begins falling meaningfully before new capacity arrives, that would suggest current scarcity is less durable than management believes.

Gross margin is second. Micron guided fiscal Q1 non-GAAP gross margin to approximately 86.25%, only modestly below the record 87% just reported. Sustaining margins anywhere near these levels would be powerful evidence that the current earnings base is not disappearing immediately.

We will also watch the growth of Strategic Customer Agreements. If customers continue locking in multi-year supply despite elevated prices, that is evidence memory has become strategically important enough that availability matters more than bargaining for the lowest possible spot price.

Then comes capacity.

Micron can announce factories today, but what matters is when those investments become usable bits of DRAM and NAND. The critical comparison is not announced capex versus current demand. It is incremental supply actually reaching production versus incremental AI memory demand.

And finally, we will watch AI infrastructure spending itself.

A significant slowdown in hyperscaler capex would attack the demand side of the Micron thesis before the supply response even arrives.

The Hayabusa Read

The most interesting thing about Micron's earnings was not that the company beat expectations again.

At this point, investors already know Micron is making extraordinary money.

The interesting part is that Micron reported $54.2 billion of quarterly revenue, 87% gross margins and $33.42 of adjusted EPS, then guided the next quarter materially above Wall Street estimates—and the stock initially barely reacted.

The battleground has moved.

The market isn't arguing about whether today's memory shortage is profitable.

It is arguing about what happens after it.

If AI has permanently increased memory intensity, if capacity remains slow to respond, and if multi-year customer agreements create a floor under future economics, then Micron's historical valuation framework may no longer fit the company.

If those forces merely extend another semiconductor cycle, the low multiple may be giving exactly the warning it appears to contradict.

That's why Micron belongs in Living Research.

We don't need to predict the answer today.

We need to keep watching the evidence.

Living Research: What We're Watching

DRAM and NAND pricing — Is pricing merely moderating, or beginning to reverse?

Gross margin — Can Micron sustain margins far above previous cycle peaks?

HBM — Pricing, allocation, product transitions and customer commitments.

Strategic Customer Agreements — Number of agreements, revenue coverage, floors, ceilings and customer deposits.

Capacity — When announced fabs and equipment spending actually become productive memory supply.

AI infrastructure demand — Hyperscaler capex, accelerator deployments and memory content per system.

Normalized EPS — The most important valuation question: what can Micron earn after shortage economics fade?

Next Catalyst: Fiscal Q1 2027 earnings.

Disclosure: Hayabusa Research publishes independent investment research for informational and educational purposes. Nothing in this article constitutes investment advice, a recommendation, or an offer to buy or sell securities. Semiconductor and memory markets are cyclical, and forward earnings estimates can change materially as pricing, demand and capacity conditions evolve.

Curiosity Before Capital. Return with evidence.