Micron's $33 EPS Is Not the Interesting Part

Micron just reported $33.42 of non-GAAP EPS in one quarter.

Q1 FY2027 guidance is $38.15.

At some point the numbers stop looking like a semiconductor company and start looking like someone dragged a decimal into the wrong cell.

Cool.

The number that changed my model was not $33.42. It was management saying 2027 and 2028 should both be tighter than 2026, while more than 75% of 2027 output is already committed and a growing chunk of future revenue now sits under take-or-pay customer agreements. (Q4 results; Q4 call transcript)

The $33 EPS tells me the cycle is incredible today. The 2027–2028 language tells me the earnings window may stay open much longer than I had been willing to assume.

Okay, the numbers first

Q4 revenue was $54.23B, up 31% sequentially and 379% year over year. Non-GAAP gross margin hit 87%. EPS was $33.42 and adjusted free cash flow was $33.2B.

DRAM revenue reached $39.8B. NAND reached $14.1B. Data-center SSD revenue was nearly $10B, more than ten times the year-ago quarter.

Then Micron guided Q1 to:

Q1 FY2027 guide	Midpoint
Revenue	$61.5B
Gross margin	~86.25%
Non-GAAP EPS	$38.15

Management says Q1 should be the gross-margin floor for FY2027 and expects sequential revenue growth every quarter of the year. (Q4 results; Q4 call transcript)

Those numbers practically beg for a very silly valuation model.

Not doing that.

What I actually changed

Rather than bury the underwriting changes in the rest of the article, this is the delta from the report and call:

Assumption	Before Q4	After Q4
Supply tightness	2027 looked very tight; 2028 was less certain	2027 and 2028 both modeled tighter than 2026
Contract visibility	~$100B of minimum-price revenue across 14 of 16 signed SCAs	26 SCAs, >35% of estimated revenue through 2030, ~$150B RPO
FY2029 revenue	~$210B / $285B / $335B	~$225B / $300B / $350B
FY2029 gross margin	~55% / 68% / 75%	Unchanged
FY2029 normalized EPS	~$80 / $150 / $205	~$90 / $160 / $215
Terminal P/E	9x / 14x / 16x	Unchanged

So the upgrade is mostly duration + a better earnings floor, with some buyback benefit in the base and bull cases. None of that requires pretending 87% gross margin deserves a permanent multiple.

The shortage moved out another year

This was the part I kept coming back to after the call.

Before earnings, the 2027 shortage already looked real. HBM was crowding conventional DRAM capacity, enterprise SSD was getting tight and the fab timelines did not magically speed up because prices were high.

Now Micron is explicitly saying both 2027 and 2028 should be tighter than 2026.

That is a meaningful change.

Idaho 1 only starts wafer output in mid-2027. Idaho 2 and the Japan expansion start in late 2028. The new Singapore NAND fab is also a second-half 2028 event. And first wafer output is not full production; capacity ramps over following quarters.

So the supply response is coming.

The important word is response, not tomorrow.

Management also expects HBM demand to keep outgrowing conventional DRAM through 2028. Even with some customers optimizing memory content, server unit growth is still expected in the high teens in 2027.

Management cannot forecast the exact end of a memory cycle three years in advance.

But the current evidence does make an early end look less likely than it did yesterday.

The contracts are turning into something I can actually underwrite

Micron now has 26 strategic customer agreements.

Management estimates those SCAs cover more than 35% of revenue through 2030. Three-quarters of that expected SCA revenue has a defined pricing framework, usually with floor and ceiling prices.

The company disclosed about $150B of remaining performance obligations based on committed volumes and minimum pricing. Customer financial commitments are now $32B, mostly cash deposits. (Q4 call transcript)

That is not a normal spot-memory-cycle fact.

Management still thinks SCA coverage could eventually reach around 50% of revenue through 2030.

Signed contracts do not turn Micron into a software company. They do make the old mental model — where basically all future economics are whatever the spot market feels like giving you — less useful.

There is a real floor being built under part of the business.

How good that floor actually is will matter a lot once supply catches up.

This is where I can still get wrecked

The bull case is easy to write right now.

Honestly, too easy.

The risk is that every supplier sees the same pricing, the same customer panic, the same AI demand and eventually does what memory suppliers do: builds.

Micron is already accelerating capex.

Q1 FY2027 capex should be around $11.5B. First-half capex should be about $25B, with second-half spending higher. Construction spending is growing faster than equipment because Micron is trying to create clean-room capacity now and decide exactly how much equipment to install as demand develops.

That sequencing makes sense.

The uncomfortable part is judging it while gross margin sits at 87%, exactly when capital discipline is easiest to praise and hardest to test.

That is why FY2029 stays the valuation checkpoint. By then, the late-2028 capacity wave has at least started showing up and the model cannot hide behind “everything is sold out.”

Q1 EPS went to $38. My FY2029 base went from $150 to $160.

The near-term earnings picture got way better.

The normalized model moved too. Just nowhere near as much.

Here is the change:

FY2029	Pre-earnings	Post-Q4
Bear EPS	~$80	~$90
Base EPS	~$150	~$160
Bull EPS	~$205	~$215
Bear value	~$720	~$810
Base value	~$2,100	~$2,240
Bull value	~$3,280	~$3,440

The terminal multiples stay 9x / 14x / 16x.

That matters.

Annualizing $38 of guided quarterly EPS and slapping a multiple on it would produce a target that makes me feel like a genius.

It would also pretend today’s pricing environment is normal.

My post-Q4 assumptions are roughly:

Bear: ~$225B FY2029 revenue, ~55% gross margin, about $90 of EPS.
Base: ~$300B revenue, ~68% gross margin, about $160 of EPS.
Bull: ~$350B revenue, ~75% gross margin, about $215 of EPS.

The base and bull also assume real share-count reduction from buybacks.

From the $1,065.08 Sep. 30 close, those values imply roughly -8.7% / +28.1% / +47.8% annualized over three years.

That is a better model than yesterday.

It is still very deliberately not a Q1-earnings multiple.

The cash pile finally has a job

Micron finished the quarter with $68.3B of net cash.

Q4 adjusted free cash flow was $33.2B. Management said Q1 free cash flow should be significantly higher than that, even with $11.5B of capex.

Then came the part I had been waiting for: Micron intends to increase capital returns starting December 9, primarily through share repurchases, and says the long-run goal is to return 100% of excess cash to shareholders. (Q4 call transcript)

That moves my Capital Allocation score from 8.25 to 8.75 / 10.

The upgrade comes from the whole setup: fund the roadmap, keep the fortress balance sheet, and then use a ridiculous amount of excess free cash flow to shrink the share count.

Now management has to do it at sensible prices.

So am I buying more?

No.

At least not because of this quarter.

MU is already a large position and the 10% hard cap stays a hard cap. The fact that the thesis strengthened does not mean position-sizing discipline gets deleted.

That is the whole reason I separate conviction from sizing.

If I had no position today, this quarter would make Micron harder to ignore. With the position I already have, the relevant question is whether MU is the best use of another dollar compared with the rest of the portfolio.

The position is already large enough to pay me if the thesis is right.

The next five checks

The next few quarters are actually pretty clean.

First: does Q1 really prove to be the FY2027 gross-margin floor?

SCA coverage and RPO also need to translate into economics, not just giant headline commitments. The HBM4/HBM4E ramp should broaden across customers, while data-center SSD keeps proving this is a memory-hierarchy story rather than one product carrying the whole thesis.

Then there is capex. As construction turns into actual equipment and actual bits, the split still needs to look disciplined.

If those keep working, the higher FY2029 floor gets easier to defend.

If the capacity build accelerates faster than the contracts and demand behind it, I will care a lot more about that than whether the next quarter beats consensus by another couple billion dollars.

So yeah, I came out of the quarter more bullish.

My position is already big enough. I do not need to turn every stronger thesis into another buy order.