Western Digital’s Valuation Highlights a Dangerous Gap Between Price and Capacity

The consensus on Western Digital (WDC) right now is as clean and unanimous as these things ever get: AI is building out, data centers need storage, WDC is the storage company, therefore the stock goes up. I’ve watched this kind of argument travel from pitch deck to portfolio to cocktail party before, and the journey rarely ends where the early believers expect. The stock closed at $483.15 per Yahoo Finance, sitting just a whisker below its 52-week high of $483.87 per stockanalysis.com — a range that stretches down to $43.60 at the trough, meaning anyone who bought the bottom has roughly ten-bagged. That kind of return doesn’t just attract capital. It attracts certainty, which is a different and more dangerous thing.

In a previous piece on WDC from early April, I flagged the AI demand narrative and the “bullwhip effect” as a silent variable worth watching. The stock has continued climbing since then, which either means I was early or wrong — and at these prices, the distinction matters less than whether the underlying argument has changed. It hasn’t. If anything, the setup has grown more stretched, and the concern sharpens rather than softens.

Here’s what the numbers actually look like beneath the narrative. Operating income swung from a loss of $403 million in fiscal 2024 to $2.334 billion in fiscal 2025 per Alpha Vantage — a reversal that is, by any honest accounting, remarkable. The operating margin sits at 30.31% on a trailing basis per stockanalysis.com, and free cash flow reached $2,905 million over the trailing twelve months per stockanalysis.com. I don’t want to wave these figures away. This is a real recovery, not a cosmetic one. But a real recovery priced at near-record multiples is still a valuation problem, and the question I keep returning to is: how much of the next two years is already embedded in the stock?

Consensus for FY26 points to revenue of $12.86 billion, up roughly 35%, with non-GAAP operating margins near 37% and EPS of $9.92 per the scenario model I’m working from. At the current price, you aren’t paying for that base case — you’re paying for something closer to the bull scenario, where AI data center demand stacks another 25% on top, pushing revenue toward $16.08 billion, margins to 42%, and EPS to $14.00, implying a fair value near $518 at a 37x multiple. That’s the optimistic arithmetic baked into the stock today, which means the margin of safety for a buyer here depends entirely on whether the AI buildout accelerates beyond what is already a generous consensus estimate. The bear case — an OEM inventory glut cutting revenue to $9.65 billion and EPS to $6.00 — implies a fair value near $222 at that same multiple. That asymmetry is the whole story.

The macro backdrop running underneath all of this is genuinely supportive, and I want to acknowledge that before I explain why it makes me more nervous, not less. High-yield credit spreads have compressed to 2.77% per ICE BofA/FRED, about as accommodating as credit markets get. When money is cheap and plentiful, hyperscalers fund capex cycles aggressively without flinching at the cost of capital, and that spending flows directly into WDC’s order book. The AI storage supercycle narrative isn’t invented — demand from cloud infrastructure is real, the NAND pricing environment has genuinely improved, and the past year’s margin expansion reflects actual pricing power, not accounting creativity. But this is precisely the moment when geopolitical supply chain risk deserves more weight than it’s getting. Escalating US-China tensions and the delayed semiconductor tariff picture add a layer of fragility to a supply chain that memory investors tend to ignore during the good years, and then can’t stop talking about during the bad ones.

What keeps me up at night is the utilization number, which almost nobody is discussing. US semiconductor capacity utilization sat at 69.7% as of Q1 2026 per FRED. Think about that alongside a stock trading near all-time highs. When a factory runs at roughly 70 cents of capacity on the dollar, the industry structurally has more room to produce than it currently needs. That’s not a picture of tight supply — it’s a picture of latent supply waiting for demand to justify it, or waiting for pricing to weaken when it doesn’t. The stock is priced like supply is constrained. The industrial data says it isn’t. I’ve seen this divergence before, and it tends to resolve in favor of the industrial data.

A similar setup played out with Micron (MU) in the 2017–2018 cycle, where a parabolic run fueled by smartphone and cloud memory demand ended in a painful drawdown once inventory accumulated faster than orders absorbed it. The pattern rhymes with WDC today in ways hard to dismiss: rapid profitability turnaround from a cyclical trough, elevated valuations, and a demand catalyst wearing the face of something newer and more permanent than it turned out to be. The crowded trade dynamic amplifies the risk. Short interest sits at 8.92% of the float per Yahoo Finance, against institutional ownership exceeding 101.7% of the float — which is a euphemistic way of saying that when sentiment shifts, there isn’t much natural buying left to cushion the exit.

The bullwhip effect — where massive AI cluster orders create the appearance of structural demand, only for hyperscalers to pause and digest inventory once build-out targets are met — is the thesis-killing scenario I’d watch most closely. Memory markets carry a long institutional memory of this pattern, even if individual investors don’t. The average analyst price target of $495.68 per Yahoo Finance sits only modestly above the current price, with a high of $660 and a low of $360 — that low already below where the stock trades. That’s not a lot of consensus cushion for a stock at an all-time high.

If WDC sustains non-GAAP operating margins above 38% for two consecutive quarters while semiconductor capacity utilization climbs back above 80% per FRED, my concern is wrong — that combination would confirm demand genuine enough to absorb the capacity overhang, and the valuation would have earned its ground.

But right now, I’d rather be the person who sold the certainty than the one who bought it.

© The Nonexpert · Original