Every Guided Line Beaten, Guide Above Consensus, Stock Down 13%: Exabyte Growth Halved and Price Did All the Sequential Work

Results vs. Consensus

Q4 FY26 Scorecard

Actual vs. Company Guidance

The guide being measured here is the one issued on the Q3 FY26 call at the end of April. Management's own summary was that revenue, gross margin and EPS "all came in at or above the high end of the guidance range," and the arithmetic supports it without qualification.

Year-over-Year Comparison

Sequential Comparison

Fiscal Year 2026 (first full year as a pure-play HDD company)

The four reported quarters sum to the fiscal-year revenue figure exactly, which is worth stating because the fiscal-year gross margin of 49.1% understates where the business now runs. Q4 alone printed 54.4%, and Q1 FY27 is guided to 55-56%. The annual average is a blend of a business that started the year at 46% and exited at 54%.

Revenue: the composition matters more than the headline

Revenue of $3,747M grew 44% against a year-ago quarter of $2,605M, and 12.3% against Q3's $3,337M. Both figures are, on their face, excellent. The disaggregation is what moved the stock. Bits shipped rose 4.1% sequentially while dollars rose 12.3%, which puts the implied realized price per terabyte up roughly 8% in three months. Management confirmed the direction explicitly, noting that the blended year-over-year price increase per terabyte "improved from high single digits last quarter to high teens this quarter."

Pricing power of that magnitude is the whole reason this stock has re-rated over the past year, so it is odd to treat it as bad news. The concern is narrower than that: in an industry where the bull case rests on structurally rising demand for capacity, the volume line is the demand signal and the price line is the negotiating outcome. When volume decelerates from +34% to +22% in one quarter and price accelerates from high-single-digit to high-teens, the revenue beat is being delivered by the less durable of the two engines. Management's answer, that customer technology mix and non-linear ordering create quarter-to-quarter noise, is credible and consistent with how hyperscale purchasing actually works. It is also unfalsifiable within a single quarter.

Margins: the incremental math is the most persuasive disclosure on the call

Non-GAAP gross margin of 54.4% expanded 1,310bp year-over-year and 390bp sequentially, clearing the 51-52% guide by a wide margin. The drivers management named were mix toward higher-capacity drives, favorable pricing across the portfolio, and manufacturing execution. Cost per terabyte fell roughly 8% year-over-year against a stated long-term glide path of about 10%.

The more useful framing came from the CFO, who redirected attention from the sequential gross-margin step to the year-over-year incremental. That series reads 60% for FY25, 75% for FY26, 84-85% in Q4 alone, and roughly 80-81% implied by the Q1 FY27 guide. Incremental gross margins in the eighties mean that four of every five incremental revenue dollars are dropping to gross profit, which is what allows a 12% sequential revenue gain to convert into a 29% sequential operating-income gain. Operating expenses fell to $382M, down $15M sequentially in absolute terms and 170bp as a share of revenue, at 10.2%. There is very little left to squeeze on the cost line, which means from here the margin story is entirely a gross-margin story.

EPS: no help from below the line, and a GAAP number to ignore

Non-GAAP diluted EPS of $3.56 rose 109% year-over-year and 31% sequentially. The build is unremarkable in the best sense: $1,655M of non-GAAP operating income, $10M of net interest and other expense, a 16.0% non-GAAP tax rate, and 388M diluted shares. Non-GAAP net income was $1,382M. Share count rose 10% year-over-year, so the EPS growth is understated relative to the profit growth rather than flattered by buybacks.

GAAP diluted EPS of $8.21 should be set aside entirely for valuation purposes. It contains a $2,050M non-cash mark-to-market gain on the retained SanDisk interest, partially offset by $362M of costs on the debt and equity transactions used to monetize it. For the full year the same line contributed $6,498M of gains, which is why FY26 GAAP EPS of $24.28 sits against non-GAAP EPS of $10.22. That gap closes permanently from here: the SanDisk position is now zero, so the mark-to-market line disappears from future statements. FY27 GAAP and non-GAAP earnings should converge to within the usual stock-compensation and realignment adjustments.

Segment Performance

WD reports three end markets. Cloud dominates and the other two are small enough that their growth rates matter more than their dollars.

The Total row is the income-statement figure. Cloud revenue is disclosed as a rounded "$3.3 billion" at 89% of total, so the value shown is the residual after the two precisely disclosed segments, and should be read as approximate to within the rounding.

Cloud

Cloud grew 43% year-over-year and held at 89% of revenue, the same share it has occupied for four consecutive quarters. This is now a nearline hyperscale business with a small consumer tail attached, and the concentration cuts both ways: it gives WD contracted multi-year visibility that no previous HDD cycle offered, and it means a handful of purchasing decisions set the quarterly volume line. The exabyte deceleration is a Cloud phenomenon by arithmetic, since Cloud is where nearly all the bits go.

Management widened the demand description meaningfully this quarter. Beyond the established hyperscalers, they cited neoclouds, sovereign buyers, frontier AI labs coming directly for supply, and physical-AI customers, with one autonomous-vehicle customer's calendar-2027 exabyte requirement described as up "multiple fold." They also flagged a reversal in on-premise enterprise, where buyers who had been moving toward all-flash arrays are pivoting back to hybrid.

"We definitely see increasing demand from neoclouds. Actually, even frontier AI labs are coming to us for supply. And as I just touched on to the earlier question, physical AI companies. So we've seen very strong demand from an autonomous vehicle company for more storage as they ramp up their autonomous vehicle capabilities."
— Irving Tan, CEO

Assessment: Customer-base broadening is the most underappreciated disclosure in the quarter. A nearline business selling to five hyperscalers is a price-taker with contracted volumes; the same business selling to hyperscalers plus neoclouds plus sovereigns plus AI labs plus physical-AI fleets has genuine allocation leverage, and management said as much when they described the tight supply environment as "an opportunity for increased pricing leverage." This does not fix the volume-growth question, but it does support the view that pricing strength is structural rather than a one-off negotiation win.

Client

Client revenue of $225M grew 61% year-over-year, the fastest growth of any line, from a base small enough that it contributes about 1.5 points to total growth. Management attributed the strength to pricing, specifically to the pricing of competing flash-based products setting a higher umbrella.

Assessment: This is a flash-scarcity spillover, not an HDD franchise improvement. It is real revenue and real margin while it lasts, and it should be modelled as cyclical. The disclosure that non-nearline gross margin now sits in the same range as nearline is the important part, because it means this incremental revenue is no longer margin-dilutive.

Consumer

Consumer revenue of $187M grew 38% year-over-year, its second consecutive quarter of acceleration after a period of decline. Same driver as Client: better pricing in a market where the flash alternative got more expensive.

Assessment: At 5% of revenue this line will not change the thesis in either direction. Its value is as a read on whether the storage shortage is broad or narrow, and the answer for two quarters running is broad.

Key KPIs

Two working-capital lines deserve a note. Receivables grew 36% year-over-year against revenue growth of 44%, so days outstanding actually improved. Inventories grew 17% against 44% revenue growth, which in a supply-constrained market is what you would expect and is the opposite of the inventory build that precedes a normal HDD downcycle. Neither line shows stress.

Key Topics & Management Commentary

Overall Management Tone: Confident on demand and margins, and noticeably framework-driven rather than quantitative when pressed on the two soft spots. Management met every question about the exabyte slowdown and the peer margin comparison with the same structure, which was to acknowledge quarter-to-quarter variability, restate a medium-term target, and decline to size the gap. That is a defensible posture for a business with genuine order lumpiness, and it is also the least satisfying answer available to a stock priced for acceleration. Tone was marginally less assured than the prior quarter, when the balance-sheet and gross-margin milestones let the numbers speak without interpretation.

1. Exabyte growth halved and management would not size the second-half reacceleration

The 231 exabytes shipped in Q4 grew 22% year-over-year, down from 34% in Q3, and the full-year figure was 25%. This was the dominant topic in Q&A, raised in different forms by two separate analysts. Management's framing was that different customers take different technologies in different quarters, so a customer weighted toward conventional recording in a given quarter yields fewer bits from the same number of drive units.

"So if there's a particular customer in a particular quarter that takes a bit more CMR products, obviously, that will mean we ship a fewer bits into the marketplace for the number of units that we have. And so in a different quarter where we have customers that are more ultra SMR heavy and for the same number of units, we're able to deliver much more exabytes into the marketplace. So there will always be variations quarter-to-quarter, and they don't buy on a very linear basis. It's a bit lumpy."
— Irving Tan, CEO

The forward commitment was specific on direction and silent on magnitude: exabyte demand growing "above 25%," accelerating in the second half of FY27 as the 40TB ePMR ramp pushes past 50% of nearline exabytes, with 44TB HAMR arriving in the first half of calendar 2027 and 50TB products in the second half.

Assessment: The mix explanation is mechanically sound. A customer buying CMR instead of UltraSMR genuinely does yield fewer exabytes per unit, and hyperscale ordering genuinely is lumpy. What management did not do is quantify how much of the 12-point deceleration was mix versus underlying demand, which they are surely able to compute. Absent that, an investor has to take the 25%-plus figure on faith for two more quarters. This is the one place where the story now requires trust rather than evidence, and it is why conviction comes down.

2. Pricing per terabyte moved from high-single-digit to high-teens year-over-year

The single largest contributor to the margin beat. Management framed pricing as the deliberate output of a "predictable and sustainable" strategy in which higher-capacity drives deliver better total cost of ownership to the customer and WD captures a share of that benefit. Nearline pricing moves on long-term agreement renewal schedules; non-nearline pricing moved faster because flash-based alternatives repriced upward.

"During the quarter, the blended average year-over-year price increase per terabyte improved from high single digits last quarter to high teens this quarter, reflecting the impact of our predictable and sustainable pricing strategy as we deliver greater value to our customers."
— Kris Sennesael, CFO

Assessment: A doubling in the rate of price increase within one quarter is a supply-shortage signature, not a contract-escalator signature, whatever the framing. Management was careful to reject the idea that contracted escalators are accelerating, which is the honest answer and also the one that makes the current pricing look more cyclical than structural. The pricing is real and will persist while supply is tight. It should not be extrapolated into the out-years at this rate of change.

3. Cost per terabyte fell 8%, short of the stated 10% glide path

Cost per terabyte declined approximately 8% year-over-year, against a mid-to-long-term target of about 10%. Management reaffirmed the 10% figure twice and attributed the shortfall to normal quarter-to-quarter variation, with the areal-density roadmap doing the work over time.

"So Aaron, the cost per terabyte in Q4 of fiscal '26, the quarter that we reported was down approximately 8% year-over-year. We've indicated before that cost per terabyte in the long term will come down on or about 10% year-over-year."
— Kris Sennesael, CFO

Assessment: An 8% cost decline against high-teens price increases produces spectacular incremental margins, so the miss against target is invisible in this quarter's results. It matters later. If pricing normalizes toward the low single digits while cost declines run at 8% rather than 10%, the gross-margin trajectory flattens faster than the current framework implies. Worth tracking as a two-variable spread rather than as a standalone metric.

4. 40TB ePMR began shipping and is the near-term volume lever

WD commenced shipments of next-generation ePMR drives at capacities up to 40TB during the June quarter and is now in volume production with two customers. The stated target is for that platform to carry more than 50% of nearline bits by the third quarter of FY27, which is also the mechanism management is relying on for exabyte reacceleration.

"For our next-generation 40-terabyte ePMR drives, we commenced shipments in our June quarter and are now entering volume production with two customers."
— Irving Tan, CEO

Assessment: This is the checkable near-term milestone, and it is a good one because it is specific and dated. Crossing 50% of nearline bits on the 40TB platform by Q3 FY27 mechanically lifts exabytes per unit shipped, so if that transition lands on schedule the volume-growth question resolves itself by the March quarter. If it slips, the volume line stays soft and the pricing-dependence problem gets worse.

5. HAMR 44TB on track for the first half of calendar 2027, and the competitive-timing question

Management confirmed the 44TB HAMR product remains on track to ship in the first half of calendar 2027, describing customer qualification feedback as "very positive," with capacity, performance and reliability exceeding customer expectations. The prior quarter's disclosure of four customers in qualification was not updated with a new count; the description shifted to a broadening qualification pipeline with additional hyperscale and cloud customers advancing.

"We are on track to ship our 44-terabyte HAMR product in the first half of calendar year 2027. Customer feedback on the qualification process continues to be very positive with the capacity, performance and reliability of our drives exceeding customer expectations."
— Irving Tan, CEO

Assessment: On-schedule is the right outcome and the qualification language is encouraging. The strategic issue is relative timing rather than absolute: WD's principal competitor is already shipping HAMR at meaningful scale, which means WD spends the next three quarters bridging with ePMR while a rival converts a technology lead into margin. WD's answer, that 40TB ePMR plus UltraSMR delivers competitive capacity economics in the interim, is supported by this quarter's gross margin of 54.4%, which was higher than the competitor's 52.7% in the same period. The gap to watch is not current margin but forward slope.

6. UltraSMR ramping with a third major customer, targeting 60% of nearline exabytes

UltraSMR is now ramping with the third major customer, and management expects the technology to account for approximately 60% of nearline exabyte shipments as WD exits fiscal 2027. This is the same adoption vector flagged in prior quarters, now with a quantified endpoint.

"We're currently ramping our UltraSMR technology with the third major customer, and we expect that UltraSMR will make up around 60% of our nearline exabyte shipments as we exit fiscal 2027."
— Irving Tan, CEO

Assessment: The 60% target is the most useful new number on the call, because UltraSMR is precisely the mix element management blamed for the exabyte shortfall. If UltraSMR reaches 60% of nearline exabytes by the end of FY27, the mix drag that suppressed this quarter's volume becomes a mix tailwind, and the exabyte line should recover without any change in underlying unit demand. This claim and the 40TB ramp are the same bet stated two ways.

7. Long-term agreements now under negotiation for calendar 2029, 2030 and 2031

Contract visibility extended again. Last quarter's disclosure was a single large-customer agreement reaching calendar 2029; management is now in active discussions covering calendar 2029, 2030 and 2031. Volumes are described as well understood; pricing is explicitly the open item.

"The last time we reported on LTAs, we talked about having one LTA of a large customer all the way up to calendar year '29. But we're very much in the throes of discussions with customers to establish LTAs for calendar '29, '30 and '31 as well. ... What we're working through with them is more in terms of the pricing commercial construct of what those LTAs would look like going forward."
— Irving Tan, CEO

Assessment: Customers volunteering to contract five years forward is the strongest available evidence that hyperscale buyers regard nearline HDD capacity as a scarce strategic input rather than a commodity purchase. It is worth being precise about what is and is not committed: management says it has "good line of sight" on volumes and is still negotiating price. Volume visibility without price visibility is genuinely valuable for capacity planning and considerably less valuable for forecasting revenue. The bull case treats these agreements as contracted revenue; they are, more accurately, contracted demand.

8. High-bandwidth drives sampling with five customers

The performance-oriented product line expanded from three customers sampling to five, targeting up to eight times the throughput of current drives without a proportional increase in power draw. Management positioned it as an entry into an additional layer of the AI storage stack rather than a capacity play.

"We are making progress on improving drive performance with our high-bandwidth drives and are now sampling with 5 customers. We are targeting up to 8x the throughput of today's drives without the corresponding increase in power draw, exactly the kind of performance AI workloads require."
— Irving Tan, CEO

Assessment: Still pre-revenue optionality, and still the most interesting thing in the pipeline that nobody is paying for. Eight times throughput at flat power would let HDDs contest workloads currently ceded to flash on performance grounds. Sampling with five customers up from three suggests genuine interest. No revenue timeline was offered, so this stays outside the model and inside the option value.

9. The demand narrative moved from training to inference, agentic and physical AI

The CEO devoted most of his prepared remarks to a structural argument: compute cycles are reused while data compounds, so as AI workloads shift from training to inference to agentic operation to physical systems, storage demand becomes a function of accumulated data rather than of infrastructure build cycles. Physical AI, meaning autonomous vehicles, robotics, industrial automation and humanoids, was given materially more airtime than in prior quarters, on the basis of improved customer visibility.

"Today, roughly 80% of data stored in a hyperscale data center resides on hard disk drives. That is likely to continue. That reflects what HDDs do exceptionally well, delivering the scale, economics and power efficiency required for long retention large-scale data storage."
— Irving Tan, CEO

Assessment: The argument is sound and it is the correct frame for a long-duration HDD thesis, because it decouples storage demand from the capex cycle that will eventually turn. The 80% share figure is the load-bearing claim, and it is one worth monitoring rather than accepting permanently: it is exactly the number that a sustained decline in flash cost per terabyte would erode. For now, flash pricing is moving in WD's favor, which is why the Client and Consumer lines grew 61% and 38%.

10. Balance sheet finished, capital returns steady, dividend held flat

The remaining 1.7M SanDisk shares were exchanged for 4.8M WD shares, completing the monetization begun three quarters ago and removing the retained interest from the balance sheet entirely. WD repurchased 2.3M shares for $1B in the quarter, a figure that includes $328M used to settle a convertible conversion premium in cash rather than stock and thereby avoid issuing roughly 773,000 new shares. The quarter closed with $1,052M of debt against $1,579M of cash. FY26 capital returns totalled approximately $3.1B. The declared dividend was $0.15, unchanged from the prior quarter.

"We also made $54 million of dividend payments, and we ended the quarter with $1.1 billion of debt and $1.6 billion of cash, resulting in a net positive cash position of $500 million."
— Kris Sennesael, CFO

Assessment: Capital allocation has been the least controversial part of this story for a year and remains so. Management explicitly declined to re-lever, committing to return free cash flow through dividends and buybacks with "no change" to strategy. Using cash to settle a conversion premium and avoid share issuance is a small, sensible, shareholder-friendly decision of the kind that tends to correlate with good capital discipline elsewhere. The flat dividend after three consecutive raises is a break in pattern rather than a signal, given that buybacks absorbed $1B in the same quarter.

11. Capacity growth without unit capacity spending

Management stated that delivering substantially more exabytes does not require capital spending to add unit capacity, with investment directed instead at heads and media operations and at automation for productivity. Q4 capex was $108M, or 2.9% of revenue, against $145M the prior quarter.

Assessment: This is the economic heart of the current cycle and the reason free cash flow margins reached 34%. Exabyte growth delivered through areal density rather than through added units means the incremental exabyte carries almost no capital cost, which is what separates this cycle from every prior HDD upcycle where capacity additions destroyed the pricing that justified them. It also means the industry's supply discipline is a technology constraint rather than a choice, which is more durable than restraint. No FY27 capex figure was given.

Guidance & Outlook

The guide is internally consistent and it is strong. Working it through from the components: $4,100M of revenue at the 55.5% gross-margin midpoint produces $2,276M of gross profit, less $395M of operating expenses gives $1,881M of operating income at a 45.9% margin, less $15M of interest and taxed at 17% leaves roughly $1,548M of net income, which on 388M shares is $3.99. The guided $4.00 is exactly what the stated components produce, with no unexplained residual.

Revenue growth of 45% year-over-year at the midpoint is an acceleration from Q4's 44%, and gross margin expands another 110bp on top of a quarter that already beat its ceiling by 240bp. Three independent characterizations of the guide place it above pre-print consensus. On its own terms this is a beat-and-raise.

Implied year-over-year incremental gross margin: approximately 80-81%, per management, versus 84-85% delivered in Q4. Slightly lower, still exceptional.

What is not guided: nothing beyond one quarter. Management was asked directly whether the September quarter represents the low point for sequential revenue growth and incremental margins, with acceleration through FY27, and declined to answer beyond restating the one-quarter-at-a-time policy and expressing "high conviction" in continued improvement.

Guidance style: demonstrably conservative and consistently so. WD has now beaten the high end of its own revenue guidance in each of the last four quarters and beat the gross-margin ceiling by 240bp this quarter. A management team with this pattern guiding 55-56% is, on the historical evidence, more likely to deliver 56-57%. That asymmetry is a meaningful part of why we stay constructive despite the volume question.

Analyst Q&A Highlights

Peer-relative margin slope and whether HAMR is required to go higher

The call opened on the competitive comparison rather than on the results, which tells you where the buy side's attention already was. The question put WD's September gross-margin guide directly against the principal competitor's, noting a gap of nearly two percentage points, and asked whether closing it requires waiting for HAMR in calendar 2027. Management's response catalogued the drivers of quarter-to-quarter margin variation without engaging the comparison itself.

Q: "It's hard not to compare your results with your main competitor where they're seeing better sequential top line growth and targeting gross margins nearly 200 bps higher than your September guide. So curious what you make of this? Is that due to the earlier ramp of HAMR? Is it a difference in timing of contracts, perhaps you're selling fewer bits into the open market? Would love to hear your thoughts there and how you think about kind of gross margin acceleration from here? Do we need to kind of wait for HAMR in calendar '27? Or are there other drivers that can allow you to push higher?"
— C.J. Muse, Cantor Fitzgerald

A: "There are always quarter-to-quarter variations in terms of gross margin because there's a whole range of factors that go into driving it. Obviously, timing of pricing in relation to long-term agreement contracts is one. So there will always be timing differences of when existing LTAs expire and new LTAs kick in with different pricing regimes. ... Right now, we are very much on track to hit 50% of nearline bits on that platform by third quarter of fiscal year '27."
— Irving Tan, CEO

Assessment: The three candidate explanations offered in the question, HAMR timing, contract timing, and open-market bit allocation, were all plausible and none was confirmed or denied. Management redirected to the 40TB ePMR milestone, which is a real answer to "what drives margin from here" and not an answer to "why is your slope shallower than theirs." Notably, WD's reported gross margin this quarter was higher than the competitor's, so the gap being discussed is entirely forward-looking. That framing was left uncorrected, which was a missed opportunity.

Exabyte deceleration: temporary mix effect or the new normalized rate

The most-pressed topic on the call, raised independently by two analysts and returned to in follow-ups. The framing was direct: 231 exabytes at roughly 21-22% growth sits well below the 30%-plus of recent quarters, so is this timing, supply, or a reset baseline. Management held the line that the medium-term rate is above 25% and that the second half of FY27 accelerates on product transitions.

Q: "Exabyte shipments were up about 21%, I think, year-over-year to 231. That's below the 30% exabyte growth you folks have had in the last several quarters. Can you just talk about, is this a temporary pause related to timing or supply? Or does this represent a more normalized growth rate? And how should we think about exabyte growth through fiscal '27?"
— Multiple analysts incl. Amit Daryanani, Evercore; Wamsi Mohan, Bank of America

A: "I think what we've guided to is that exabyte, we see demand growing at above 25%, right, 25% plus going forward. So we clearly have a road map that's designed to support that. That will, again, as I mentioned, similar to margins, be quarter-to-quarter fluctuations in terms of exabyte growth rates depending on the mix of products that we ship. And as we move into the second half of the year, as we really ramp up the 40 terabyte shipments, as I mentioned, that will be over 50% of the nearline exabytes we ship. We expect exabyte growth rate to accelerate."
— Irving Tan, CEO

Assessment: Management answered the "what happens next" half and left the "what just happened" half unquantified. The follow-up pressed specifically on whether there is a near-term ceiling on bits before HAMR ramps, which is the sharpest version of the question, and the reply again went to customer lumpiness and the medium-term target. An investor is left holding a directional promise with a specific mechanism (40TB crossing 50% of nearline bits by Q3 FY27) and no sizing. The mechanism is checkable, which is the saving grace.

Whether long-term agreements cap the pricing upside

A more sophisticated line of questioning than the simple pricing-trajectory question: if roughly 90% of bits go to cloud and most of that is under long-term agreement, then locking in multi-year contracts during a shortage should mechanically slow the velocity of price increases. Management confirmed the structure without conceding the conclusion, noting that agreements stagger their start and end dates so repricing occurs continuously through the year.

Q: "With 90% of your bits going into cloud and most of them in LTAs and those LTAs extending longer, the velocity of price increases that we are seeing going from high single digits to high teens last quarter and probably 20% in September quarter, would that slow down because most are under LTA? Or do you think there's still flexibility where the velocity of price increases could still increase?"
— Krish Sankar, TD Cowen

A: "Obviously, not every LTA starts and ends at the same time. So we will see price adjustments from LTAs even as we go through the fiscal year, depending on when the contracts kick in with the new pricing. ... Where we've seen a bit more ability to drive more rapid price increases has been in the non-nearline space."
— Irving Tan, CEO

Assessment: The most revealing exchange on the call. Management conceded that the fastest pricing gains came from the non-contracted, non-nearline business, which is 11% of revenue and the most cyclical part of the mix. That concession implicitly confirms the questioner's premise: the contracted 89% reprices on a schedule, at a deliberately "predictable" pace, so the high-teens blended figure is being pulled upward by the small volatile tail. It argues for modelling nearline price increases as smoother and lower than the blended number suggests.

Cost per terabyte through the 40TB product transition

A question about the cost side of the equation during a platform changeover, noting that cost per terabyte looked roughly flat to slightly down sequentially. Management gave the year-over-year figure of approximately 8%, restated the roughly 10% long-term expectation, and tied the trajectory to areal density.

Q: "It seems to me like you've got a product transition dynamic going on with the 40-terabyte ePMR drives. If I look at your cost structure, it looks like you saw a cost per terabyte that was more maybe flattish, down a little bit sequential. I'm curious, as we move through this product transition, how should we think about cost per terabyte or cost of goods sold per terabyte progressing through '27?"
— Aaron Rakers, Wells Fargo

A: "So Aaron, the cost per terabyte in Q4 of fiscal '26, the quarter that we reported was down approximately 8% year-over-year. We've indicated before that cost per terabyte in the long term will come down on or about 10% year-over-year. And the main driver ... is the move to higher capacity drives, better areal density, which gives us a substantial reduction on a cost per terabyte."
— Kris Sennesael, CFO

Assessment: A straight answer with a number, which was not the norm on this call. The 8% against a 10% target is a small miss that nobody minds while pricing runs at high teens. The exchange is useful mainly for establishing the spread: roughly 8% cost deflation against high-teens price inflation is what produces 84% incremental gross margins, and any modelling of FY28 onward should stress-test what happens when the price side reverts toward the cost side.

Capital structure now that the debt is nearly retired

With debt down to about $1.1B against $1.6B of cash, the question was whether the company intends to run at net cash permanently, re-lever, or simply return everything it generates. Management gave an unambiguous commitment to the existing framework.

Q: "You've done a lot on the balance sheet, right? You've got a $1.1 billion left on kind of the debt. You've got $1.6 billion in cash. I'm just curious, how are you thinking about the capital structure from here? Is it pretty much 100% free cash flow return? Do you want to put more debt on the balance sheet? I'm just curious if any updated thoughts on that."
— Aaron Rakers, Wells Fargo

A: "No change there to our strategy and our commitment to return the free cash flow back to the shareholders consistently with what we have been doing over the last 4 or 5 quarters through a combination of our dividend payments as well as our share repurchases. And so we're fully committed to those two programs."
— Kris Sennesael, CFO

Assessment: Exactly the right answer for a business at this point in its cycle, and the clearest commitment given on the call. No leveraged buyback, no acquisition ambition, no balance-sheet financial engineering while the operating cycle is doing the work. Against FY26 free cash flow of $3,511M, a full-return policy funds meaningful share-count reduction at the current de-rated price.

Non-hyperscale demand: sovereigns, neoclouds, frontier labs and physical AI

A follow-up on a prior-quarter comment about early interest from sovereign and neocloud buyers, asking what they buy, how price-sensitive they are, and how large they could become. The response converted a tentative prior disclosure into a materializing one and added frontier AI labs and physical-AI customers.

Q: "Last quarter, you mentioned starting to see some interest from sovereign and neocloud data centers for nearline drives in your systems. Can you maybe just tease that out a bit more now that we're 3 months beyond that comment? Just what are they buying? How price sensitive are they? How big are they? How big could they get, et cetera?"
— Erik Woodring, Morgan Stanley

A: "I think we looked at that as an opportunity and that opportunity is materializing today. ... We are seeing more demand for sovereigns, from neoclouds, from AI frontier labs as well as they particularly see the economic benefits from hard drives and the increasing storage demands that their business models are creating as well. In terms of pricing, I would say they are in the very tight supply environment that we are in, that provides us an opportunity for increased pricing leverage."
— Irving Tan, CEO

Assessment: The most genuinely new information on the call, and it was volunteered rather than extracted. Buyer-base diversification away from a handful of hyperscalers changes the negotiating structure of the business, and the explicit statement that these buyers offer "increased pricing leverage" is the clearest indication that pricing is supply-driven. No sizing was given, which keeps it out of the model, but it materially supports the durability half of the thesis.

Whether the September quarter is the trough for sequential growth

A direct attempt to extract a shape for FY27: is Q1 the low point for both sequential revenue growth and incremental margins, with acceleration after. Management declined on policy grounds while signalling the direction.

Q: "Given the product ramp and improving demand environment, should we expect the first quarter fiscal to represent the low point for both sequential revenue growth and incremental margins with acceleration as we move throughout the year with regard to FY '27."
— Ben Reitzes, Melius

A: "As you know, we only guide one quarter at a time. ... We have high conviction that we can continue to ship a lot more exabytes at better pricing and driving down the cost over time, which will lead to continuous strong revenue growth as well as further gross margin improvement. And I'll leave it at that."
— Kris Sennesael, CFO

Assessment: "I'll leave it at that" is a deliberate stop, and a reasonable one for a company that has never guided beyond a quarter. The substance leans affirmative without committing. For a stock that had appreciated more than fivefold over twelve months entering this print, a refusal to validate the acceleration shape was always going to be read as a soft negative, regardless of policy consistency.

What They're NOT Saying

1. How much of the exabyte deceleration was mix versus demand. Management has the internal data to decompose a 12-point growth swing into product-mix and underlying-unit components. It was asked three different ways and answered qualitatively every time. This is the most consequential omission of the quarter.
2. Any FY27 framing at all. No annual revenue, EPS, exabyte or margin guidance, and an explicit refusal to confirm the shape of the year when asked directly. Defensible as long-standing policy, poorly timed for a stock that needed the reassurance.
3. Pricing on the calendar 2029-2031 agreements. Management volunteered that volumes have "good line of sight" and that the "pricing commercial construct" is still being negotiated. The extended visibility is therefore demand visibility, not revenue visibility, and the distinction was not drawn for listeners.
4. An updated HAMR qualification count. The prior quarter gave a specific number of customers in qualification. This quarter substituted qualitative language about a broadening pipeline. The schedule was reaffirmed, so this is most likely presentational, but dropping a quantified metric that had been trending favorably is a choice.
5. Any response to the competitive margin-slope comparison. The first question of the call laid out a specific forward gap against the main competitor and offered three possible explanations. None was addressed. WD's current-quarter margin is actually the higher of the two, which was a straightforward rebuttal available and unused.
6. FY27 capital expenditure. Management said unit-capacity spending is not required and named heads, media and automation as investment areas, without a dollar figure. Capex has run between $108M and $145M quarterly; the absence of a number is notable mainly because capital intensity is central to the free-cash-flow case.
7. Customer concentration. Roughly 89% of revenue is cloud, and management repeatedly referenced "our three largest customers" without disclosing what share they represent. Standard practice, and still the largest unquantified risk in the story.
8. Why the dividend was held flat. After three consecutive quarterly increases, $0.15 was declared unchanged with no commentary. Buybacks took $1B in the quarter, so capital return did not slow, but the streak ended without acknowledgement.

Market Reaction

• Pre-print setup: WD closed at $519.17 on August 5, the day of the release. The stock entered the print up 201.4% year to date against 12.8% for the S&P 500, and up 584.6% over twelve months. It was, however, already correcting: down 10.1% over the trailing 30 days from $577.46, and 30.4% below its 52-week closing high of $746.23. The regular session on release day itself fell 5.4% from the prior close of $548.56.
• Reaction session (August 6): opened at $428.89, a 17.4% gap down, traded a range of $407.48 to $477.63, and closed at $451.52, down 13.0% or $67.65. Volume was 16.2M shares against an 8.7M 30-day average, at 1.9 times normal.
• Follow-through (August 7): a further decline of 3.8% to $434.30, leaving the shares 41.8% below the 52-week closing high and 20.8% below the pre-release August 4 close.
• Index and peers: the S&P 500 fell 0.2% on the reaction session. The decline was concentrated in storage and memory rather than in WD alone, with the principal HDD competitor and the flash names also lower on the day. WD was the weakest performer in the group.

A 13% single-session decline on a quarter that beat consensus on both lines, beat the high end of every guided metric, and guided the following quarter above consensus requires an explanation that is not in the numbers. There are three, and they compound.

The first is positioning. A stock up nearly sixfold in twelve months and up 201% year to date does not clear bars, it clears expectations, and the expectations embedded at $519 were for acceleration on every axis simultaneously. Sell-side price targets averaged well above the pre-print price entering the quarter, which is a description of consensus optimism rather than of upside. When the whole float is long and the print is merely very good, the marginal buyer has already bought.

The second is the exabyte line, and this is the part that is analytically legitimate rather than merely technical. Volume growth halving while price growth doubles is a genuine change in the character of the revenue, and management declined to quantify how much of it was transitory. For a thesis whose central claim is structural demand growth for capacity, the volume metric is the thesis metric. One soft quarter is not a break, but it converts a fact into a promise, and promises are worth a lower multiple than facts.

The third is peer-relative. The competitive comparison was the first question asked, before anyone discussed WD's own results, and it went substantively unanswered. Investors holding both names had a fresh datapoint from a rival that reported a week earlier and rose on it, and a forward margin slope that looks steeper. In a sector where positioning was crowded, relative preference is expressed by selling the second-favourite.

What none of the three amount to is a deterioration in the business. Every operating line improved sequentially and year-over-year, the guide implies further improvement, and the balance sheet finished the quarter in net cash. The de-rating is a change in what investors will pay for those results, from 29 times forward earnings before the print to 24 times after. That is a multiple event, and multiple events are where the opportunity usually is.

Street Perspective

Debate: Is the exabyte slowdown a mix artefact or the beginning of demand normalization

Bull view: The mechanism management described is real and arithmetically verifiable. A customer taking conventional recording instead of UltraSMR yields fewer exabytes from the same units, and hyperscale ordering has never been linear. With 40TB ePMR crossing half of nearline bits by the third fiscal quarter and UltraSMR reaching roughly 60% of nearline exabytes by the end of FY27, the mix drag becomes a mix tailwind on a known schedule. Full-year exabytes still grew 25%.

Bear view: Volume is the demand signal and it halved. Revenue growth is now carried by price in an industry that has destroyed its own pricing in every prior cycle. Management could quantify the mix effect and chose not to, which is what companies do when the residual is unflattering. Two quarters of high-teens price increases with decelerating volume is the classic profile of a shortage in its late innings rather than a structural re-rating.

Our take: The bull explanation is more likely correct and the bear objection is better positioned. The mechanism is credible, dated and checkable, which is unusual and valuable. But the refusal to size it means the market is right to apply a discount until the Q1 or Q2 print either shows the reacceleration or does not. We weight the mix explanation at roughly two-to-one and size the position accordingly rather than paying up for it.

Debate: Does the competitor's HAMR lead translate into durable margin advantage

Bull view: It has not so far. WD's gross margin of 54.4% in the June quarter exceeded the competitor's 52.7% in the same period, despite the rival being further along on heat-assisted recording. WD's bridge of 40TB ePMR plus UltraSMR is delivering competitive capacity economics without the yield and reliability risk of a new recording physics, and 44TB HAMR arrives on schedule in the first half of calendar 2027 with qualification feedback described as exceeding expectations.

Bear view: Current margin is the wrong metric; slope is the right one. The competitor is guiding a materially steeper sequential gross-margin step and a higher operating margin on the same September revenue guide, off a smaller base growing faster. Areal density compounds, so a two-year technology lead becomes a permanent cost-per-terabyte advantage that shows up exactly when pricing normalizes and cost position determines who keeps the margin.

Our take: The bear case has the better argument on the long-run economics and the bull case has the better evidence on the near term. WD's cost per terabyte fell 8% against a 10% target, which is the metric where a density disadvantage would first appear, and it is a small miss rather than a divergence. We treat this as the primary structural risk to the thesis and the main reason not to hold this name at a premium multiple. It is a reason to own less, not a reason to own none.

Debate: Is 24 times forward earnings cheap or still expensive for a cyclical

Bull view: The business earns 44% operating margins, converts a third of revenue into free cash flow, holds net cash, returns everything it generates, and has demand contracted in principle out to 2031. FY27 earnings power near $18 puts the shares at roughly 24 times after a 42% drawdown from the high, with estimates rising rather than falling. Sell-side targets have barely moved and no downgrades have followed the print.

Bear view: This is a two-player hardware industry that has spent thirty years proving it cannot hold pricing, and 24 times is a growth multiple applied to peak-cycle earnings. If pricing reverts toward cost deflation, FY28 earnings could be materially below FY27 rather than above it, which makes the current multiple 24 times a number that never repeats. The correct comparison is not to a compounder but to prior HDD peaks, where the stock traded at single-digit multiples of cycle-peak earnings.

Our take: Both sides are arguing about the durability question rather than the valuation, which is correct, because at 24 times the entire answer depends on whether FY27 is a peak or a waypoint. The structural evidence, that exabyte growth is delivered through areal density without unit capacity additions, is the strongest reason to believe this cycle differs from its predecessors: the supply discipline is a physics constraint rather than a promise. We think that argument holds and that 24 times is reasonable rather than cheap. It is not the multiple to be maximally long at, which is why conviction moderates while the rating holds.

Model Update Needed

Valuation impact: Fair value range moves to $465 to $575, from $500 to $620 previously. The range comes down modestly even though the earnings estimate rises sharply, because the multiple assumption falls further than the estimate rises. The prior framework applied 38-48 times to FY27 earnings of $13 to $15; both halves of that were wrong in offsetting directions. Earnings power is materially higher than we assumed, and the multiple the market will pay for it is materially lower.

Against the August 7 close of $434.30, the revised range implies 7% downside protection at the low end and 32% upside at the high end, with the $520 midpoint implying roughly 20% upside. The shares trade at 24.3 times FY27E, against 29.0 times at the pre-print close and 41.7 times at the 52-week high.

Thesis Scorecard Post-Earnings

This is the first quarter WD has been scored against a formally recorded thesis rather than a scorecard assembled at the time of writing. The pillars below are the ones carried forward from the prior quarter's coverage.

Overall: Thesis intact on execution, weakened on durability evidence. Four of five bull pillars confirmed outright and one of them, the balance sheet, now complete. The single pillar that weakened is the one that matters most, because structural capacity demand is the reason to own an HDD manufacturer at any multiple above cyclical trough. Both technology-related bear points moved in the wrong direction without producing any actual damage to reported results.

Action: Hold the position, do not add at the current level, and treat the Q1 FY27 exabyte figure as the decision point. Maintaining Outperform on a 12-month view given 24 times FY27E earnings, a 45% revenue growth guide, net cash, and a full free-cash-flow return policy. Conviction moves to Moderate from High because the volume line converted the central claim of the thesis from demonstrated to promised. A Q1 print showing exabyte growth back above 25% restores High conviction; a second quarter in the low twenties with pricing still carrying the revenue line would move this to Hold regardless of the reported margin.