WDC Q1 FY27 — Price Did the Work Twice. The December Guide Asks It to Do It Again.

The setup

Western Digital’s December quarter will be carried by price again: exabyte growth stays below 25% while revenue lands inside the guide, which is the outcome management has just told investors not to expect. The September quarter is why we think so, because exabytes shipped grew 22% for the second quarter running, below the 25% the September dive set as the line, while revenue landed at $4,381M against a $4.1B guide. Stated as a test: Western Digital reports fiscal second-quarter exabyte growth below 25% year on year, which on the 215 exabytes of the December 2025 quarter means fewer than 268.75 exabytes, while revenue lands inside its $5.0B to $5.2B range. The 215-exabyte base is the figure reported in summaries of the January call; we have not confirmed it against a primary transcript. We are wrong if exabytes print at 25% or more, because that is the quarter management has now told us to expect.

The September piece put the test in one sentence. If first-quarter exabytes came back at or above 25% with revenue at the $4.1B guide, the fourth quarter’s 22% was customer-mix lumpiness and the read was wrong. If exabytes printed below 25% again and revenue still landed, price had done the work in back-to-back quarters. On October 22, CFO Kris Sennesael said: “We delivered 248 exabytes in the quarter, up 22% year-over-year and up 8% sequentially.” Revenue was $4,381M, 6.85% above the $4.1B midpoint and 4.31% above the top of the range. The test resolved on the confirming side. (The September piece gave October 29 as the report date from a vendor calendar; the company reported a week earlier.)

None of that makes the quarter a bad one. It was the best quarter in the company’s post-separation history on every line it reports: revenue up 55%, GAAP gross margin 57.8%, non-GAAP operating margin 49.1%, non-GAAP EPS of $4.64 against a $4.00 guide, operating cash flow of $1,604M and free cash flow of $1,377M. The point is narrower and it is about the ingredients. Sennesael gave both of them on the call. Exabytes were up 22%. “Year-over-year growth in average price per terabyte further increased to 28%.” Multiply 1.22 by 1.28 and you get 1.56, which is the 55% the release printed, within rounding. Three-quarters of the dollar growth in the quarter was price.

The stock had already done some of this arithmetic. WDC closed at $441.57 on September 3, the day the September dive published, and at $354.19 on October 22 before the release, down 19.79%. It fell 6.43% in the regular session of the release day itself, from $378.51. On October 23 it opened at $345.98, down 2.32%, traded to $339.55, and was at $354.06 at 9:06 a.m. PT, flat. A beat of 6.85% on revenue and 16.0% on EPS against guidance, and the stock is unchanged. The market is not arguing with the numbers. It is arguing about which of the two ingredients it is paying for.

The numbers

Read the Sankey from the right, and notice what is missing. A year ago net income of $1,182M sat on operating income of $792M because $611M of it was a mark on the retained Sandisk stake. In the June quarter the same line carried a $2,050M gain. This quarter interest and other was an $8M expense, and the 46.7% operating margin passed almost untouched to a 40.6% net margin, with GAAP and non-GAAP EPS one cent apart. For the first time since the separation, the GAAP income statement is just the drive business. That makes it the cleanest quarter to decompose, and the decomposition is where the argument lives.

The September test, graded

The standing rule here is to read the ingredients, not the level: when revenue is two disclosed variables multiplied together, a steady headline growth rate says nothing about whether the business underneath it is steady. Western Digital gives both variables every quarter. Put them in one table across four quarters and the story is already written.

In the December 2025 quarter revenue grew 25.2%. Exabytes grew 22% on 215 shipped, which leaves about 2.5% for price. In March revenue grew 45.5% on exabytes up 34%, and CEO Irving Tan, asked directly, said pricing was up 9%. In June revenue grew 43.8% on exabytes up 22%, which leaves 18.0% for price, the “high teens” Sennesael described. In September revenue grew 55.5% on exabytes up 22% and price up 28%. Exabyte growth printed 22% in three of the four quarters. Price per terabyte went from roughly 2.5% to 28% across the same four. The revenue line in the chart below does not follow the blue line. It follows the gold one.

Sennesael on the volume line. On the call he put the exabyte number in context before an analyst could: growth is “slightly below the 25%” because the company is “in between those 2 product transitions,” and “as the product transitions are getting more traction, we will exceed that 25%.” Decode it against the September call. In August the explanation for 22% was customer mix: a large customer taking more conventional-recording drives, and therefore fewer bits per unit. In October the explanation for 22% is a product transition. Both can be true. But two different explanations for the same number in consecutive quarters is the pattern the house rule about mix exists for, a mix-driven move is direction-agnostic and you do not bank it until it prints twice. It has now printed twice, in the direction the explanation said it would not.

Sequentially the picture is the same. Exabytes went from 231 to 248, up 7.36% on the rounded figures (management said 8%). Revenue went from $3,747M to $4,381M, up 16.92%. Revenue per exabyte rose 8.91% in one quarter. A little over half the sequential revenue gain arrived without an additional exabyte leaving a factory.

The end markets say it in the 10-Q’s own words. Cloud revenue, 88% of the total, rose 54% “driven by a more than 20% increase in exabytes shipped along with an approximately 25% increase in average selling price per exabyte.” Client rose 119% on about 15% more exabytes and about 90% more price per exabyte. Consumer rose 19% on about 30% fewer exabytes and about 70% more price, and the filing names the cause: “general softness in demand in these end markets as a result of the increased pricing levels.” That last sentence is the first place in a Western Digital filing this year where price has visibly cost volume. It is 5% of revenue and nobody should build a thesis on it. It is still worth writing down, because it is what price elasticity looks like when it starts at the edge of a market rather than the middle.

Every point of margin since January came from the price line

The September dive found that cost per terabyte had slowed its decline from about 10% a year to about 8%. The October call took it further. Sennesael: “in the near term, our cost per terabyte will be less than 10% down year-over-year.” And then the reason, which is new: “We are facing some input cost increases,” because “we use some DRAM and some NAND inside our hard disk drives.”

The company does not disclose cost per terabyte as a number, but it discloses both halves of a cruder version every quarter. Divide GAAP cost of revenue by exabytes shipped and you get an all-in cost per exabyte. It is crude because it covers every product, not nearline drives alone, and it moves with mix. It is still the only version a reader can compute, and its trend is unambiguous.

The 10-Q on gross profit. The filing attributes the $1.31B gross profit increase to “improved pricing, an increased volume of shipments, and a lower cost structure on our newer generation products.” Decode the third item against the series. Cost of revenue per exabyte was $7.61M in the December quarter, $7.48M in March, $7.44M in June and $7.45M in September. It fell 1.7% from December to March, 2.2% across the whole span, and has been flat for three quarters. Against the implied year-ago figure of $7.83M it is down 4.84%, which matches the 4% to 5% decline an analyst put to management on the call and which management did not dispute. Revenue per exabyte, over the same five quarters, went from $13.86M to $17.67M. A “lower cost structure” is real at the product level, and the newer drives are cheaper per terabyte to make. At the level of the whole income statement it has been offset by mix and by input costs since January. The gross margin went from 45.7% to 57.8% over those three quarters on a cost line that did not move. Every point of it came from the top line.

The incremental margin says the same thing from the other side. Year on year, revenue rose $1,563M and gross profit rose $1,307M: an incremental gross margin of 83.6%, which is Sennesael’s “well above 80%.” Sequentially it was 79.8%. An incremental margin above 80% on a manufactured product is only possible when most of the incremental revenue is price. If it were volume, each new exabyte would bring its $7.45M of cost with it and the incremental margin would sit near the average margin, not 25 points above it.

Inventory, and the input Sennesael named. Total inventories rose 17.87% in the quarter to $1,781M, slightly faster than revenue’s 16.92%. The cash-flow discussion explains it as a ramp: “a $269 million increase in inventories as we ramped production in response to growing demand.” Decode it by component. Finished goods rose 3.53%, to $411M. Work-in-process rose 12.79%, to $935M. Raw materials and component parts rose 52.63%, from $285M to $435M. A production ramp shows up in work-in-process. A 52.63% build in purchased components, in the quarter the company first told investors that DRAM and NAND inside its drives were getting more expensive, looks like buying the input ahead of the next price rather than staging the next quarter’s units. That is a sensible thing to do, and it is the same move this site has watched server makers make all autumn. It is also a cost that arrives in cost of revenue a quarter or two later, at the price paid now. Days of inventory rose to 87.7 from 80.0 on cost of revenue.

Put those together and the margin structure is clear. Western Digital’s cost per exabyte is no longer falling at the company level, its stated near-term cost-down is below its long-run 10%, it is paying more for two of its components, and it has stocked up on them. The guide for a 60% to 61% non-GAAP gross margin in December, up from 58.1%, therefore assumes price rises again. That is not a criticism of the guide. It is a description of what the guide is made of.

“The floor” and the two transitions

Tan on the exabyte line. Asked by Bank of America’s Wamsi Mohan whether fiscal 2027 exabyte growth would be near the long-term range, Tan said: “We sort of see the current quarter as being sort of the floor in terms of exabyte growth.” Decode it against what the company has said is coming. The 40-terabyte ePMR platform is expected to be “close to 30% of our nearline shipments by the second quarter and 50% by the third quarter of fiscal 2027.” In August the same platform was targeted at more than 50% of nearline bits by the third quarter. The target is unchanged; the second-quarter waypoint is new. First-generation HAMR finishes qualification “with 2 hyperscale customers in the first quarter of calendar 2027” and begins volume shipments in the first half of calendar 2027, with two more customers starting qualification in the same half. A second-generation HAMR drive at 5 terabytes per platter and 11 disks starts qualification in the second half of calendar 2027.

All of that is a density roadmap, and the standing rule about density growth is that it is elective at the format level and factory-paced at the drive level. The September dive noted that UltraSMR, the shingled format that adds about 20% capacity only if a customer chooses to run it, was the half of the supply plan with a customer-adoption dependency inside it. UltraSMR did not come up on this call at all. HAMR did, at length, and HAMR is the better half: drive-level density needs no customer permission; it needs factory time. Every exabyte of acceleration Tan is promising therefore runs through heads and media that Western Digital makes itself. Tan said so: “all the high-cap leading-edge platforms that we introduce are all on our own internal heads.” And then, on whether adding head capacity adds drives: “that does not translate into more heads and therefore, more units.”

So “the floor” is a statement about the calendar, not about demand. Exabytes grow when the 40-terabyte platform reaches 30% and then 50% of nearline shipments and when HAMR ships in volume. On the company’s own waypoints the first of those lands in the December quarter. That is why the December quarter is the right test. If the floor is the floor, December exabytes print above 22%, and on the trajectory management described, at or above the 25% it says the business will exceed.

UBS’s Timothy Arcuri asked whether investors should expect something like 35% exabyte growth six quarters out, the gap to the rival maker closed. Sennesael’s answer was that the company is “only guiding one quarter at a time.” That is a fair answer and an informative one. A company with seven customers it says it can see to 2030 and 2031 chose not to put a number on volume growth beyond the current quarter. It did put a number on price: “for the next 3, 4 quarters, there’s definitely further room to expand our ASP per terabyte year-over-year growth.” One ingredient got a four-quarter outlook. The other got a quarter.

Capex doubled; the unit count did not

The September dive’s other finding was that Western Digital was not building: fiscal 2026 capex of $418M against $412M, 1.12 times depreciation, construction-in-process down 14.10% to $457M, and facility projects cancelled and impaired inside the year. One quarter of capex proves nothing until the series agrees, and the September quarter added a point to that series that runs the other way.

Capital expenditure was $227M, the largest quarter in the five-quarter series, 2.10 times June’s $108M and 2.18 times the quarter’s $104M of depreciation and amortization. Construction-in-process turned up, to $486M from $457M, a rise of 6.35%. Machinery and equipment rose $114M gross. The 10-Q restates the policy: “We expect capital expenditures to average 4% to 6% of revenue over the next five years; however, we anticipate capital expenditures as a percentage of revenue will be modestly above that range over the next two years as we invest in heads and media technology and factory automation.” At 5.18% of revenue the quarter sat inside that range, not above it, which suggests the larger quarters are still ahead.

Tan on capacity. On the call Tan said the company remains “committed to not investing in any incremental unit capacity.” Decode it alongside the capex line. These two statements are consistent, and together they say something precise. Western Digital is spending more, and it is spending it on the parts of the drive that set terabytes per unit (heads, media) and on automation that sets cost per unit, not on the assembly capacity that sets units. That is the HAMR maker’s capital posture: Seagate, which this site wrote up in September, runs capex at 2.1 times depreciation with construction-in-process rising for the same reason. Price does the work the capex hasn’t done yet was the September description of Western Digital. The correction this quarter is that the capex has started; it is the density capex, and it will take the HAMR calendar to show up in exabytes.

What has not changed is the asset base relative to the earnings it produces. Gross property, plant and equipment was $9,628M against accumulated depreciation of $7,042M: 73.14% written off, against 73.81% in July. A business that earned $2,153M of non-GAAP operating income in a quarter on $2,586M of net property is not capital-constrained in any financial sense. It is constrained by the time a new head and media generation takes to qualify, which is the constraint that sets the exabyte line.

The customers changed under the concentration number

The September dive counted three customers at 44% of revenue and the top ten at 73%. The 10-Q now reports the top ten at 64% for the quarter, against 73% a year ago, with two customers at 14% and 13%, against four at 16%, 13%, 13% and 12% a year earlier. That looks like a large diversification in one year, and some of it is. Read the footnote first: “Beginning in the three months ended October 2, 2026, we updated our methodology for identifying customers for purposes of revenue concentration disclosure,” attributing sales to “the ultimate end customer when products are purchased on its behalf by distributors, contract manufacturers, or other resellers.” Prior periods were revised on the same basis, so the 64% against 73% is like for like. The 73% the September dive quoted was on the old basis, and the two are not comparable.

On the like-for-like basis the diversification is real and management described where it is coming from. Tan named frontier AI labs, neocloud providers and hybrid cloud OEMs, and told Cantor’s C.J. Muse: “we definitely see, I would say, a diversification of our customer base.” The geography agrees. Americas revenue rose 89.18% to $2,081M, 47.50% of the total against 39.03% a year ago; Asia rose 45.36% and EMEA 9.02%. The 10-Q attributes the shift to “the deployment and build plans of our larger Cloud customers.”

This matters to the price argument in one specific way. A diversifying customer base is a customer base with more buyers bidding for the same fixed number of drive units. That is a reason the price ingredient can keep rising, and it is the strongest version of management’s case. It is also the reason the price ingredient is the one to watch: new buyers without long-term agreements pay spot, and Sennesael confirmed that the uncontracted exabytes are seeing “a very strong price environment.” Spot is the part of the book that reprices fastest in both directions.

Where the cash went

Free cash flow was $1,377M, a 31.43% margin, against 34.19% in June; operating cash flow was 0.90 times net income, held down by the $269M inventory build and a $206M rise in receivables. Days sales outstanding fell to 46.4 from 49.2. The quality of the earnings is still not in question: nothing was pulled forward and nothing was financed by a customer.

The financing side spent more than the business generated. Repurchases took $1,001M, dividends $56M, settlement of convertible notes $570M, and taxes paid on vested employee stock $317M: $1,944M in all, $567M more than free cash flow. The gap was funded by $300M of net borrowing on the revolver and a $265M fall in cash. The revolver, which matures in January 2027, went from $350M to $650M after $900M of draws and $600M of repayments. The convertible notes are nearly gone: $141M remained at quarter end, of $710M in July, after $191M was exchanged and $378M was settled on conversion, and the company issued a notice on September 14 to redeem whatever is left on November 16. Net of $1,314M of cash and $789M of debt, the company holds $525M of net cash.

The buyback, priced. The release calls the capital return a statement of “our confidence in the business and our commitment to return capital to shareholders.” Decode it with the 10-Q’s repurchase table. The company bought 0.7M shares in July at an average of $526.29, 0.5M in August at $488.15 and 0.8M in September at $461.59, about 2.0M shares under the program. At $354.06 those shares are worth 32.73%, 27.47% and 23.30% less than was paid for them. That is the ordinary risk of buying back stock in a cyclical business at the top of its margin range, and it is not a scandal. It is a reminder of who carries the price-cycle risk when a manufacturer returns 76.76% of its free cash flow at peak margins: the remaining shareholders. The dividend rose 17% to $0.175 a quarter, $0.70 a year, a 0.198% yield at today’s price. The authorization has $2.26B left.

The guide and the December bar

The guide: revenue of $5.1B ± $100M, non-GAAP gross margin of 60% to 61%, operating expenses of $400M to $410M, interest and other of about $15M, a 17% tax rate and non-GAAP EPS of $5.80 ± $0.20 on about 380M diluted shares. At the midpoint that is revenue up 69.04% on the December 2025 quarter’s $3,017M, non-GAAP gross profit of about $3,085.5M, and non-GAAP operating income of about $2,680.5M, a 52.56% margin.

Now compute what the two ingredients have to do. The December 2025 quarter shipped 215 exabytes, as reported in summaries of that call. If exabytes grow 25%, the floor-plus case, December ships 268.75 exabytes and the guide needs revenue per exabyte of $18.98M, up 7.42% sequentially from $17.67M; year on year, price per terabyte has to rise about 35.2%. If exabytes grow 22% again, December ships 262.3 exabytes and revenue per exabyte has to reach $19.44M, up 10.07% sequentially, with price up about 38.6% year on year. If price per exabyte were frozen at September’s level, the guide would need 288.7 exabytes, 34.28% more than a year earlier and 16.41% more than September. Nobody, including management, has suggested that.

Every one of those paths has price per terabyte growing faster in December than the 28% of September. The guide is a price guide. The volume assumption inside it is not disclosed, but the sensitivity is: each percentage point of exabyte growth below 25% has to be made up by roughly a point and a quarter of additional price. The gross margin guide carries the same assumption on the cost side. At 60.5% on $5.1B, non-GAAP cost of revenue is about $2,014.5M. On 268.75 exabytes that is $7.50M per exabyte; on 262.3 it is $7.68M. September’s non-GAAP figure was $7.41M. Either way the guide assumes cost per exabyte flat to slightly up, consistent with the input-cost warning and the component build.

That is the December bar, stated with its assumptions. Revenue inside $5.0B to $5.2B with exabytes at or above 268.75 means volume is finally carrying its share and Tan’s floor was the floor. Revenue inside the range with exabytes below 268.75 means price did the work a third time in a row, at a rate above 35%.

How peers read this

Seagate reports its September quarter on October 27, and this site froze a test for it in September: fewer than 228 exabytes confirms a price-led September for Seagate too; 237 or more refutes it. The two makers reached the same price-led composition from opposite capital postures, Seagate building HAMR capacity with rising construction-in-process and Western Digital, until this quarter, not building. Seagate exited June with HAMR at about 40% of its nearline exabyte run rate, against Western Digital’s first HAMR volume in the first half of calendar 2027. If Seagate prints exabytes at the top of its range next week, the volume gap between the two widens for at least two more quarters, and Western Digital’s price ingredient is doing work that its rival’s volume ingredient does for it.

Several analysts on the call asked about a third maker adding capacity. Ben Reitzes of Melius asked whether a smaller competitor could move from about 10% to 30% of capacity share. Tan: “I can’t comment about somebody else doing it,” and, on how Western Digital would respond to a rival adding unit capacity, “our response is going to be the same as we’ve communicated over the last year.” That response is no new units. In a market where both large makers are holding units flat and pricing up, the only supply-side risk to price is a third party willing to build. The question was asked three different ways on one call. That is the market telling you where it thinks the price ingredient breaks.

For memory, the read runs the other way. Sennesael’s input-cost comment is a small, clean data point for the DRAM and NAND cycle: a hard-drive maker, which buys those parts as minor components, says their cost is now large enough to slow its cost-down. The 52.63% rise in raw materials and components inventory is that comment turned into a balance-sheet line. Hard-drive makers are not large memory buyers, and none of this moves a memory maker’s numbers. It does say the component inflation the server makers described this autumn has reached the parts of the data center that use memory incidentally.

For the hyperscalers, the read is in the price line. Cloud revenue rose 54% on more than 20% more exabytes and about 25% more price per exabyte. The buyers of nearline storage are paying a quarter more per terabyte than a year ago, with top accounts under contract and new buyers paying spot. Who ultimately pays for the capex is usually a question about who funds the GPUs. Here is a small piece of the answer for storage: the cloud buyers are, through price, and the drive maker is returning it to its own shareholders.

The strongest case against this read

Here is the bull case at full strength. Tan said that for the company’s top seven customers it has “good visibility up to 2030, 2031 for across all of them.” Long-term agreements carry base and upside volumes and pricing with periodic reviews. A 40-terabyte platform is in the fastest ePMR transition the company has run; HAMR qualification is on schedule with no feedback “that would give us any reason for pause”; the customer base is broadening to labs, neoclouds and OEMs; and management has now said in writing and on the call that exabyte growth will exceed 25%. On this reading the decomposition this site has run for two quarters is describing the gap between two product generations. Price ran ahead of volume because volume was between platforms, the December quarter is when the 40-terabyte drive reaches 30% of nearline shipments, and by the March quarter exabytes are growing in the high 20s while price is still growing in the 20s. A company growing exabytes at 25% and price at 25% grows revenue at about 56% with an incremental margin above 80%, and at 15.26 times annualized December EPS that is not expensive. The September dive said the test was one line; the line came in as feared, and the stock is down 19.79% since. A bear who was right on the quarter may be wrong on the year.

That case has the strongest numbers on this page behind it, and it may be right. Here is why we still land where we land. First, the visibility Tan described is visibility of demand, and demand has never been the variable in doubt. Sennesael told Bernstein’s Mark Newman how much of the price is fixed: “The further you go out, the less of it is covered under LTAs.” A commitment is an allocation claim before it is a cost, and what is being committed here is volume; the pricing is reviewed periodically and, beyond the near quarters, is not contracted. Second, the cost line has stopped falling and the company has said why. If price per terabyte decelerates in calendar 2027 the way Sennesael said it would — he expects the growth to “eventually… slow down a little bit” — the margin has no cost-down underneath it to hold the level. Third, the transitions that are supposed to lift exabytes are the same transitions that consumed the August explanation and the October one. A floor is a claim about the next print, and the next print is ten weeks away. We would rather grade it than assume it.

Valuation reality check

At $354.06 on 374.32M shares outstanding as reported by the market-data vendor, Western Digital’s market value is about $132.53B. With $1,314M of cash and $789M of debt, enterprise value is about $132.01B.

On trailing figures: non-GAAP EPS for the four quarters to September was $13.05 ($2.13, $2.72, $3.56 and $4.64), a multiple of 27.13 times. In September, on the June quarter’s trailing figures, the multiple was 43.2 times; the compression came from both directions, a lower price and a fast-rising denominator. Trailing free cash flow was $4,289M ($653M, $978M, $1,281M and $1,377M), a multiple of 30.90 times and a yield of 3.24%. Trailing revenue was $14,482M, 9.15 times.

On run-rate figures the stock looks cheap. September’s $4.64 annualized is $18.56, a multiple of 19.08 times; the December guide’s $5.80 annualized is $23.20, a multiple of 15.26 times. December’s implied non-GAAP operating income of about $2,680.5M annualized puts enterprise value at 12.31 times. Western Digital reports one segment, so there is no segment split to value; the end-market split is 88% Cloud, and the Cloud line is the business.

Which basis is honest depends entirely on the ingredient argument. If exabytes compound at 25% and price holds, the run-rate multiples are the right ones and 15 times forward earnings for a business with a 49% operating margin is low. If price is the cycle and it mean-reverts, the annualized December quarter is the peak quarter of a price cycle, and peak-quarter multiples on a cyclical are the classic trap: the multiple looks lowest exactly when the earnings are least repeatable. The trailing multiple of 27.13 times sits between the two and is the one we would anchor on until the December exabyte line comes in.

What we’d watch from here

• December-quarter exabytes against 268.75. That is the call: below it. At or above it, the floor was the floor and we are wrong.
• Revenue per exabyte against $18.98M. The guide needs at least that at 25% volume growth. A print at or above $19.44M with exabytes near 22% is the third price-led quarter.
• Cost of revenue per exabyte against $7.45M. A rise above $7.68M means the input-cost warning is in the income statement, not only in the inventory.
• Raw materials and components against $435M. A further large build says the company still expects component prices to rise; a fall says it bought ahead and is now consuming the position.
• Price per terabyte growth against 28%. Management gave it a three-to-four-quarter runway. The first quarter it decelerates is the first quarter the margin needs volume.
• HAMR qualification with two hyperscale customers in the first calendar quarter of 2027. A slip moves the exabyte acceleration with it.
• Seagate’s exabytes on October 27 against 228 and 237. The rival’s volume is the best available read on whether the industry’s exabyte growth is supply-paced.
• Top-ten concentration against 64%. On the new basis, further diversification supports the spot-price argument; a reversal would say the new buyers were a one-quarter event.

The trade

Evidence-led, not a chase. The reaction day says the market priced the quarter as fully known: a 6.85% revenue beat, a guide 69.04% above a year ago, and a flat stock after a 19.79% decline since September. Analyst framing, not advice: a holder who believes the floor argument is being paid at 15.26 times annualized December EPS to wait for the December exabyte line, and the downside if it misses is that the price cycle gets marked as a price cycle, which is what the last seven weeks have already partly done. A holder who believes the ingredient argument has a cleaner expression than the shares. The December print is binary on one disclosed metric, and defined-risk optionality into late January expresses a view on that metric without underwriting the level of the price cycle. For anyone pairing the two drive makers, Seagate’s print next week comes first and sets which of the two is carrying volume.

The risks to the call are in the company’s own waypoints. If the 40-terabyte platform reaches 30% of nearline shipments in December as stated and UltraSMR adoption at the third large customer completes in the same quarter, exabytes can clear 25% without HAMR, and we are wrong. A customer pulling in drives ahead of further price increases would do the same, for a quarter. Those are the conditions under which the call fails, and we would say so in the January grade.

Bottom line

Western Digital reported the best quarter in its post-separation history, and every reported line is excellent: revenue of $4,381M, up 55%; a 57.8% gross margin; $1,377M of free cash flow; net cash of $525M. It also reported exabyte growth of 22% for the second straight quarter, against price per terabyte up 28%, on a cost per exabyte that has not moved since March. The September test asked which ingredient would carry the September quarter, and the answer was price. Management now says this quarter is the floor for volume and has guided a December that needs price to rise faster still. We think price carries December too. The late-January print settles it, on one number: exabytes against 268.75.