Built by Grok 4.6 High

Western Digital Corporation reported fiscal Q1 FY2027 on Thursday, October 22, 2026. Analysis prepared Monday, October 26, 2026. The quarter ended in early October 2026 (calendar 2026Q3). The fiscal year ends in late June. Flash left with the SanDisk separation on February 21, 2025, so this is an HDD continuing-operations print. FY2026 was a 53-week year and FY2026Q1 had 14 weeks, so the +55.5% compare is against a longer quarter, not a short one.

WDC | Earnings Review — FY2027 Q1

The growth rate turned back up, and the turn is price. Continuing-ops revenue was $4,381 million, +55.5% versus $2,818 million in FY2026Q1. That is +1,163 bps faster than FY2026Q4's +43.8%. The path into the trough was +76.2%, +30.9%, +30.0%, +27.4%, +25.2% (FY2026Q2). The up-leg is +45.5%, +43.8%, +55.5%. Three prints, one pause, not a one-quarter bounce. Non-GAAP diluted EPS was $4.64, +160.7% versus $1.78. On a clean HDD base that rate has accelerated for three straight quarters: +100.0%, +114.5%, +160.7%.

Bits did not do it. Exabytes shipped were 248, +21.6% versus 204, the same year-over-year rate as FY2026Q4. Nearline, 223 of those exabytes, was +21.9%. Calculated price per terabyte (revenue divided by exabytes) was +27.9% year over year, after +2.5%, +8.8%, and +18.3%. Four straight accelerations in price. Management's blended figure on the call was +28%, versus high teens in the prior quarter. That +22% and +28% multiply to about +56% and describe the quarter just printed. The calculated +21.6% and +27.9% multiply to the +55.5% print. December at +69% still needs a higher bit rate, a higher price rate, or both. Cloud is the book: $3,868 million, +54.1%, 88% of revenue and about 87% of the dollar increase. Client $320 million grew +119% and is 7% of sales. Consumer $193 million slowed to +19.1%.

Margins are still widening, and one spoken sentence is wrong. Non-GAAP gross margin was 58.1%, +1,420 bps versus 43.9%. The prepared remarks say "up 34 basis points." That figure is not usable. A +34 bp change cannot produce the incremental gross margin management also claimed, which the filing math puts at about 84% and the call called "above 80%." Non-GAAP operating margin was 49.1%, +1,870 bps versus 30.4%. Non-GAAP operating income was $2,153 million, +151.5%. GAAP diluted EPS was $4.63. The one-cent gap to non-GAAP means this beat is not an add-back. Through FY2026, GAAP EPS was the SanDisk stake. This is the first quarter in the window where the two meet.

The beat reversed a shrinking surprise, and it cleared their own high. Versus FMP, revenue was +$237 million / +5.7% and non-GAAP EPS was +$0.53 / +12.9%. Both also cleared the August guide high: revenue $4.1 billion ± $100 million, EPS $4.00 ± $0.15, gross margin 55%–56%. Actuals were about +$181 million, +$0.49, and +210 bps past those highs. FMP was already at the midpoint, so this is not a beat of a sandbagged estimate. Post-spin revenue surprises had compressed from +5.4% to +1.3%. This quarter re-widened them. Non-GAAP EPS is 12 of 12 over the last twelve quarters. Pattern: consistent beater. The mechanism is price and mix, not a volume surprise and not cost-down. Cost per terabyte is a headwind: the long-run target is still about a 10% decline, and near term it will be less than that because DRAM and NAND inside the drive are inflating.

December is priced. The path after it is not. The new guide, the only guide they give, is revenue $5.1 billion ± $100 million, gross margin 60%–61%, and EPS $5.80 ± $0.20. At the midpoint that is +69% revenue and +172% EPS versus FY2026Q2. Post-print FMP (October 26) is $5,127 million and $5.87, inside both bands. There is no full-year guide. FMP's FY2027 year is $21.05 billion, about +62.9% versus FY2026 revenue of $12,919 million, and the quarterly path fades from +69.9% to about +67% and then +63% by FY2027Q4. Management is repeating two things that path only half-accepts: this +22% exabyte print is the floor, and price per terabyte still has room for three to four quarters and then slows. That is the disagreement. It is not "the street is low on December."

What else changed, and what was not answered. Tone versus August is the same framework, said harder, with three new boundaries: price growth is not the steady state, cost-down misses the 10% target near term, and capex runs modestly above the 4%–6% five-year band for two years. The January call had said the HAMR ramp would stay inside that band. Seventeen of twenty-three street questions were answered. Ackerman asked once and had no follow-up. The six deflections are the weights a buyer cannot see: neo-cloud and physical-AI size, how wide the supply gap is, whether out-year hyperscale price talks have moved, and the contract terms for the newer customer set. Five document conflicts will corrupt a model if the wrong sentence is used. Use +1,420 bps, not +34. HAMR volume is calendar first half 2027, not fiscal first half 2027. 44TB is first-generation HAMR, not the 40TB ePMR ramp that is supposed to be about 30% of nearline bits this quarter and about 50% in FY2027Q3.

HDD only. Pre-spin consolidated figures that still include flash are not in this table. Dollar and margin cells link to the current figure. A year-over-year percent is calculated from the linked year-ago level. Exabytes are the shipment count management stated. Daloopa's unit field says Million. The value is exabytes, not millions of exabytes. Price per terabyte is calculated, not a Daloopa fundamental.

Non-GAAP EPS before FY2026Q2 is not a clean HDD year-over-year rate. The prior-year print in that series is still the consolidated company, except the restated FY2025Q2 base of $1.20. FY2025Q2 in the EPS row above is that restatement, not the $1.77 the street graded. Operating-margin percents before FY2026Q1 are calculated from the cited operating income and revenue. Derived EBITDA is GAAP operating income plus depreciation and amortization, not the old adjusted-EBITDA series. This quarter it foots: GAAP operating income $2,046 million plus D&A $104 million equals $2,150 million. FY2025Q1 continuing-ops revenue, outside this eight-quarter window, was $2,212 million, +85.3% versus $1,194 million. That is the rate the FY2025Q2 deceleration is measured against.

End markets, above the total

Cloud carries the company. Client is the growth-rate spike and is still too small to be the story. Consumer re-accelerated and then gave part of it back.

The segment price bridge Daloopa tags for this quarter is not the blended +28%. Cloud price impact was +25%, client +90%, and consumer +70%. Do not average those into the blended rate. They do explain why client revenue can grow +119% while client exabytes are not a separate surge. Two quarters ago the tagged company bridge was the other way around: exabyte impact +34% and price-per-exabyte impact +9% in FY2026Q3, versus +22% and +2% in FY2026Q2. The mix of the bridge has flipped. Non-GAAP opex was $391 million versus $381 million, +2.6% against +55.5% revenue. Operating leverage is the quiet half of the margin print. Top-two customers were 14% and 13% of revenue, versus 18% and 16% a year ago.

Five completed years

Annual revenue growth re-accelerated into FY2025 and then slowed. The quarterly turn happened inside that slower year and after it. FY2022–FY2024 are the restated HDD view, not the flash-inclusive company. Non-GAAP EPS before FY2025 is left blank because the old per-share series is still consolidated.

FY2026's +35.7% includes a 53rd week, so the slowdown versus FY2025's +50.7% is larger than the headline. Annual gross-margin expansion narrowed from +1,070 bps to +970 bps. Quarterly expansion did the opposite. It got wider. FMP's FY2027 revenue average of $21.05 billion is about +63% versus $12,919 million. FMP's FY2027 EPS average is $24.32 versus $10.22. Neither is a company guide.

Wide, price-led double beat, above the company's own high, after the revenue surprise had been shrinking. FMP is the street. There is no Visible Alpha or Bloomberg gross-margin consensus, so margin is scored against the company guide only. FY2027Q1 is highlighted.

FY2025Q2 revenue and EPS in that table are the pre-spin consolidated print. Do not score the later continuing-ops restatement of $2,409 million and $1.20 against the combined-company estimate. That would invent a miss. FY2025Q3 is the first HDD-only print and the only revenue miss in this eight-quarter window. FY2026Q4 revenue of $3,747 million was about $3 million inside a $3.75 billion guide high. Management called it at or above the high end. EPS still cleared its high.

The four quarters before that window, still flash-inclusive, are part of the twelve-quarter rate and are not an HDD run-rate.

Rates. Revenue versus the street is 9 of 12 over twelve quarters and 4 of 4 over the last four. EPS is 12 of 12 and 4 of 4. Revenue has not printed below the guide low in the six quarters that can be scored, and it cleared the high in five of those six. EPS cleared the high in six of six. The FY2027Q1 gap versus the revenue high, about +$181 million, is the widest of the six.

Magnitude. Post-spin revenue surprises went +5.4%, +3.2%, +3.0%, +2.8%, +1.3%, then +5.7%. Dollar surprises +$133 million, +$88 million, +$88 million, +$91 million, +$47 million, +$237 million. FY2026Q4 was the exhaustion print. This quarter reversed it. It is not a smooth four-quarter improvement. HDD-era EPS surprises in dollars: +$0.24, +$0.18, +$0.19, +$0.20, +$0.33, +$0.25, +$0.53. One dip, then a new high. The percent stays in a high-single-digit to mid-teens band because the base is rising.

What management said caused it. They do not say "versus the street." They say the print was ahead of their own expectations. Three drivers, all structural to this cycle rather than one-time: contractual and spot price, mix onto the up-to-40TB ePMR ramp, and more exabytes from areal density without a matching unit add. Cost per terabyte did not drive the beat. An analyst put the September cost decline at about 4–5%. The CFO did not confirm the figure and did not argue with the direction. No tax gift and no share-count trick. Diluted shares were 383 million versus 369 million a year ago, so EPS growth is not a buyback. The CFO's own duration on the price piece: further room for three to four quarters, then the rate of increase slows.

FMP's stored revenue actuals for the April, July, and October 2024 prints have been overwritten to later HDD restatements. Diffing those against the combined-company estimates still on the file invents misses of roughly 45%. This history uses the original revenue. EPS on that file was not overwritten.

Western Digital does not guide a year. It guides one quarter, non-GAAP, and replaces that guide every print. There was no prior numerical guide for FY2027Q2. Comparing the new $5.1 billion with the old $4.1 billion is two different quarters, not a revision.

The midpoint bridge is a calculation, not a second guide. Gross profit at 60.5% on $5.1 billion is $3,085.5 million. Less $405 million of opex and $15 million of interest, taxed at 17%, on 380 million shares, is $5.82. That is a $0.02 residual versus the $5.80 guide, inside the ±$0.20 band. Operating margin on that bridge is 52.6%. Year-over-year incremental gross margin on that midpoint is about 81%, the same zone as the 80–81% the CFO said was embedded in the quarter just reported. An analyst's "75%" is the sequential increment from FY2027Q1, which is the comparison the CFO has said he does not run. FMP's EBIT field does not reconcile to this non-GAAP bridge and is not used. The FMP revenue range on this quarter is $4,045 million to $5,674 million. The average is inside the guide. The low end is not, so at least some models in the blend were not revised.

Why the midpoint is not the bull case

They have printed at or above the high end of gross margin and EPS on each of the last four guides. The latest revenue beat versus the midpoint, +$281 million / +6.9%, is the largest of the four.

Waterfall — the year-over-year rate, not a same-quarter revision

The step from the quarter just delivered to the new guide is +1,350 bps of revenue growth and about +12 points of EPS growth, from +160.7% to +172%. Gross-margin expansion barely steps up, from +1,420 bps to +1,440 bps. The full-year FMP rate is 610 bps slower than the quarter being guided. FMP's own quarterly path, same pull: FY2027Q3 revenue $5,583 million is about +67% versus $3,337 million, and FY2027Q4 revenue $6,124 million is about +63% versus $3,747 million. H1 on company numbers is $4,381 million plus the $5.1 billion midpoint, about $9.5 billion. The $21.05 billion year leaves a second half in line with those quarterly averages. Nobody is hiding a different year. They are hiding a milder second half than "accelerating exabytes and more price" would produce if both legs keep moving.

What +69% requires. Holding exabytes at +22% and getting revenue to +69% takes price-per-terabyte growth of roughly +39%. If exabytes reaccelerate only to the "greater than 25%" long-term rate, price growth still has to be in the mid-30s. Management would not stamp either figure. They did say both legs move the right way near term, and that the price leg fades after three or four quarters. December is the quarter where that claim has to start showing up in exabytes, not only in price.

What they refused. Full-year revenue, EPS, margin, exabytes, and ASP. "We only guide one quarter at a time." Other forward comments that are not a quarterly number: up-to-40TB ePMR near 30% of nearline shipments in FY2027Q2 and about 50% in FY2027Q3; HAMR qualifications with two hyperscalers in calendar Q1 2027 and volume in calendar first half 2027; dividend raised to $0.175 from $0.15, payable December 17, 2026. Versus the year-ago $0.125 the new rate is +40%. The board's "17%" is versus the prior quarter's rate.

Tone. August was confident and still qualifying the slope. October is the same framework with more boundaries. Results were ahead of expectations. Visibility for the top seven customers runs toward the end of the decade, with the demand signal still rising. A neo-cloud hiccup, they said, is absorbed by the other three demand engines. The new cautious sentences are the price window, the cost-down miss, and capex above the band. A year ago the guide was about +20% revenue and a mid-40s gross margin. The migration to a high-60s growth guide and a low-60s margin guide, with the team beating the top of each range, is four quarters of delivery, not one adjective.

Continuing-ops revenue growth spent four quarters decelerating, from +76.2% in FY2025Q2 to a trough of +25.2% in FY2026Q2. The first step, −4,529 bps into FY2025Q3, was mostly the compare getting harder after a depressed HDD year, not a new downturn. The year-over-year dollar add was still about $540–$610 million a quarter through the trough. The regime changed in FY2026Q3. The rate jumped +2,023 bps to +45.5%, and the dollar add crossed $1 billion. FY2026Q4 was a pause, not a reversal: −163 bps to +43.8%, with the dollar add still rising, from $1,043 million to $1,142 million. FY2027Q1 re-accelerated +1,163 bps to +55.5%, and the dollar add widened again to $1,563 million.

Non-GAAP EPS, once both sides are HDD, is a cleaner acceleration than revenue: +77.5%, +100.0%, +114.5%, +160.7%. The last step is the largest, +4,622 bps. The dollar gap widened too, from +$0.93 to +$1.36 to +$1.90 to +$2.86. The base rose from $1.36 to $1.78 and the rate still rose. This is not a tiny-base trick, and it is not the share count. Gross-margin expansion itself widened from +1,040 bps to +1,310 bps to +1,420 bps. GAAP EPS should not be used for the rate until the SanDisk stake is out of the compare. FY2026Q3 and FY2026Q4 GAAP diluted EPS were $8.20 and $8.21 against non-GAAP of $2.72 and $3.56.

The FY2027Q2 guide midpoint of $5.1 billion versus $3,017 million would be another +1,350 bps of revenue-rate acceleration, to +69%. The EPS guide of $5.80 versus $2.13 is +172%, another step up from +160.7%. FMP is on that guide, not on a different near-term trajectory. The later risk management has already named is price year-over-year rolling off while bit growth is only a low-20s number, unless the exabyte re-acceleration they are describing actually shows up. It has not shown up in the last two prints.

The quarter now underway is FY2027Q2, period ending around January 2, 2027. Consensus has already moved onto the guide. The catalysts that can still change the year are the ones the quarterly model does not contain.

Where they disagree. On December, they do not. The contrarian reading is not a low street number for the quarter being guided. It is whether the exabyte floor and the three-to-four-quarter price window are real, because that is what would stop the year-over-year rate from rolling over in the March and June quarters the way FMP has penciled. Management has earned the right to be believed on a one-quarter gross-margin and EPS beat. They have not yet printed the exabyte re-acceleration they are now calling the base case. The last two exabyte prints went the other way, or rather they went nowhere: +21.6%, then +21.6%.

What would show management is early: a December exabyte rate clearly above +22% and a price-per-terabyte rate still at least the +28% just reported, with gross margin at the top of 60–61%. What would show the street is right to fade the duration: revenue at the guide low on bits stuck at 22% and price no longer accelerating, or gross margin at 60% only because of mix while cost per terabyte is flat or up. The dividend of $0.175 is a cash date inside the quarter, not an earnings catalyst. Free cash flow of $1,377 million is funding both the head-and-media step-up and the higher dividend. A capex ratio that breaks out above "modestly above 4–6%" would be the tripwire. It is not what December EPS assumes.

Twelve speakers, twenty-three questions. Ackerman asked once and had no follow-up. The Bernstein line was taken by Scott, not by Mark Newman. Scored 17 answered and 6 deflected. The clean answers are the price mechanics, the refusal to discount HAMR, and what the extra capex is for. The deflections are one cluster: weights the buyer cannot see.

The six deflections are not six unrelated slips. Irving acknowledged "a lot of fear" on neo-clouds and then declined the weight. A tight market that cannot be sized cannot be distinguished from a slogan. New demand is only contracted demand if duration, price, and deposits exist, and they were not described. The gross-margin trajectory is not one of the deflections. Year-over-year expansion is +1,420 bps to 58.1%, and a "75% incremental" headline is a sequential calculation management did not adopt.

Do not lift speech-to-text as facts. "HAMMER" is HAMR. "Hidden media" is head and media. "Up to 2020, 2030" is planning visibility through 2030. "Up 34 basis points" is not the margin change.

Five conflicts. Two will move a model by quarters or by about 1,400 basis points if the wrong sentence is taken literally. The operating story is otherwise consistent: the revenue, exabyte, segment, and dividend figures match the filing. Labeled guidance updates are not in this list. The January capex sentence was not labeled as a change, so it is.

Checked and not contradictions: end-market growth matches the spoken $3.9 billion, +54%, $320 million, +119%, and $193 million, +19%. Exabytes "248, up 22%" matches 248 versus 204. The December gross-margin range matches 60%–61%. The long-term exabyte frame moving from the mid-teens and low-20s up to "greater than 25%" was raised over several calls and is an updated target, not two descriptions of one unchanged number. C.J. Muse's 62 / 80 / 75 incremental path was an analyst frame the CFO did not adopt. The FMP company description that still cites flash and SanDisk is stale vendor text, not a management statement on this print.

This was not a rates, CPI, tariff, or consumer-confidence call. Nothing was said about the Fed, Treasury yields, household sentiment, retail traffic, PMI, or trade policy. The macro they described is a tight storage market: four demand engines, exabyte supply that is not yet accelerating, and a price per terabyte that is. The only inflation they named is DRAM and NAND inside the drive.

Rates. No rate level. The only financing comments are their own: interest and other expense expected at $15 million next quarter, and a net-cash comment on the call that is not a Daloopa series here. An analyst tried to hang a late-cycle worry on capex. The answer was heads and media, not units into a peak. On this telling, a higher cost of capital is a filter for which neo-clouds get supply, not a reason the storage cycle rolls over.

Input costs. No CPI. DRAM and NAND inside the drive are tight, value engineering is still in qualification, and near-term cost per terabyte will fall by less than the long-run 10%. That is better for unnamed memory suppliers and a real but second-order cost for Western Digital. It has not stopped the margin: non-GAAP gross margin +1,420 bps to 58.1%, with the next quarter guided at 60%–61%. August was explicit that client and consumer price was up because flash-based alternatives had repriced. October did not repeat that. Do not carry "HDD price is up because NAND price is up" forward as an October statement. October's NAND sentence is a cost, not a competing-media read-through. No memory vendor was named.

Not a consumer call. Client revenue +119% to $320 million is mostly price and mix. The client price-bridge impact was +90%. Non-nearline exabytes only went from +10% to +19%. Do not treat +119% as evidence that PC demand doubled. Consumer revenue decelerated, from +37.5% to +19.1%, to $193 million. Geography, same quarter: Americas $2,081 million versus $1,100 million, +89.2%. Asia $1,708 million versus $1,175 million, +45.4%. EMEA $592 million versus $543 million, +9.0%. The growth is the Americas. Asia is not the region carrying the print, which matters for the autonomous-vehicle anecdote.

Companies. Only one operating company was named.

Named last quarter and dropped this quarter, which is silence, not a downgrade: flash price as the umbrella for client and consumer HDD price; enterprise OEMs leaving all-flash arrays; sovereigns and China open-source labs; one large-customer LTA through calendar 2029. October's "top seven" is broader on count and vaguer on contract, and the CFO then cut price duration back.

Do not treat "thinking about it in the right direction" as guidance to 35% exabyte growth. The latest reported exabyte year-over-year rate is +21.6%, flat versus the prior quarter. Do not treat top-seven visibility to 2030 as a priced book. Do not read a flat bit rate as the acceleration they guided the language toward. The acceleration they have already delivered is in price and in the revenue year-over-year rate. Bits are the promise for the next several quarters, tied to qualifications they have not finished.