Record Revenue, Flat Profit Dollars: A Tariff Refund Carries Both the Beat and the Raise, While Personal Systems Adds $1.8 Billion of Revenue and No Operating Income

Coverage Continuity

Our last published note on HP was the Q1 FY26 recap on February 25, 2026, where we maintained Underperform with a $20-$26 fair value range. We did not publish on Q2 FY26. The Q2 comparatives used throughout this note come from the comparative columns of the current quarter's earnings exhibit rather than from any prior work of ours.

That Q1 note set out, explicitly, what would move us off Underperform. Two of the six conditions have now been met or nearly met. FY26 EPS guidance has landed above the $3.15 threshold we named, and AI PC penetration has reached 46% of shipment mix against the 50%-by-year-end test we set. Four have not: there is still no permanent chief executive, the postponed Investor Day has not been rescheduled, memory costs have not stabilized, and buyback has not accelerated to the $1B-per-quarter pace we flagged.

We also got two things materially wrong. We modeled FY26 revenue at $55.0-$55.5B; nine months are already at $44.52B and the year will finish near $60B. We modeled FY26 free cash flow at $2.7-$2.8B; the company now guides $3.0-$3.2B. Pricing pass-through drove far more revenue, and working capital discipline drove far more cash, than our framework allowed. The margin call was right. The earnings and cash call was not, and the stock has moved against us accordingly.

Results vs. Consensus

Q3 FY26 Scorecard

Consensus for this print was unusually scattered. Independent providers carried non-GAAP EPS estimates of $0.66 and $0.69 and revenue estimates ranging from $14.21B to $14.60B. We use $14.43B for revenue, the figure two independent sources agree on within a hundredth of a billion, and show the EPS estimate as a range because the disagreement is material relative to a guidance band only ten cents wide.

Year-over-Year Comparison

Sequential Comparison

The sequential table is the more honest one. Revenue rose 8.8% quarter over quarter and gross profit fell 2.4%. Every profitability line moved the wrong way against a quarter that itself carried a large restructuring charge. Gross margin gave up 215 basis points in three months, and non-GAAP operating margin gave up 95, in a quarter that included a tariff windfall.

Revenue

An 8.6% revenue beat is a large number for a company of this size, and it deserves to be taken seriously before it is discounted. Growth was broad by geography, with constant-currency revenue up 22% in APJ, 10% in EMEA and 5% in the Americas. It was HP's ninth consecutive quarter of top-line growth and the tenth for Personal Systems. Management's stated growth areas, which span AI PCs, advanced compute, hybrid systems and Workforce Solutions, grew 46% collectively.

The problem is what sits underneath. Personal Systems delivered 18% revenue growth on 16% fewer units. That combination implies an average selling price increase of roughly 41%, and management declined to decompose it. Some of the increase is genuine mix, with premium share up 2.6 points sequentially and workstation share up 1.8 points. The rest is memory cost passed to the customer. Revenue built that way is not durable: when component costs normalize, the pricing reverses, and the unit base destroyed on the way up does not automatically return.

Margins

Gross margin of 18.8% is the weakest of the three quarters of fiscal 2026, against 20.9% in Q2 and 19.6% in Q1, and the compression accelerated: down 168 basis points year over year, down 215 sequentially. Management attributed it to higher commodity costs and to a heavier Personal Systems mix, which is arithmetically correct since PS now represents 75% of revenue and carries roughly a quarter of Printing's operating margin.

Operating expense discipline is the offset and it is genuine. Operating costs fell 3.8% in absolute dollars and 223 basis points as a share of revenue, with R&D down 4.2%. That is how a 168-basis-point gross margin decline becomes only a 63-basis-point non-GAAP operating margin decline. It is also a lever with a floor. R&D at 2.5% of revenue is thin for a company positioning itself as an edge AI platform, and the year-over-year decline in absolute R&D dollars sits awkwardly beside the strategic ambition described on the call.

EPS

The distance between the two EPS figures is the story of this quarter. Reported non-GAAP EPS grew 10.7%. Excluding the tariff refunds that management itself quantified, it fell 4.0%. HP grew revenue by $1.75 billion year over year and produced less underlying profit per share than it did a year ago.

GAAP EPS fell 11.3% to $0.71, though that comparison is unhelpful in the other direction: the prior-year quarter carried a $139M tax benefit and a litigation gain, both flagged in the release. The cleanest read is non-GAAP operating profit, which rose 2.6% to $1,021M on 12.5% more revenue, and which itself includes the tariff benefit.

Segment Performance

Segment revenues sum exactly to $15,679M; a $(2)M "Other" reconciling item brings the total to the $15,677M reported on the income statement. Effective in Q1 FY26 HP moved its Print-as-a-Service business from Corporate Investments into Printing, and restated prior periods on an as-if basis, so all comparatives above are on the current basis.

Personal Systems

Record third-quarter revenue of $11.77B, up 18%, with commercial up 22% and consumer up 10%. Commercial now represents 73% of segment revenue, up from 71% a year ago, though down from 76% in Q2 as consumer rebounded 29% sequentially. Premium share rose 2.6 points sequentially and workstation share 1.8 points. Attach businesses, meaning peripherals, collaboration and services including the Workforce Experience Platform, contributed one third of segment gross profit.

Against all of that, segment profit was $537M versus $541M a year ago. Revenue rose $1,836M and segment earnings fell $4M.

"PS operating margin of 4.6% was below our long-term range as expected and down year-over-year from higher commodity costs and variable compensation, which we worked to partially offset with repricing actions and other cost reductions."
— Karen Parkhill, CFO

Two details in that sentence deserve attention. The first is "as expected," which is accurate: management guided PS below the 5-7% range for the remainder of FY26 back at the Q1 print, and has delivered on that unhappy promise. The second is "variable compensation," which is a cost of the revenue beat itself. HP is paying out on a top line that generated no incremental segment profit.

Assessment: This is the single most important number in the release and it is not in the headline. A business that adds $1.8 billion of revenue and no profit is not converting its pricing actions into economics; it is running to stand still. The mix improvements are real and should eventually matter, but at present they are being fully consumed by component costs. Until incremental revenue carries incremental profit, PS revenue growth is a measure of inflation rather than of value creation.

Printing

Revenue of $3.91B declined 2%, or 4% in constant currency, with hardware units down 7%. Supplies, which is 65% of segment revenue and the profit engine, declined 3% and fell 8% sequentially. Reported operating margin of 18.1% sits in the upper half of the 16-19% long-term range and improved 1.1 points year over year.

That improvement is where the tariff refunds landed.

"It does largely benefit Print. There is a smaller portion that benefited PS, but it is largely benefiting Print. And so you saw our Print margins in Q3 be higher than expectations, driven by that benefit. But once you exclude those benefits, they were largely in line with expectations."
— Karen Parkhill, CFO

Management's framing in prepared remarks was that ex-refund Printing operating margin was "in line with our guidance at the low end of our long-term range." The low end is 16%. So the reported 18.1% and the underlying figure are roughly two points apart, and the year-over-year improvement disappears once the refund is removed.

The operational progress in Printing is nonetheless real and worth separating from the margin optics. Tank printer units grew 42% and HP took another 4 points of share in that category, placing hardware that carries profit upfront rather than through a supplies annuity. Industrial Printing posted its twelfth consecutive quarter of revenue growth, and HP signed a three-year, $100 million agreement with a large print services provider. Management reaffirmed the 16-19% range for FY27.

Assessment: Printing is doing what a mature cash business should do, which is decline slowly and pay for everything else. The concern is not this quarter but the composition of the decline: supplies fell faster than hardware, and commercial hardware weakness was concentrated in the North American and Chinese office markets. The tank-share gains are the right strategic response, but they trade a high-margin annuity for an upfront margin, and that is a mix headwind that compounds quietly over several years.

Key Topics & Management Commentary

Overall Management Tone: Confident and considerably more assured than at the Q1 print, where the leadership change dominated and the memory framework was being revised in real time. Management now speaks from a position of having called its own margin trajectory correctly for three quarters and returned repeatedly to that record, most directly in the claim that the company did what it said it would do. The confidence thinned in exactly one place: every question that reached fiscal 2027 economics was answered with directional language and a deferral to the Q4 call, and the one analyst who pressed for the specific mechanics of margin recovery received a list of levers rather than a bridge.

1. The Tariff Refund and What the Beat Actually Contains

Tariff refunds contributed $0.11 to both GAAP and non-GAAP EPS in the quarter, with $0.19 embedded in the raised FY26 outlook and $0.08 assumed in the Q4 guide. Management disclosed all three figures clearly in the release, and the CFO addressed the ex-refund performance directly in prepared remarks rather than waiting to be asked.

"Our net earnings per share at $0.83 grew 11% and include $0.11 related to tariff refunds. Importantly, without the tariff benefit, we still delivered EPS above the top end of our guidance range."
— Karen Parkhill, CFO

The claim is accurate and the margin is one cent: $0.72 against a $0.71 ceiling. It is also worth noting that the same adjustment, applied to the year-ago quarter, converts an 11% gain into a 4% decline.

Assessment: Credit where due, HP volunteered the ex-refund comparison instead of hiding behind the reported number, which is better disclosure than many companies offer on one-time benefits. But an operating beat of one cent on a ten-cent guidance band is a rounding outcome, not evidence that the mitigation program is outrunning the cost curve. The refunds are non-recurring by construction, and the FY27 base does not include them.

2. Memory Costs and the End of the Cheap-Inventory Benefit

Through the first half of FY26, HP was consuming inventory purchased before the memory price spike, which cushioned reported margins. That cushion is now gone, and management said so plainly.

"Yes, we were benefiting earlier in the year from that lower cost of inventory that was on our balance sheet and flowing through to the P&L. While we still have a mix of lower cost of inventory, I would say that benefit is largely behind us. And we now have higher cost of inventory that's working through our P&L, as I said."
— Karen Parkhill, CFO

This explains the shape of the year precisely. Gross margin was 20.9% in Q2 and 18.8% in Q3, a 215-basis-point step down in a single quarter with revenue up 8.8%. The Q4 guide embeds a further decline. Management expects input costs to keep rising in FY27, though at a slower rate than in FY26.

Assessment: The Q1 note treated the memory cycle as the binding bear point, and it remains so. What has changed is the mechanism: the risk is no longer that costs surprise to the upside, since management has calibrated that well for three consecutive quarters. The risk is that the cost curve stays elevated long enough for the price increases to permanently reset the unit base lower. HP is managing the P&L through this competently. It cannot manage the demand response.

3. Unit Destruction and the Elasticity Problem

Personal Systems units fell 16%, with consumer down 19% and commercial down 14%. Printing hardware units fell 7%. Management guided the industry PC unit market down high teens for the second half of the calendar year and Q4 revenue below normal seasonality.

"By segment, in Personal Systems, we remain aligned with industry experts projecting the PC unit TAM to decline high teens year-over-year for the second half of the calendar year. Given the impact of commodity-driven price increases, we expect below seasonal revenue performance in Q4."
— Karen Parkhill, CFO

The Head of Personal Systems went further, naming deferral explicitly as one of two demand vectors shaping fiscal 2027: "there has been demand delay or as some of the customers deferred product refresh given the price increases this year."

Assessment: This is the first quarter in which HP has conceded, on the record, that its own pricing is suppressing demand. The concession cuts both ways. It is intellectually honest and it validates the framework management has been describing, but it also means the revenue growth the market is being asked to reward carries a demand cost that shows up later, in a deferred refresh cohort competing with the AI PC upgrade cycle for the same budget.

4. Pricing as the Last Lever, and the Channel Lag

Management was pressed on whether pricing has caught up to costs or whether more increases are coming. The answer placed pricing at the end of the mitigation sequence, not the start.

"On pricing, we have been increasing pricing. We expect to continue to increase pricing as input costs rise. We use that as a last lever after we have focused on demand shaping and product reconfiguration and taking costs out everywhere where that we can. We use pricing as the last lever."
— Karen Parkhill, CFO

The Head of Personal Systems added the mechanical detail that matters for modeling: HP sells through online channels, traditional channel partners, enterprise customers and contractual end-user deals, and "the lag can range from immediately to a few months before price changes will be fully reflected for all the customers."

Assessment: The lag structure means reported margin troughs after cost peaks, which is consistent with management's own Q4-trough framing. It also means the contractual book is the last to reprice and the last to normalize downward, so a genuine memory cost decline would show up in HP's margins with the same several-month delay, in the favourable direction. That is the mechanical basis for the FY27 recovery case, and it is more credible than the qualitative version management gave when challenged directly.

5. AI PC Mix: The Pillar Compounding on Schedule

AI PCs reached 46% of shipment mix, against a 40-50% FY26 forecast, with management reiterating the 50% year-end target and extending the trajectory well beyond it.

"AI PCs were a strong performance also for us this quarter. It contributed 46% of our mix, in line with our 40% to 50% forecast for FY '26. And we are expecting it to get up to 60% to 70% in 2027 and more than 70% in 2028."
— Ketan Patel, Head of Personal Systems

The economics were also framed with some precision. Attach rates on AI PCs run "very similar" to the rest of the PC business, around a third, and management characterized the category as margin accretive. HP has more than 150 independent software vendor partners working on the platform. The Windows 11 refresh, which carried the last two years, is roughly 70% complete.

Assessment: This pillar is intact and tracking, and it is the strongest argument for owning HP through the memory cycle. The caution is that AI PC mix has risen from above 35% at Q1 to 46% at Q3 while Personal Systems operating margin has fallen from 5.0% to 4.6% over the same span. Premium mix is supposed to be margin accretive and is currently being fully absorbed by component costs. The pillar will only prove itself when the commodity headwind abates and mix shows through to segment margin.

6. Q4 as the Declared Trough

Management reaffirmed a position it first took at the Q2 call, that the October quarter is the low point for Personal Systems margins.

"As we look ahead in Q4 at our margins, I would say that -- we said that we expected Q4 margins to be a low point last quarterly call. We continue to believe that. But we have high confidence that it will improve from there."
— Karen Parkhill, CFO

The Q4 guide is consistent with that framing and with nothing better. Non-GAAP EPS of $0.69-$0.79 includes $0.08 of assumed tariff refunds, so the ex-refund band is $0.61-$0.71, identical to the range HP guided for Q3 three months ago. The midpoint of $0.74 is 10.8% below the $0.83 just reported and 20.4% below the $0.93 delivered in Q4 FY25.

Assessment: A declared trough is a testable claim and we get the first read on it in November. The credibility case is that management has now called this trajectory correctly three quarters running. The caution is that the trough was originally expected earlier, and the FY27 recovery depends on cost increases decelerating, which is a supplier decision rather than an HP one.

7. What FY27 Margin Recovery Actually Requires

Management named four mechanisms for returning Personal Systems margin to the 5-7% range: input cost increases moderating, long-term contract repricing landing, product redesign and platform optimization reaching the cost line, and continued premium and attach mix. Only the last is fully within HP's control and it is already running hard.

"So on input costs, we said we still expect them to rise in FY '27 and in Q4, but at a slower rate than we've seen to date. So our focus, as I said, is just going to be to recover PS margins back to the long-term range as quickly as possible while also continuing to drive profitable growth."
— Karen Parkhill, CFO

Note the construction. Costs still rise in FY27; the improvement comes from the second derivative. Note also "as quickly as possible," which is an intention rather than a timeline, and which management declined to convert into one on two separate occasions during the call.

Assessment: The bridge is plausible in direction and unquantified in magnitude. A 40-basis-point recovery in Personal Systems margin on roughly $44B of annual segment revenue is worth about $176M pre-tax, or roughly fifteen cents of EPS after tax on the current share count. On a $3.00-$3.10 base that is a 5% swing, and it is most of the difference between a stock that works from here and one that does not. Management has the levers and has earned some benefit of the doubt on execution. It has not yet shown the arithmetic.

8. Free Cash Flow and Who Is Financing It

Free cash flow of $1.57B in the quarter and $2.52B over nine months, against $1.44B in the prior-year nine months, is the strongest operational result in the release and drove the raised FY26 outlook of $3.0-$3.2B. Management attributed it to Personal Systems growth and working capital discipline, and pushed back on the suggestion that tariffs explained it.

"Our cash conversion cycle is negative 37 days, and we continue to have a strong focus on that. So it was driven by, yes, more than the tariffs. Tariffs helped a little bit but clearly driven by underlying performance."
— Karen Parkhill, CFO

The balance sheet shows where the cash came from. Over nine months, accounts payable supplied $3,323M while receivables consumed $1,531M and inventory consumed $1,902M. Payables funded the entire working capital build to within $110M. Payables now stand at $21,383M and 151 days, against inventory of $10,322M at 73 days and receivables of $7,168M at 41 days.

Assessment: The cash generation is genuine and management is right that it is not a tariff artifact. It is, however, a supplier-financed model operating at an unusually stretched setting. A negative 37-day cash conversion cycle is an enviable structural position and also one that cannot extend indefinitely. Inventory dollars are up 21% since the October year-end while days held flat, which correctly identifies cost inflation rather than overstocking as the driver, but it does mean HP is carrying $1.8B more working capital at exactly the point where units are falling.

9. Capital Return and the Balance Sheet

HP returned $574M in the quarter through a $274M dividend and $300M of buyback covering approximately 12.2 million shares, and has returned $1,550M over nine months. It paid down slightly more than $500M of debt maturities in the quarter, net of new issuance. Diluted share count fell 2.8% year over year. Gross cash ended at $4.2B and stockholders' deficit narrowed to $(92)M from $(346)M at the October year-end.

Management reiterated the framework of returning approximately 100% of free cash flow over time, conditioned on gross leverage remaining under 2x and the absence of better opportunities, and said the company ended the quarter within its target leverage range.

Assessment: Steady rather than notable. At a $300M quarterly pace, buyback retires roughly 4% of the share count annually at the current price, which nets to the 2.8% year-over-year reduction actually reported once stock compensation is issued back. A meaningful but not transformative EPS lever. The Q1 note flagged an acceleration toward $1B per quarter as one condition that would change our view; that has not happened, and with FY26 free cash flow guided to $3.0-$3.2B against roughly $1.1B of annual dividends, the capacity exists without being used. Being back inside the target leverage range removes the excuse.

10. The Leadership Vacuum, Six Months In

Enrique Lores departed at the Q1 print after a 36-year career. Bruce Broussard remains interim chief executive. The search update on this call contained no timeline.

"Before I do, I know some of you are interested in an update on the CEO search. I'm sure you can appreciate that I am unable to share details or a time line, but the search is proceeding well, and we continue to make good progress on finding the right next leader for HP."
— Bruce Broussard, Interim CEO

The interim chief executive did offer four observations from his months running the company: advancing the operating infrastructure, making devices AI leading, accelerating toward integrated solutions, and providing greater strategic clarity on where to play and how to allocate capital. The fourth is the most revealing, since it is an admission that such clarity is currently absent.

Assessment: Two full quarters of interim leadership with no announced timeline is longer than it should take, and the Investor Day postponed from April has still not been rescheduled. The practical consequence showed up repeatedly in the Q&A: nobody on this call could commit to fiscal 2027 economics, and it is difficult to see how they could before a permanent chief executive owns the plan. That is a real discount factor, and it is the main reason we stop at Hold rather than going further.

11. Edge AI as the Strategic Frame

The prepared remarks devoted substantial airtime to positioning HP as the trusted edge AI platform, spanning the ZGX Fury desk-side system, Z Boost GPU sharing, the Workforce Experience Platform with Poly Lens integration, the OmniBook Ultra and OmniDesk Mini, and a scaled internal deployment with OpenAI. Management framed the customer logic as token economics, latency, security and data governance rather than as a hardware upgrade.

"For customers already seeing the cost of cloud-based AI add up, the economics alone makes Edge AI compelling. It can also improve security, latency and strengthen data governance."
— Bruce Broussard, Interim CEO

Asked whether customers are quantifying returns or simply defaulting to AI-capable specifications in refresh cycles, management pointed to specific deployments in engineering workflows, customer service, manufacturing quality inspection, retail and healthcare, describing adoption as developing in stages.

Assessment: The strategic frame is coherent and better articulated than in prior quarters, and the token-cost argument is a genuine reason a corporate buyer would specify a more expensive device. What is missing is any financial disclosure that would let an investor size it. There is no edge AI revenue line, no attach-rate disclosure specific to the category, and no margin quantification beyond the assertion that the mix is accretive. Until there is, this remains a narrative that supports pricing rather than a segment that can be valued.

Guidance & Outlook

The headline raise is $0.29 at the midpoint, and $0.19 of it is tariff refunds. Removing them leaves a range of $3.00-$3.10 against a prior range of $2.90-$3.10. The floor rose ten cents and the ceiling did not move. That is a narrowing of downside risk earned by three quarters of delivered execution, which is worth something, and it is a materially smaller event than the headline implies.

The more useful comparison is against what HP actually earned last year. FY25 non-GAAP diluted EPS was $3.12. The raised FY26 guide of $3.19-$3.29 is above that, by 2% to 5%. The ex-refund range of $3.00-$3.10 is below it, by 1% to 4%. So on the measure that carries into the FY27 base, HP is on track to earn slightly less this year than last, on roughly $5B more revenue. Two further items flatter the FY26 non-GAAP figure against the FY25 one: the tariff refunds, and a definitional change disclosed in this release under which CEO transition costs became an excluded item effective Q2 FY26.

A structural oddity worth flagging for modeling purposes: the Q4 GAAP EPS guide of $0.74-$0.84 sits above the non-GAAP guide of $0.69-$0.79, because the non-GAAP adjustments for the quarter net to $(0.05). Nearly all of the $0.67 of FY26 exclusions has already been recognized in the first nine months, where the GAAP-to-non-GAAP gap was $0.71. Q4 therefore carries a net non-GAAP credit rather than a charge, driven by retirement-related credits and tax adjustments exceeding a much-reduced restructuring line.

Implied Q4 ramp: Nine-month non-GAAP EPS is $2.50, so the FY26 range of $3.19-$3.29 implies exactly the $0.69-$0.79 Q4 guide with no residual. Revenue math is looser, since HP does not guide revenue, but Printing at normal seasonality and Personal Systems growing year over year while below seasonal implies roughly $15.5-$16.0B, taking FY26 to approximately $60.0-$60.5B against $55.3B in FY25.

Guidance style: Conservative and consistently beaten. HP has now delivered EPS at or above the top of its guidance range for three consecutive quarters, and the CFO said so directly. The pattern argues for treating the Q4 band as a floor rather than a midpoint expectation. It does not argue for extrapolating the tariff refunds, which management has quantified and bounded at $0.19 for the year.

Where the Street sits: The pre-print consensus of $0.66-$0.69 for the quarter was below the guidance midpoint, which is itself unusual and suggests the sell-side was modeling a guidance miss on memory costs. That expectation was wrong, which explains the size of the reported beat. Post-print, the live question is not FY26, which is now three quarters closed and effectively guided, but the FY27 base that starts from $3.00-$3.10 with no tariff refunds and an unresolved margin bridge.

Analyst Q&A Highlights

Whether the Units-Down, Revenue-Up Pattern Persists Into Fiscal 2027

The call opened on the question that defines the investment case: is a quarter of 18% revenue growth on 16% fewer units a transitional artifact or the new operating pattern? Management confirmed the shape for Q4 and explicitly refused to extend it a quarter further, deferring all fiscal 2027 framing to the November call.

Q: "But I'd love to understand, do you see that fiscal Q4 trend line of units being down high teens, but revenue still growing by a few points, persisting through fiscal '27? Or is that more in Q4?"
— Amit Daryanani, Evercore ISI

A: "On Q4, our outlook reflects the industry's view that PC units will decline high teens in the second half as pricing actions pressure demand. As we said, though, we do still expect to drive year-over-year revenue growth. We expect it to be down quarter-over-quarter, but growing year-over-year with a richer mix of higher-value categories, including premium and commercial and consumer PCs, AI PCs, workstations and attach offerings, along with pricing and all of that to more than offset the lower units. I would also note that we're maintaining our discipline to prioritizing profitable growth and edge AI-driven demand rather than chasing low-margin share. Our comments on revenue growth relate to Q4. It's too early for us to be giving FY '27 guidance."
— Karen Parkhill, CFO

Assessment: The answer is complete for Q4 and deliberately closed for FY27, and the closing sentence was volunteered rather than extracted. Management is unwilling to commit to the durability of the pattern, which is the honest position given that it depends on memory pricing. Investors should read the refusal as information: the company does not yet know whether it can hold revenue when pricing stops rising.

Whether the Low-Cost Inventory Cushion Is Exhausted and Where Pricing Stands

A recurring line of questioning sought to separate the memory cost headwind that has already flowed through the P&L from the portion still to come, and to establish whether HP's price increases have caught up to its cost increases or still have ground to make up. The answer settled the first question and left the second deliberately open.

Q: "On the timing of memory cost increases, is the inventory -- the lower-cost memory inventory benefit done? Or in the Q3 that you just printed, is there still some cost benefit? I'm just trying to figure out on the cost line, are there more headwinds for you to normalize to the market price of memory?"
— Mark Newman, Bernstein

A: "While we still have a mix of lower cost of inventory, I would say that benefit is largely behind us. And we now have higher cost of inventory that's working through our P&L, as I said. But on the cost side, we have other things that we are doing to help drive costs down. Things like product redesign and platform optimization, those things take some time, and we'll begin to start seeing the benefit of some of those things going forward."
— Karen Parkhill, CFO

Assessment: This is the most useful disclosure on the call for modeling purposes. It confirms that the reported Q3 margin still contained some cheap-inventory benefit and that Q4 will contain less, which is the mechanical reason the guide steps down. It also confirms that the cost-reduction offsets are the slowest-acting of the four mitigation levers, which pushes the recovery into the back half of FY27 rather than the front.

The Direct Challenge to the Fiscal 2027 Margin Bridge

The sharpest exchange of the call laid out the bear case in full and asked management to answer it: units falling, component inflation continuing, higher-cost inventory flowing through, elasticity worsening, and a peer in the same supply chain guiding to margin pressure from the same cause. The response listed levers without sizing any of them.

Q: "But if we think about unit declines accelerating, component inflation is continuing, you're working through higher cost inventory, and it's fair to probably imagine demand elasticity likely increases due to these factors. It's not totally clear to me what the factors are that allow Personal Systems operating margins to improve. Even NVIDIA tonight is guiding to margin pressure next year from memory costs. So can you maybe just help me please better understand the offsets to some of these pressures that we're thinking about?"
— Erik Woodring, Morgan Stanley

A: "So obviously, we talked about the fact that we expect costs to continue to increase, but at a slower rate. So I'll start there. And at the same time, the mitigation actions that we've been taking are really starting to kick in. So things like long-term contract revisions on the pricing side, which take some lead time and other cost actions that we've been working through like product redesign and platform optimization that can also take some time starting to kick in. And I would say, importantly, we're also going to continue to focus on what we've been driving so far, which is an increased mix of premium products with AI PCs and workstations and also a greater mix of attach offerings where we've got plenty of opportunity to drive even more."
— Karen Parkhill, CFO

Assessment: The question was quantitative and the answer was not. Every lever named is real and each has been named before; none was sized, sequenced or dated. The most important words are the first, because the entire bridge rests on the rate of cost increase decelerating, which is a memory supplier decision. This exchange is the strongest argument for waiting until the November call, when a fiscal 2027 framework is promised, before underwriting the recovery.

Where the Tariff Refunds Actually Landed

A clarifying question late in the call produced the most consequential disclosure of the hour: the segment attribution of the tariff benefit, which management had not provided in prepared remarks and which materially changes how the Printing margin should be read.

Q: "Can you help us walk through your comments in greater detail about protecting print margins? Does that exclude the benefit of the tariff contract count in '26? ... And if you could help us understand, I jumped on late and I apologize, was the tariff refund largely in Print? I would imagine there's a big chunk in PC also, but I didn't quite hear that in the prepared remarks."
— David Vogt, UBS

A: "When we talk about our long-term 16% to 19% range, that is without or excluding any one-time benefit that we might get from tariff refunds. We did have some benefit from tariff refunds this quarter and have signaled more next quarter. It does largely benefit Print. There is a smaller portion that benefited PS, but it is largely benefiting Print. And so you saw our Print margins in Q3 be higher than expectations, driven by that benefit."
— Karen Parkhill, CFO

Assessment: Printing's reported 18.1% operating margin and its 1.1-point year-over-year improvement are substantially a tariff artifact, and management confirmed it only when asked. The underlying figure sits near the 16% floor of the long-term range. This also means the Personal Systems margin of 4.6% is close to clean, which makes it worse rather than better: the segment carrying the revenue growth received almost none of the windfall and still could not hold its profit dollars flat.

Decomposing the Average Selling Price Increase

With revenue up 18% and units down 16%, the implied pricing move is the single largest input to any fiscal 2027 model, and management was asked to rank its components. It declined to quantify but did disclose the sequential share gains and the attach contribution.

Q: "I was wondering if you could help quantify or otherwise rank how like-for-like price increases, portfolio mix shift benefits and then the increased attach of related services benefited that implied ASP increase in the quarter? And if 1/3 of gross profit in this segment coming from those attached businesses is still the right way to think about the mix?"
— Katherine Murphy, Goldman Sachs

A: "I would say all of the things that we are doing really drove our revenue growth, and we're not going to quantify how much was related to each. But clearly, mix played a role as we drive more premium shares, more AI PCs, more hybrid, more workforce experience platform, all of those things played a role along with pricing. We've been continuing to increase pricing given our higher input costs, and that also played a role, too."
— Karen Parkhill, CFO

Assessment: An explicit refusal, and the most important one on the call. The distinction between price and mix determines whether the revenue base survives memory normalization: mix is durable, cost pass-through is not. Management supplied the supporting detail it was comfortable with, including premium share up 2.6 points and workstation share up 1.8 points sequentially, and withheld the ratio itself. That withholding is a reasonable competitive decision and it leaves the central modeling question unanswerable from disclosure.

AI PC Configuration and Attach Economics

Questioning probed whether the AI PC mix figure is being flattered by high-end workstations and what the category actually earns, since a mix statistic without attach economics says little about margin. The answer separated the two and put attach rates at parity with the rest of the portfolio.

Q: "Bruce, like previously during your prepared remarks, you talked about a 50% AI PC mix target by year-end. I'm just kind of curious like in terms of the configurations of those AI PCs, does that include a lot of AI workstations? And what's sort of the attach rate of discrete AI accelerators within that AI PC mix?"
— Steven, for Krish Sankar, TD Cowen

A: "Well, thank you for the question. And really is oriented to the AI PCs and in totality, not the workstation. So just to provide that context relative to the attach rate, the attach rate on the AI PCs are very similar to the attach rate that we see in other parts of the PC business, and that's traditionally around 1/3 of 30% margin kind of opportunity for us."
— Bruce Broussard, Interim CEO

Assessment: A clean answer that removes a plausible objection. The 46% figure is not inflated by workstations, so the mix statistic is comparable across quarters. The attach economics being merely at parity with the rest of the portfolio, rather than superior, tempers the accretion argument somewhat: AI PCs help through price point rather than through a structurally richer attach.

Channel Inventory in an Inflationary Cost Environment

With inventory dollars up sharply and units falling, channel health is the natural concern, since a revenue beat built on sell-in rather than sell-through would eventually reverse. Management addressed both segments and pre-empted the obvious inference from the higher inventory balance.

Q: "I was wondering if you could talk about channel inventory levels in both PC and Print and how they're shaking out maybe relative to where you would ideally like them?"
— Wamsi Mohan, Bank of America

A: "And in an inflationary cost environment, we would expect inventory to be higher than normal, and that's exactly what we see in PS right now. But that said, the inventory levels in the channel remain well controlled. They're within our demand outlook, and they continue to support the supply continuity that we really want in a constrained memory environment. And in Print, our channel inventory remains at healthy levels and within the range that we would consider normal."
— Karen Parkhill, CFO

Assessment: The answer is reasonable and consistent with the balance sheet, where inventory days held flat sequentially at 73 even as dollars rose, which is the signature of cost inflation rather than accumulation. The acknowledgment that Personal Systems channel inventory is above normal is nonetheless a watch item: if the deferred-refresh cohort persists into Q4, above-normal channel stock in a falling unit market becomes a discounting risk precisely when margins are already at their trough.

What They're NOT Saying

1. The price-versus-mix split in the Personal Systems ASP increase. Asked directly to rank like-for-like pricing, portfolio mix and attach, management declined to quantify any of the three. This is the number that determines whether 18% revenue growth survives memory normalization, and it is the one number withheld.
2. Any fiscal 2027 guidance, on two separate attempts. Both the revenue trajectory question and the unit volume question were deferred to the November call, with management noting it is still in its planning period. A company two quarters into an interim chief executive is not in a position to commit to a plan, which is itself the disclosure.
3. A CEO search timeline. The only update offered was that management is "unable to share details or a time line". Six months of interim leadership without a stated timeline is longer than the market expected at the Q1 print.
4. The Investor Day. Postponed from April 23, 2026 and not mentioned once on this call. The forum where a fiscal 2027 and 2028 framework would be articulated remains unscheduled.
5. Where Personal Systems margin troughs, in numbers. Asked point-blank where PS margins would bottom, management confirmed Q4 as the low point and would not name a level. "Below Q3" is the entirety of the disclosure, and Q3 was 4.6%.
6. The size of the tariff refunds in dollars. The benefit is disclosed as $0.11 per share for the quarter and $0.19 for the year, but never as a dollar amount, never split between the segments beyond "largely benefit Print," and never broken into the portion already received versus still expected.
7. Any sizing of the edge AI opportunity. Extensive strategic airtime, several named products, one named customer agreement, and no revenue line, attach rate or margin figure attached to any of it.
8. Why research and development spending fell. R&D declined 4.2% year over year in absolute dollars, to 2.5% of revenue, in the same quarter management described an ambitious edge AI platform strategy. The decline was neither mentioned nor explained.

Market Reaction

• Pre-print setup: HPQ closed at $30.52 on August 26, 2026, up 37.0% year to date against the S&P 500's 12.1%, up 13.0% over trailing twelve months and up 15.0% over the trailing thirty days. The stock entered the print near the top of a 52-week closing range of $18.20 to $31.30.
• Reaction session: The shares gapped down 8.7% to open at $27.86 on August 27, traded a range of $26.73 to $29.73, and closed at $29.63, down 2.9% or $0.89, on a day the S&P 500 rose 0.7%.
• Volume: 40.8 million shares against a 30-day average of 13.9 million, or 2.9 times normal.
• Intraday shape: The low of $26.73 marked a 12.4% decline from the pre-print close before the stock recovered roughly three quarters of the drawdown into the close.

A stock that beats revenue consensus by 8.6%, beats EPS by roughly 20%, and raises full-year guidance does not normally open down 8.7%. That gap is the market pricing the composition of the beat rather than its size, and it happened fast enough that it must have been driven by the release itself rather than by the call.

The recovery through the session is at least as informative as the gap. Buyers stepped in below $27, roughly nine times the ex-refund FY26 guide and a 4.4% dividend yield, and pushed the stock back to within 3% of where it started. That is the signature of a valuation floor rather than a change of mind about the quarter. The market did not decide the print was better than it first appeared; it decided the price had already discounted enough.

The setup matters for interpreting all of this. HP came into the print up 37% year to date, having tripled the market's return, on a thesis that pricing power would offset memory costs. The quarter confirmed the pricing power and disconfirmed the offset. With the shares already reflecting the good half of that trade, a beat of this composition had very little room to work with.

Street Perspective

Debate: Is the revenue growth demand or inflation?

Bull view: The bull case being made on the Street holds that the mix shift is genuine and durable. Commercial is now 73% of Personal Systems revenue, premium and workstation share rose sequentially, AI PCs are 46% of shipments heading toward 50%, and attach businesses generate a third of segment gross profit. On this reading, HP is deliberately trading low-margin units for high-value ones, and the unit decline is a feature.

Bear view: The bear camp contends that a 16% unit decline against an 18% revenue gain is cost pass-through wearing mix as a disguise, and that management's refusal to split the two is the tell. When memory normalizes, price comes out of the model and the destroyed unit base does not return, leaving a smaller company with a better mix and a worse absolute earnings stream.

Our take: The bears have the better of this on the current evidence, though not decisively. If mix were the dominant driver, it would be visible in segment margin, and Personal Systems margin fell 88 basis points year over year while mix improved on every metric management disclosed. That is only consistent with cost pass-through being the larger force. The honest position is that both are operating and that the ratio is unknowable from disclosure, which is precisely why the refusal to quantify was the most important moment of the call.

Debate: Does the fiscal 2027 margin recovery bridge hold?

Bull view: The optimistic case rests on management's forecasting record. It called the margin trajectory correctly for three consecutive quarters, delivered EPS at or above the top of guidance each time, and has now declared Q4 the trough with high confidence. Four mitigation levers are in flight, the slowest of them are just beginning to reach the cost line, and input cost increases are expected to decelerate in fiscal 2027.

Bear view: Skeptics note that the bridge depends on a second derivative that HP does not control. Costs still rise in fiscal 2027 on management's own guidance; only the rate of increase improves. Pricing is described as the last lever and is already deployed. The contractual book reprices with a lag in both directions. And the company would not size the recovery when asked directly.

Our take: Split. The direction is credible and the timing is not underwritten. Management has earned the benefit of the doubt on execution and has not earned it on quantification, largely because it cannot: the input is a supplier pricing decision. The recovery is worth roughly fifteen cents of EPS at a forty-basis-point Personal Systems improvement, which is most of the gap between a stock that works from $29.63 and one that does not. That asymmetry argues for waiting for the November framework rather than paying for the bridge now.

Debate: Is roughly nine times earnings with a 4% yield cheap enough?

Bull view: The value case is straightforward. At $29.63 the shares trade at 9.6 to 9.9 times the ex-refund FY26 guide, carry a 4.05% dividend yield, and offer an 11.0% to 11.7% free cash flow yield on the raised $3.0-$3.2B outlook. Roughly 100% of free cash flow returns to shareholders over time, leverage is inside the target range, and the share count is shrinking near 3% annually. Cyclically depressed margins on a trough multiple is the classic setup.

Bear view: The skeptical case is that the multiple is not the constraint, the earnings base is. Free cash flow is payables-financed at 151 days, a structural position that cannot extend much further. The stock has already risen 37% this year, so the cyclical rerating has substantially happened. And a company running on an interim chief executive with no timeline, no Investor Day and no fiscal 2027 framework is not a company an investor can underwrite at any multiple.

Our take: The valuation is fair rather than cheap, which is the substance of our move to Hold. Nine to ten times a trough earnings base with a well-covered 4% yield limits downside meaningfully, and we no longer see the path to the $20-$26 range we were carrying. But the free cash flow yield overstates the case, because the working capital contribution behind it does not repeat at this magnitude, and the governance discount is real while the search remains open. Fair value $27-$32, with the stock inside it.

Model Update Needed

Valuation impact: Fair value range raised to $27-$32 from $20-$26. The low end assumes Personal Systems margin stays below 5% through FY27 and units continue contracting, putting normalized EPS near $3.10 at roughly 8.7 times. The high end assumes the declared Q4 trough holds and Personal Systems recovers to the low end of its range during FY27, supporting roughly $3.40 at 9.4 times. At $29.63 the shares sit near the middle of that range, which is the definition of a Hold.

The single largest swing factor is the price-versus-mix composition of the Personal Systems ASP increase, which management declined to disclose. If it is predominantly mix, the FY27 revenue base holds near $62B and the margin recovery arrives with operating leverage. If it is predominantly cost pass-through, FY27 revenue compresses as memory normalizes and the recovery in rate is offset by a shrinking base.

Thesis Scorecard Post-Earnings

Overall: The thesis is unchanged in its operating substance and broken in its price conclusion. Every margin and volume concern we carried into this print was confirmed, several of them more emphatically than at Q1. What failed was the inference we drew from them: we expected the margin compression to produce falling earnings and a falling stock, and instead pricing power protected the earnings dollars and the shares rerated 37%. A bear case that is right about the mechanism and wrong about the outcome is not a bear case worth holding.

Action: Upgrading to Hold from Underperform, with fair value raised to $27-$32 from $20-$26. We are not moving further because the underlying earnings are flat to lower, Personal Systems is generating no incremental profit on record revenue, Q4 is guided down, and the company has no permanent chief executive and no fiscal 2027 framework. The November print is the decision point: it brings the FY27 outlook, the first test of the declared Q4 trough, and, plausibly, a named chief executive. An upgrade to Outperform requires two of the three to land well. A return to Underperform requires the Q4 trough to fail.