Apple and Huawei Renewed, Data Center Finally Billing: Upgrading Qualcomm Into the Chip Selloff

Results vs. Consensus

Q4 FY2026 Scorecard

Qualcomm does not publish a consolidated gross margin, so the gross-margin and earnings-before-taxes margin lines above are derived from the filed income statement rather than quoted. Providers measure the revenue beat against different bases, one against the $10.44 billion non-GAAP figure and one against the $10.32 billion GAAP figure, which is why the beat reads as 2.0% or 2.9% depending on the source. Both agree on direction and on a low-single-digit magnitude. The $120 million wedge between the two revenue lines is a payment to a customer arising from a legal settlement, which the company excludes from non-GAAP revenue. The GAAP revenue and handset consensus marks each come from a single provider, so treat those two comparisons as indicative.

Year-Over-Year Comparisons

The prior-year GAAP loss per share of ($2.89) is not a comparable operating result. Fiscal 2025's fourth quarter carried a $5.7 billion charge to establish a valuation allowance on federal deferred tax assets, and fiscal 2026's second quarter released it. The GAAP earnings comparison across these two years is therefore a tax-accounting artefact in both directions, which is why the non-GAAP line carries the analysis.

Quarter-Over-Quarter Comparisons

This is the arithmetic that explains the quarter better than any single headline. Revenue rose 5.0% sequentially on a non-GAAP basis and QCT revenue rose 6.0%, yet earnings per share fell 1.8%. Every dollar of incremental revenue arrived at a worse margin than the dollar before it, and 202 basis points of QCT segment margin went out the door on a 6% revenue increase. Operating leverage ran backwards for the fourth consecutive quarter.

Fiscal Year Comparisons

The sum of the four reported quarters ties to the fiscal-year figure exactly at $43,120 million, and the fiscal 2026 non-GAAP earnings of $10.54 per share land $0.03 above the ~$10.51 we published on August 1. Fiscal 2026 was therefore a year in which Qualcomm delivered what it said it would deliver, and earned 12.4% less for it. Both facts matter for the fiscal 2027 argument.

Performance Against the Company's Own Guide

The GAAP shortfall against its own guided range deserves a sentence, because it is the only line in the quarter that failed a company-set bar. The July guide assumed a zero contribution from strategic investments and $(0.11) per share of other items. The quarter delivered $(0.09) from strategic investments and $(0.20) of other items, a combined $0.18 worse than guided, against non-GAAP earnings $0.02 better than the midpoint. The driver is $235 million of restructuring and restructuring-related charges plus $141 million of acquisition-related charges, neither of them fully contemplated three months ago. Restructuring charges are now a recurring feature: $343 million across fiscal 2026, of which $235 million landed in this quarter alone.

Assessment: Revenue

The revenue line did what the July guide promised and slightly more, and the composition is the story. Qualcomm is now a company whose sequential growth comes from automotive and whose largest segment is shrinking, and this quarter is the first in which that was visible in a single pair of numbers: automotive up 25.3% sequentially and 89.0% year over year, handsets up 1.0% sequentially and down 26.2% year over year. The handset miss against consensus is small in dollars and large in signal, because last quarter management committed to double-digit sequential growth in handset revenue from Chinese OEMs and this quarter delivered a 1.0% increase in the total handset line without disclosing the Android component. Either Android grew strongly and Apple fell faster than the Street modelled, or Android underdelivered. The company chose not to resolve which.

Assessment: Margins

Two things happened to margin this quarter, and only one of them is bad. The bad one is the level: gross margin has now fallen in every quarter of fiscal 2026, from 54.55% to 50.75%, and the fourth-quarter decline of 230 basis points sequentially was the steepest of the four. The good one is the disclosure. The release commits in writing to the full benefit of the pricing actions showing up in QCT gross margins in the second half of fiscal 2027, and the call added that the benefit begins in the second fiscal quarter. For two consecutive quarters we have written that a margin recovery with no date attached is the shape of disclosure that costs this stock money. The date now exists. What came with it is a quieter downgrade of the destination, from the 48% to 50% historical band to the low end of that band, with the custom-ASIC mix removing a further 150 to 200 basis points on top. Blended QCT gross margin, on management's own arithmetic across two calls, does not return to its historical range at all.

Assessment: EPS

Earnings per share of $2.17 beat by a cent and fell 27.7% year over year, and the fiscal-year figure of $10.54 fell 12.4%. The important number is not either of those; it is the first-quarter guide of $2.15 to $2.35 against the $3.50 Qualcomm earned in the first quarter of fiscal 2026, a 36% year-over-year decline at the midpoint. That is the real magnitude of what the Apple product exit plus the input-cost shock is doing to this company's earnings, and it is why the stock has spent a year going nowhere while the business got structurally broader. The offset, which is the whole basis of the upgrade below, is that the first quarter of fiscal 2027 is very likely the bottom, and for the first time Qualcomm has told us when the recovery starts and shown us a data center revenue line to put against the hole.

Segment Performance

Reporting Segments

The segment total does not foot to the consolidated figure because corporate items and the strategic-investments segment sit outside QCT and QTL. Segment EBT margins are derived to the basis point from the filed revenue and EBT lines; the company rounds them to whole percentage points.

QCT Revenue Streams

Non-handset revenue reached 43.0% of QCT in the quarter and 35.3% for the fiscal year. Management guides both past 50% of QCT in fiscal 2027 once data center is included.

Fiscal 2026 Quarterly Trajectory by Stream

Read down the columns rather than across the rows and fiscal 2026 resolves into one picture: automotive compounded sequentially in every quarter and IoT in the last three, the largest segment fell by a third and then stabilised, gross margin fell every time the business reported, and earnings fell in each of the last three quarters. The fourth quarter is the first in which the growth segments were large enough to push total revenue up while margin still fell.

QCT Handsets

Handset revenue of $5,137 million fell 26.2% year over year and rose 1.0% sequentially, the first sequential increase in three quarters. Fiscal 2026 handset revenue of $24,072 million fell 13.4%. Management now plans on global handset units flat to down in calendar 2027, after a low-teens to mid-teens unit decline in calendar 2026, and reiterated that the June quarter was the bottom for China Android.

"Going forward, our plan assumes global handset units will remain flat to down in calendar 2027. The timing and shape of the Apple modem transition is fully incorporated in how we are planning the business and its well understood at this point."
— Cristiano Amon, President and CEO

The Samsung relationship was reframed usefully. Asked whether the 75% share framework still applied to the next Galaxy generation, management moved the conversation from a share estimate to a planning baseline.

"If you think about our business in the past, the baseline was 50%. This baseline is no longer 50%. it is north of 70%, as we have been doing."
— Cristiano Amon, President and CEO

In July the Samsung framing had softened from "greater than 70%," which management had called the framework for this year and next, to "approximately 70% of their flagship devices," and we flagged the softening as a bear data point. It has firmed back up, and management added that it is now winning Samsung wearables content that it did not previously supply.

Assessment: The handset segment has stopped being a forecasting problem and become an arithmetic one, which is an improvement even though the arithmetic is unpleasant. A flat-to-down unit market, a premium Android share that management says will strengthen, and an Apple product line ramping to immaterial gives a handset revenue base somewhere in the low twenty-billions for fiscal 2027 and a floor after that. The segment missed the Street this quarter and the company declined to split Android from Apple, which is the single most useful disclosure it could make and the one it has now withheld for two quarters.

QCT Automotive

Automotive revenue of $1,990 million grew 89.0% year over year and 25.3% sequentially, rose sequentially in all four quarters of fiscal 2026, and took the fiscal year to $6,004 million, up 51.7%. Management headlines it as the first $2 billion automotive quarter, which rounds $1,990 million up. Three months ago the company raised its exit run rate to roughly $7 billion; it exited at approximately $8 billion.

"In automotive, we exit fiscal 26 at an annualized revenue run rate of approximately $8 billion well above the $7 billion expectations we provided last quarter."
— Cristiano Amon, President and CEO

The mechanism is content, not units. Commercial shipments of the fifth-generation Snapdragon digital chassis began in the September quarter, and management put the silicon content increase from the third-generation platform at eight times. The products are also changing shape, from systems-on-chip to integrated compute modules, which captures more of the bill of materials.

"there is a lot more a lot more compute content. Into the automotive and most of our automotive growth is really share where the design win pipeline translates into revenue. that is where we are starting to see step function increases on the SoC revenue and that will continue."
— Cristiano Amon, President and CEO

Assessment: This is the most reliably executed part of the company and the second consecutive quarter in which it has beaten a target management had only just raised. An eight-fold content increase on a platform transition that is one quarter into a multi-year ramp is a durable revenue mechanism rather than a cycle, and the claim that Qualcomm becomes one of the largest automotive semiconductor suppliers by revenue in fiscal 2027 is now credible on the run rate alone. The question put to management, whether the beat reflects pre-buying ahead of price increases, got a content-based answer rather than an order-book answer, which is the right answer but not a verifiable one.

QCT IoT

IoT revenue of $1,888 million grew 4.5% year over year and 3.2% sequentially, and rose sequentially in each of the last three quarters of fiscal 2026. The internal composition is sharply divergent: the industrial, networking and robotics category grew 20% year over year, offset by memory constraints on consumer products.

The robotics line is where the forward optionality sits. The premium IQ-10 platform begins its commercial ramp, and ecosystem partners demonstrated models of up to 80 billion parameters running directly on the device during the quarter.

Assessment: A 4.5% growth rate flatters nothing and conceals a genuine bifurcation, with a 20% industrial business carrying a consumer business that memory pricing has pushed into decline. The segment is the smallest contributor to the fiscal 2027 non-handset inflection and the one most exposed to the input-cost cycle at the low end, since memory is a larger share of the bill of materials in cheap connected devices than in premium handsets. Guidance has automotive and IoT combined roughly flat sequentially in the first quarter, which implies IoT gives back some of its ground while automotive holds.

QTL Licensing

QTL revenue of $1,304 million fell 7.4% year over year and rose 2.0% sequentially, with EBT margin of 70.40%, down 164 basis points year over year but up 146 basis points sequentially. For the fiscal year QTL produced $5,556 million of revenue and $4,022 million of EBT at a 72.39% margin. Guidance for the first quarter is $1.3 billion to $1.5 billion at a 72% to 76% margin, which management attributes to normal seasonality.

Assessment: The financial performance of this segment is the least interesting thing about it this quarter. What matters is that 31.0% of fiscal 2026 non-GAAP earnings before taxes, generated on 12.9% of revenue, was sitting on contracts whose renewal was an open question in every note we have written on this company. That question is now closed, and the section below is where it is argued.

Key Topics & Management Commentary

Overall Management Tone: Management was the most unambiguously forward-leaning it has been in four quarters, and for the first time the posture was earned rather than asserted: the call opened by listing the year's uncertainties as closed rather than as managed. Analyst pushback concentrated almost entirely on whether the handset outlook and the automotive beat reflect pre-buying ahead of announced price increases, and the answers there stayed framework-level rather than quantified. The one visible retreat was on the March quarter, where management declined to repeat the sequential-growth framing it had volunteered in July.

1. The Licensing Overhang Closed, and It Was a Third of the Profit

Qualcomm completed long-term renewals with Apple, Huawei and Transsion. The combination matters more than any one name. Apple was the renewal that the market had been pricing as a live risk; Huawei had gone unmentioned on two consecutive calls, which we flagged in July as increasingly conspicuous; Transsion is the volume leader in the emerging-market tiers where unit growth actually lives.

"We are extremely pleased with our renewals with Apple, Huawei and Transsion. These agreements provide stability for QTL as we move to the 6G era. They also demonstrate the enduring value of our IP and patent portfolio."
— Cristiano Amon, President and CEO

The commercial terms held. Asked directly whether the rate per device survived, management confirmed that the run rate is unchanged because the rate is unchanged, which is a materially stronger statement than a revenue reassurance. The 10-K puts the term structure in writing: key OEM patent licence agreements now expire at varying dates between fiscal 2030 and fiscal 2033.

"we have now completed long term license agreements with Apple, Huawei, and Transsion. And do not have any upcoming license renewals with major handset OEMs for the next several years. Following these agreements, QTL revenue run rate and long term revenue forecast remains unchanged."
— Akash Palkhiwala, CFO and COO

Put the figures against it. QTL generated $4,022 million of earnings before taxes in fiscal 2026, 31.0% of the company's non-GAAP total, on 12.9% of revenue at a 72% margin. That cash stream is now contracted for four to seven years at held rates.

Assessment: This is the single largest change to the investment case since we initiated, and it is being almost entirely ignored because it arrived on the same day as a sector meltdown. A third of the profit, at the highest margin in the company, moved from contractually unresolved to contractually locked through at least fiscal 2030, at unchanged rates, with no major handset renewal in between. The bear argument that survives is that the renewals remove a tail risk without adding growth, and that the average-selling-price cap on the royalty base still blocks pricing-driven upside. That argument is correct and much smaller than the risk it replaces.

2. The Gross-Margin Trough Finally Has a Date, and a Worse Destination

The defining financial problem of fiscal 2026 was an input-cost shock that migrated from handset volumes to gross margin and then sat there without a recovery date for two quarters. The release attaches one in writing.

"The semiconductor industry is experiencing a broad-based increase in input costs across wafer fabrication, assembly, test, advanced packaging, memory and other materials. As previously disclosed, we are taking actions to reflect the higher input costs in our product pricing. We anticipate the full benefit of these actions to be reflected in QCT gross margins in the second half of fiscal 2027."
— Qualcomm fiscal fourth quarter 2026 results release, Business Outlook

The call added the front end of that schedule: the benefit begins to be realised in the second fiscal quarter and reaches full effect in the second half. Against that, the destination moved. Through fiscal 2026 the stated baseline was a 48% to 50% QCT gross margin band; the framework offered this quarter returns the baseline to the low end of that range, and the custom-ASIC mix then removes a further 150 to 200 basis points, a figure management quantified in July and reaffirmed in substance here.

"we expect that to be something that is accretive our operating margins, but dilutive to gross margins. And so that calibrates the gross margin down from the frame. Framework I just outlined."
— Akash Palkhiwala, CFO and COO

Meanwhile the level kept deteriorating. Derived GAAP gross margin of 50.75% is the lowest quarter of fiscal 2026, the fifth consecutive sequential decline, and 459 basis points below the prior-year quarter. Cost of revenues rose 1.0% while revenue fell 8.4%, the same perverse signature as the third quarter.

Assessment: The disclosure we asked for twice has arrived, and it is worth less than we hoped because it comes attached to a lower endpoint. Blended QCT gross margin, on management's own two-call arithmetic, does not return to its historical band: the pricing action restores the low end and the data center mix takes 150 to 200 basis points back off. The thing that has genuinely improved is bounding. An investor can now model a trough in the first half of fiscal 2027, a recovery slope through the second half, and a terminal margin below the old normal. That is a worse business than the one the bulls described in April and a far more investable one than the undated hole we were handed in July.

3. Data Center Converted From a Promise Into an Invoice

For four quarters the data center story has been a sequence of commitments: a target, then an unnamed customer, then two customers with purchase orders and started wafers. This quarter it becomes revenue. First-quarter guidance includes more than $650 million of data center revenue, custom accelerators began shipping in October, and the full-year target was reiterated with an exit rate attached.

"Our data center business is off to a strong start. We are reiterating our $5 billion revenue target for fiscal 2027 with healthy order coverage supporting continued revenue growth in fiscal 28."
— Cristiano Amon, President and CEO

"We expect data center revenues to scale throughout fiscal 27, exiting the year at a $7 billion run rate."
— Akash Palkhiwala, CFO and COO

Amazon Web Services is now named, inside a framework with a disclosed ceiling. The relationship spans custom silicon for AI inference and optical connectivity across multiple product generations, with a potential revenue opportunity of up to $60 billion over ten years. A second optical connectivity customer was secured in the quarter, and a third customer sits on the accelerator side with advanced discussions on subsequent custom silicon.

"Our agreement spans multiple product generations across custom silicon for AI inference and optical connectivity."
— Cristiano Amon, President and CEO

The composition of the $5 billion was disclosed for the first time: mostly custom silicon for the main AI accelerator at the two hyperscalers, with connectivity the next material category. The fiscal 2029 $15 billion target then layers in merchant CPU and accelerator products. And on the upside case, management inverted the usual constraint.

"we expect to be constrained by supply rather than demand for upsides beyond the $5 billion."
— Akash Palkhiwala, CFO and COO

Assessment: The transition from design win to revenue is the hardest step in a semiconductor growth story and Qualcomm has now taken it, on the schedule it set in April. Two things still need saying. The $5 billion is concentrated in two customers and in the lowest-margin product form the company sells, which is why it dilutes gross margin while helping operating margin. And the path from $5 billion to $15 billion runs through merchant products whose only named customer is Meta. The step taken this quarter does not de-risk fiscal 2029; it de-risks fiscal 2027 and 2028, which is where the valuation argument actually lives.

4. The Amazon Warrant Is Contra-Revenue, and Nobody Asked

The warrant issued to Amazon came up on the call only as a question about whether it is a template for other customers. Management said it is not, describing it as "the exception" and noting the warrants were issued "at the money." The 10-K carries the terms, and one of them changes how the data center revenue line should be modelled.

Qualcomm issued a warrant in September 2026 to an affiliate of Amazon Data Services for up to 25 million shares at an exercise price of $161.26, with a ten-year exercise and vesting term. Vesting is tied to commercial arrangements, binding purchase orders and actual purchase milestones up to a maximum of $60 billion of payments. Approximately 4 million shares vested on issuance. Grant-date fair value was $71.57 per share, $1.8 billion in total. The accounting treatment is the point: the warrant shares "are recorded as a reduction to revenues as the related revenues are recognized during the vesting term."

Assessment: Two consequences follow and neither was discussed. The reported data center revenue line will be net of warrant vesting, so the $5 billion fiscal 2027 target and the reported revenue that lands against it are not the same quantity, and a model that treats them as identical will overstate the revenue print. Second, full vesting and exercise represents 2.33% dilution at a strike of $161.26, which is above where the stock closed. The structure is defensible as a way to align a hyperscaler to a multi-generation commitment, and the exercise price is above the current price, which is the right side of the trade for existing holders. It is still a cost of the deal, and it is being carried in a line item most readers of the revenue target will never see.

5. The Segmentation Change That Costs Three Points of QCT Margin

From the first quarter of fiscal 2027 the data center business moves inside the QCT segment. QCT EBT margin is guided to 20% to 22%, against 23.76% reported this quarter, and the guide explicitly includes a three-percentage-point drag from the segmentation change itself.

"And just that change kind of reduces QCT gross margin operating margin by 3%, as, of course, we are investing ahead of revenue in a couple product lines. It does not change the overall Qualcomm numbers."
— Akash Palkhiwala, CFO and COO

Add the three points back and the guide is 23% to 25% on a basis comparable with the 23.76% just reported. The segment margin is roughly flat at the midpoint, not down another two to four points.

Assessment: This is the most mis-readable number in the guide and it will be misread. The headline comparison, 24% falling to 20% to 22%, looks like a further leg of deterioration. On a like-for-like basis it is stabilisation, which is exactly what the pricing-action timetable implies should be happening. Management also left the long-term QCT target at 30%, and tied it to revenue scaling across "the next 2 to 3 years." The honest reading is that segment margin has stopped falling and that the route back to 30% now runs through revenue scale.

6. Apple Is Leaving as a Product Customer and Staying as a Licensee

The 10-K discloses that revenues from Apple and Samsung each exceeded 10% of consolidated revenue in fiscal 2026, and the concentration table shows two customers at 22% and 20%. Since only those two cleared the threshold, Apple and Samsung are those two, in unstated order, at a combined 42% of consolidated revenue.

Two readings follow. The first is that Apple's product exit is not a customer exit, because the licence that just got renewed keeps Apple inside the top-two concentration table regardless of what happens to modem purchases. The second is the row nobody mentioned: a customer or licensee that supplied 13% of revenue in fiscal 2025 fell below the 10% disclosure threshold in fiscal 2026. That disclosure exists only in the annual report.

Assessment: The Apple relationship is bifurcating rather than ending, and the half that is ending is the low-margin half while the half that was just locked to at least fiscal 2030 is the 70%-margin half. That is a materially better outcome than the framing of the past year has implied. The unexplained departure of a 13% customer from the concentration table is the opposite: it is the clearest single piece of evidence for how violent the fiscal 2026 handset reset actually was at the customer level, and it was disclosed in a filing rather than discussed on a call.

7. The Inventory Build Is in Raw Materials, Not Finished Goods

Inventories closed the year at $8,640 million against $6,526 million, up 32.4%, and consumed $2,049 million of operating cash flow. Days of inventory, computed on the quarter's cost of revenues, rose to roughly 155 from roughly 118. Those headline numbers look like a demand problem. The composition says otherwise.

Of the $2,114 million build, $1,786 million or 85% sits in raw materials and work-in-process. Finished goods grew less than 15% while raw materials nearly doubled.

Assessment: This is the signature of a company buying ahead of rising input costs and staging production for a data center ramp, not of unsold product accumulating. In a quarter where the entire bull and bear debate is about whether customers are pre-buying ahead of price increases, it is notable that the clearest evidence of pre-buying in the filings is Qualcomm doing it to its own suppliers. The risk is real and worth watching, because 155 days of inventory in a flat-to-down handset market is a write-down exposure if the mix is wrong. It is a different risk from the one the headline number suggests, and no analyst raised it.

8. Capital Return Keeps Shrinking While the Authorisation Sits Idle

The company returned $1,998 million in the quarter, $1,011 million through repurchases of 6 million shares and $987 million in dividends at $0.92 per share. That is the smallest repurchase quarter of fiscal 2026 and the second consecutive sequential decline, down 58.6% from the $2,443 million repurchased in the prior-year quarter.

Three facts sit badly together. Capital returned reached 134.9% of free cash flow for the year, against 98.3% the year before. Net debt rose $5,098 million, counting the $2,323 million of restricted cash held a year earlier as cash, with $2,492 million of short-term debt appearing where there had been none, including $499 million of commercial paper. And the $7,817 million of repurchases reduced shares outstanding by 2 million, or 0.19%, because 46 million shares were issued, including $4,512 million of stock used to pay for acquisitions. Meanwhile most of the $20 billion authorisation announced with the March-quarter results remained available at year end.

Assessment: Qualcomm spent $7.8 billion on its own stock in fiscal 2026 and finished the year with essentially the same share count, which is the clearest available statement of where the capital actually went: into acquisitions, paid in equity, with the buyback running behind to absorb the issuance. That is a legitimate strategy for building a data center franchise and it is not the same thing as returning capital. With the buyback now at a $1 billion quarterly pace against a $20 billion authorisation that is mostly unused and net debt up five billion dollars, the valuation support that the repurchase used to provide has largely gone. We keep this pillar at risk, and we note that the company has not framed fiscal 2027 capital return at all.

9. Agentic AI as the Handset Content Argument

The strategic answer to a flat-to-down unit market is content per device, and management spent a substantial share of prepared remarks building that case: agentic smartphones launched on Snapdragon during the quarter, Meta Muse passing 10 million downloads in 40 days, and a widening set of intelligent wearables.

"Heading into fiscal 2027, we are pleased with our premium tier Android share position and we expect it to strengthen as the transition to Agentic smartphones drive an increase in computing content per device."
— Cristiano Amon, President and CEO

"For perspective, leading AI companies are targeting smartphones by 2028 capable of running 100 billion parameter models locally with an all day battery life. Reaching this goal will require a step function change in on device compute performance."
— Cristiano Amon, President and CEO

The design-win pipeline across the company grew 70% year over year, which is the only quantified support offered for any of it.

Assessment: The logic is sound and the timing is distant. If on-device inference of hundred-billion-parameter models becomes the premium handset specification by 2028, the company that sells the application processor captures a content increase that does not depend on unit growth, and Qualcomm is better placed for that than anyone except the vertically integrated. None of it shows up in the fiscal 2027 model, and a 70% increase in an undefined design-win pipeline is a direction rather than a number. We treat this as free optionality that justifies nothing in the valuation and should not be written off either.

10. The SpaceX Infrastructure Win Is Real and Mostly a Fiscal 2028 Event

SpaceX selected Qualcomm's Dragonwing cellular infrastructure platforms to power Starlink Mobile's nationwide terrestrial 5G network in the United States. Qualcomm supplies radio silicon including massive MIMO, the accelerator for the digital unit, and commercial-grade layer 1 software, under a multi-year agreement with deployment beginning in 2027 across what both parties describe as one of the largest 5G build-outs in the country.

"Under a multiyear agreement, deployment begins in 2027 and is expected to be 1 of the largest 5G deployments in the United States."
— Cristiano Amon, President and CEO

Asked to size it, management was unusually direct about the timing.

"You should think about from a financial standpoint, it will start to appear as we exit fiscal 27. it is more of a fiscal 2028 from a revenue standpoint."
— Cristiano Amon, President and CEO

Assessment: A greenfield nationwide RAN build is a genuinely new revenue category for Qualcomm and the architectural argument behind it, that an AI-native and open-RAN approach can displace dedicated incumbent equipment, is the same argument that makes the data center business plausible. Management deserves credit for volunteering that the revenue is a fiscal 2028 story rather than letting the announcement inflate the fiscal 2027 number. Nothing for the current model, and a second named anchor customer in a new category within two quarters.

11. The Arm Verdict, and the Part Management Left Out

A jury found that Arm breached its long-standing architecture licence agreement with Qualcomm, and management devoted a paragraph of prepared remarks to the outcome.

"We are gratified that the jury found ARM breached its long standing architecture license agreement ALA with Qualcomm. Failed to honor its commitments. Intentionally acted to disrupt Qualcomm's business and engage in unfair competition both as a supplier and a competitor."
— Cristiano Amon, President and CEO

Not mentioned in the prepared remarks: the remedies phase remains pending. It entered the call only because an analyst raised it, and management's answer was that there are no implications for the CPU roadmap.

Assessment: The verdict matters for roadmap certainty. Management says Qualcomm can continue developing and shipping its Oryon CPUs, which is the outcome that was genuinely at risk in the dispute, and that certainty underpins both the handset content argument and the merchant CPU line inside the fiscal 2029 data center target. The presentation was one-sided. A verdict framed as a victory for innovation and competition still has a remedies phase to run, and the fact that it had to be surfaced by an analyst rather than by the company is the kind of omission worth noticing in a quarter otherwise full of good news.

Guidance & Outlook

The shape of this guide is the opposite of July's. Three months ago the revenue midpoint sat slightly above the Street while the earnings midpoint sat roughly 9% below it, and the entire gap was cost. This time the revenue midpoint sits roughly 3% above consensus and the earnings midpoint roughly 1% to 3% below it, depending on whose consensus mark you use. Providers disagree on the first-quarter earnings estimate, at $2.27 and at $2.32, so the shortfall is either two cents or seven cents. Either way the gap has narrowed at least threefold, which means the cost pressure has stopped getting worse relative to expectations.

Two further items inside the guide are worth separating. The guided share-based compensation drag rises to $(0.99) per share from the $(0.72) actual in the fourth quarter, a 37.5% increase. And the seasonality of the business has formally changed.

"typically in previous years, December has been our strongest quarter in the year. Now with Apple product revenue ramp down, we do not expect December to be the high quarter for us."
— Akash Palkhiwala, CFO and COO

Implied quarterly ramp: Management framed fiscal 2027 with four constraints: non-handset QCT revenue including data center grows more than 60% year over year starting in the first quarter; non-handset exceeds 50% of QCT revenue for the year; data center delivers $5 billion and exits at a $7 billion run rate; and handsets fall below 50% of total revenue. From the $10.8 billion first-quarter midpoint, data center alone adds roughly $1.1 billion of quarterly revenue by the fourth quarter, with automotive and IoT adding perhaps $0.8 billion against modest handset erosion. That path produces something near $10.8 billion, $11.3 billion, $11.9 billion and $12.4 billion, or roughly $46.4 billion for the year. The segment arithmetic implies more: 60% growth on the $13,136 million non-handset base gives $21 billion or better, which at above half of QCT caps handsets near $21 billion and puts total revenue closer to $47 billion to $48 billion. We use $46 billion to $48 billion, or growth of roughly 6% to 11%, and flag that the two routes disagree.

Street at: The guide is above consensus on revenue and marginally below on earnings. The first-quarter earnings midpoint of $2.25 against the $3.50 earned in the first quarter of fiscal 2026 is a 36% year-over-year decline, and that is the number driving the six price-target cuts that followed the print.

Guidance style: Consistent with the pattern. Qualcomm guided the fourth quarter to $9.7 billion to $10.5 billion of revenue and delivered $10,442 million, at the 93rd percentile of its own range, and guided QCT to $8.4 billion to $9.0 billion and delivered $9,015 million, just above the top. Earnings landed at the 60th percentile. This company has beaten the top end of its segment revenue guide and the midpoint of its earnings guide in consecutive quarters, which argues the first-quarter range should be read from the upper half. The exception is GAAP earnings, which came in below the guided range because restructuring charges were not contemplated in the guide.

Analyst Q&A Highlights

Whether Gross Margin Keeps Falling Through the Data Center Ramp

The first question of the call went straight to the central tension: getting to a $7 billion data center exit run rate requires the business to roughly triple through the year, and the first data center dollars are the most gross-margin dilutive the company sells. The answer separated the two forces cleanly for the first time. The pricing actions return the baseline to the low end of the historical range, with the benefit starting in the second fiscal quarter and reaching full effect in the second half, and the data center mix then calibrates down from that baseline. Management offered no blended number and no quarterly path.

Q: "last quarter, you guys kind of telegraphed the gross margin step down that was expected from the ASIC business starting to ramp. How should we think about it through the rest of the year though?"
— Joshua Buchalter, TD Cowen

A: "As you know, we mentioned this last quarter, we are implementing price increases to reflect the higher semiconductor cost across the industry. And so we start realizing the benefit of that in our second fiscal quarter. And we will see the full benefit of that show up in second half of 27."
— Akash Palkhiwala, CFO and COO

Assessment: The timing answer is new and the level answer is a quiet downgrade. An investor leaves this exchange able to model a first-half trough and a second-half recovery, which was impossible after the last two calls, and also knowing that the endpoint is the low end of a band rather than the band itself, with 150 to 200 basis points of mix taken off after that. That is the trade this quarter offered: certainty about when, in exchange for a worse where.

Whether the Royalty Rate Survived the Renewals

The most valuable answer of the call came in response to the shortest question. Management had said repeatedly that the QTL revenue run rate was unchanged after the Apple, Huawei and Transsion renewals, which could be true on volume growth even if rates had been conceded. Asked to confirm the rate itself, the answer closed the ambiguity.

Q: "In terms of the royalty rate, you sort of said that revenue run rate of QTL is unchanged. That the rate per device kind of similar as you look at these new contracts that you signed? Is there any change there?"
— Joseph Moore, Morgan Stanley

A: "We are indicating that the revenue run rate remains the same. Which you should take it as we were able to hold the rate and the royalties as we renewed the contracts."
— Alex Rogers, President, Qualcomm Technology Licensing

Assessment: This is the single most important sentence spoken on the call, and it got the least airtime. Holding rate in a renewal negotiated against a counterparty that has just insourced the modem, in a declining unit market, is a demonstration of patent-portfolio leverage rather than of goodwill. It converts a third of the company's earnings from a negotiation risk into a contracted annuity through at least fiscal 2030.

Whether the Handset Outlook Is Just Pre-Buying Ahead of Price Increases

The sharpest challenge of the call, and the one that most deserved a number. The argument put to management was internally consistent: the handset market is running down more than 20% year over year, the company is simultaneously raising prices and reporting growing China Android revenue, and those two facts together look like customers pulling orders forward. The response came in two parts. The financial answer was that fiscal 2026 suffered from channel inventory drawdown as well as market decline, that channel inventory is now thin, and that the company cross-checks it against licensing data. The strategic answer was that the price increases are being applied to new platforms rather than current ones, which makes them harder to front-run.

Q: "I wanted to dig into your handset market outlook a little bit. I am a little surprised still with the outlook for flat, especially given where we seem to be running at the end of the year, we are probably down more than 20% year over year in truth here, and things do not look better almost all that great in Q4. You also said your own China Android revenues were actually starting to grow in, I guess, June or September and December. You are also doing price increases. Like how do you know in the context of everything that is going on that is just not pre-buys in front of the price increases in Android?"
— Stacy Rasgon, Bernstein Research

A: "And we do not expect that impact in 2027 because we have been following channel inventory. As you know, we get QTL information as well that we can check against. And so the channel inventory continues to be thin."
— Akash Palkhiwala, CFO and COO

Assessment: The channel-inventory cross-check against licensing data is a genuinely privileged information source and the strongest available rebuttal, because Qualcomm sees royalty-bearing device sell-through that its chipset customers cannot easily misrepresent. It is also unverifiable from outside, and the company declined to quantify thin. The pricing-on-new-platforms point is the more durable answer. Note the asymmetry in what the filings show: the clearest evidence of pre-buying this quarter is Qualcomm's own raw-material inventory nearly doubling, not its customers' order patterns.

Whether the March Quarter Is Still Guided Up

The only visible retreat of the call. On the July call management volunteered that revenue would be up slightly sequentially into March, framing a new post-Apple seasonality in which growth runs from December into the rest of the year. Pressed twice to repeat it, with the questioner supplying the supporting logic about easy Apple comparisons and improving Android, management declined to recommit and instead endorsed the framework in the abstract.

Q: "Is that because last call you did sort of say that March would be up. I guess you do not wanna commit that now?"
— Timothy Arcuri, UBS

A: "Yeah. I think it is the reasonable framework that I think the way people have looked at the business is a reasonable way of looking at it."
— Akash Palkhiwala, CFO and COO

Assessment: A non-answer, and the only one in the call. Reading it charitably, the company has just told investors that December is no longer its peak quarter and that revenue scales through fiscal 2027, which makes a March increase the default implication and an explicit guide unnecessary. Reading it less charitably, a management team that will tell you the second-half gross margin schedule, the fourth-quarter data center exit rate and the full-year non-handset growth rate, but will not repeat a one-quarter sequential direction it offered unprompted three months ago, is signalling that the March quarter is tighter than the framework implies. We assume the second reading and model March up only modestly.

How Concentrated the $5 Billion Data Center Target Really Is

A recurring line of questioning across the call pressed on how many customers actually underwrite the fiscal 2027 target, after management referenced a third customer in the slide deck. The answer narrowed it rather than broadening it: two customers carry the $5 billion at significant scale and extend into fiscal 2028 and 2029, with the third sitting on the accelerator side and in advanced discussion on subsequent custom silicon that contributes to the later years. One of the two at scale is Amazon.

Q: "Was wondering if you can maybe just kind of clarify relative to the 3 distinct hyperscale customers that you are ramping through fiscal 27, will all 3 of those be at some kind of revenue materiality exiting the year? Or will it just be maybe 2 of them and 1 will kind of fall outside the year?"
— Jim Schneider, Goldman Sachs

A: "You should think of 2 of those customers at very significant scale, making up that $5 billion, and they will extend into fiscal 28 and then 2029 as well."
— Akash Palkhiwala, CFO and COO

Assessment: Management answered honestly and the honest answer is a concentration risk. Two customers account for essentially all of a $5 billion revenue line that is the entire basis of the fiscal 2027 growth story, and both are hyperscalers whose capital expenditure plans are precisely what the market was repricing on the day this call happened. The mitigant is that these are multi-generation custom silicon programmes with binding purchase orders, which are far stickier than merchant sockets. The exposure is still that a single customer deferral moves the number materially.

Whether the Amazon Warrant Becomes the Template

A short exchange with a larger implication than it received. The question was whether the equity component attached to the Amazon relationship is how Qualcomm intends to win hyperscaler business generally. Management said no: the engagement model across data center is conventional semiconductor commercial practice, the warrant is the exception, and the warrants were issued at the money. Future structures will be assessed deal by deal.

Q: "on the Amazon deal, the warrant structure that you have there, you view that as kind of the model for approaching other customers? Do you expect to have those types of warrant arrangements? Or is that unique to that specific relationship?"
— Joseph Moore, Morgan Stanley

A: "And I think as we consider data center engagements going forward, we will we will look at other structures on a deal by deal basis, but you should think of our framework as being aligned with the traditional semiconductor model."
— Akash Palkhiwala, CFO and COO

Assessment: The right answer, and incomplete. Confining equity sweeteners to the anchor customer protects the economics of every subsequent deal, and an at-the-money strike above the current share price is the least bad version of the structure. What went unsaid, and unasked, is that the warrant is recorded as a reduction to revenue as the related revenue is recognised, which means the revenue target and the reported revenue line are not the same number. That is a modelling consequence of a deal structure management describes as an exception, and it survives in the accounts for ten years.

What the Accelerator Silicon Is Actually Doing in the Lab

The high-bandwidth compute programme cleared its most important gate. Against an Investor Day design target of roughly 0.5 picojoules per bit, the silicon is running at about 0.3 in the lab, yields are exceeding expectations, and the workload where the efficiency advantage shows up most is decode. Two major prospects were added in the quarter, and the technology is now entering custom silicon conversations as a component to be combined with customers' own accelerators.

Q: "it is good to see the HBC demos that you outlined, earlier this week. Can you maybe speak to what type of workloads you are seeing the traction for your accelerator parts initially?"
— Joshua Buchalter, TD Cowen

A: "We are engaged with customer. It changes the conversation when you have a PowerPoint to when you have silicon."
— Cristiano Amon, President and CEO

Assessment: Beating a power-efficiency design target by 40% with yields ahead of plan is the strongest technical datapoint this programme has produced, and specialising in decode is a credible wedge because decode is where inference economics are decided at scale. The upgrade trigger we set in July was a silicon demonstration that converts a named merchant customer. Half of that has been delivered: the silicon works better than specified, and the customer list is still two prospects and no names. The fiscal 2029 $15 billion target remains the least underwritten number in this company's plan.

The Automotive Beat and Whether It Can Be Pre-Bought

Automotive beat a target that had been raised one quarter earlier, which invited the same pre-buying question as handsets given announced price increases. The answer moved the ground from orders to content: the growth is the fifth-generation platform transition translating a design-win pipeline into revenue, with an eight-fold silicon content increase from the third-generation platform, and the product moving from chips to integrated compute modules.

Q: "I wanted to ask about automotive. With regard to the beat in the quarter versus expectations. I know you are talking about IoT, automotive and data center all growing 60% for the year. But specifically in automotive, it is particularly strong versus expectations."
— Benjamin Reitzes, Melius Research

A: "there is a lot more a lot more compute content. Into the automotive and most of our automotive growth is really share where the design win pipeline translates into revenue. that is where we are starting to see step function increases on the SoC revenue and that will continue."
— Cristiano Amon, President and CEO

Assessment: The content answer is the right rebuttal and the quarterly series supports it. Automotive rose sequentially in all four quarters of fiscal 2026, from $1,101 million to $1,990 million, which is not the shape of a pull-forward; pre-buying produces a spike and a payback, not four consecutive increases inside a platform transition. The guide for automotive and IoT combined to be roughly flat sequentially in the first quarter is the honest counterweight, and it means the 89% growth rate decelerates sharply from here even if the revenue level holds.

What They're NOT Saying

1. QCT gross margin, as an actual number: For two consecutive quarters management has discussed QCT gross margin relative to a 48% to 50% band without once disclosing where it is. The segment disclosure stops at earnings before taxes, and consolidated cost of revenues cannot be split, so the metric at the centre of the investment debate is unobservable by design.
2. Android handset revenue, separated from Apple: Last quarter's explicit commitment was double-digit sequential growth in handset revenue from Chinese OEMs. This quarter delivered a 1.0% increase in total handset revenue, a miss against the Street's roughly $5.25 billion, and the statement that revenue grew into September and December. The two components move in opposite directions and only the sum was disclosed, which makes last quarter's commitment impossible to grade.
3. Any fiscal 2027 earnings or full-year margin framework: The third consecutive quarter without one. Fiscal 2027 revenue was framed in unusual detail, with non-handset growth, segment mix, data center exit rate and seasonality all quantified. The margin and earnings side got one quarter of guidance and a 30% QCT target with a two-to-three-year timeframe.
4. The pending Arm remedies phase: Management framed the Arm verdict as a victory for innovation and competition across a full paragraph of prepared remarks without mentioning that the remedies phase is pending. It entered the call only because an analyst raised it.
5. The 155-day inventory position: Inventories rose 32.4% to $8,640 million and consumed $2,049 million of operating cash flow, in a quarter whose entire debate was about pre-buying ahead of price increases. Neither management nor any analyst mentioned it. The composition is reassuring and the level is not.
6. The customer that fell out of the concentration table: A customer or licensee representing 13% of revenue in fiscal 2025 dropped below the 10% disclosure threshold in fiscal 2026. That appears in the annual report and was not raised on the call, and it is the single clearest measure of how severe the handset reset was at the customer level.
7. Fiscal 2027 capital return, and the warrant's revenue treatment: No framing of buyback pace for the coming year, with most of the $20 billion authorisation unused, the quarterly repurchase down to $1 billion, capital returned at 134.9% of free cash flow and net debt up $5.1 billion. Separately, the Amazon warrant is recorded as a reduction to revenue as related revenue is recognised, a fact that appears only in the 10-K and that makes the $5 billion data center target larger than the revenue line it will produce.

Market Reaction

• Pre-print setup: Closed at $173.19 on November 4, up 1.3% year to date against the S&P 500's 12.3%, up 0.2% over twelve months and down 4.2% over thirty days, inside a 52-week closing range of $124.07 to $251.02. The stock had rallied 17.3% from the $147.61 close recorded in our third-quarter note on July 31.
• After-hours move: Down roughly 5% to $164.50 on the release, extending to more than 6% lower after the call. This is the earnings-attributable reaction and the only web-sourced level in this section.
• Reaction session: Opened at $150.06 on November 5, a 13.4% gap down, traded between $148.25 and $160.80, and closed at $156.75, down 9.5% or $16.44.
• Volume: 40.0 million shares against a 30-day average of 10.8 million, 3.7 times normal.
• Index context: The S&P 500 fell 4.4% and the Nasdaq Composite 5.5% in the same session, with the Nasdaq entering correction territory. The Philadelphia Semiconductor Index fell roughly 12%.
• Peer reaction: Every major semiconductor name fell harder than the market and most fell harder than Qualcomm. On the same session, memory and compute peers declined between 9% and 22%, with the broad semiconductor ETF complex down about 11%.

Separate the two moves. The reaction to this print was about 5% to 6%, delivered in after-hours trading on November 4, and it is explicable: handsets missed consensus, the earnings guide landed marginally below the Street, and the first-quarter midpoint implies a 36% year-over-year earnings decline. That is a reasonable repricing of a transition year. What happened next was not about Qualcomm. The November 5 session was a market-wide repricing of the artificial-intelligence capital-expenditure trade following the midterm elections, with concern about AI regulation and the pace of data center build-out reversing the prior day's post-election rally and erasing something in the order of two trillion dollars of market value. Qualcomm opened down 13.4% in that tape and closed down 9.5%, recovering $8.50 from its low.

The relative performance is the signal. On the day its own sector suffered its worst session in years, on its own earnings day, with six price-target cuts landing in the same session, Qualcomm fell less than the semiconductor index, less than the merchant compute names, and far less than the memory complex. That is not what a stock does when the market has decided the quarter was bad. It is what a stock does when the quarter removed risk and the tape was busy with something else. The stock has been a relative safe haven inside its own sector precisely because it does not depend on hyperscaler capital expenditure to the degree its peers do, which is an irony worth noting on the day its own data center business started billing.

What the price-target cuts actually say. Six firms lowered targets in a single session and not one changed its rating. That is the signature of an estimate revision rather than a thesis revision: the Street cut its fiscal 2027 earnings base and mechanically repriced. The surviving dispersion is extraordinary, with new targets spanning a 37% range against a $156.75 close and two desks still carrying targets more than a quarter above the stock. A market that cannot agree on fair value within 37% after a quarter that resolved three of the four open questions is a market that has not finished processing the quarter.

Street Perspective

Debate: Is Fiscal 2027 a Trough or Another Leg Down?

Bull view: The bull case on the Street is that the first quarter is the bottom and everything after it improves mechanically. Pricing actions begin benefiting gross margin in the second fiscal quarter and reach full effect in the second half, data center scales from $650 million to a $7 billion run rate within the year, automotive compounds off an $8 billion exit rate, and the Apple product revenue being lapped shrinks toward irrelevance. On that path the second half of fiscal 2027 earns at a run rate well above the first, and fiscal 2028 starts from a $7 billion data center base with pricing fully in effect.

Bear view: The bear camp points out that fiscal 2027 earnings are likely to be below fiscal 2026 even with revenue growing 6% to 11%, which is the definition of a business whose unit economics are deteriorating faster than its scale is improving. The first-quarter guide annualises to roughly $9, the remaining three quarters have to average about $2.76 to reach a flat year, and nothing in the guide underwrites that ramp beyond management's assertion. Estimates are still being cut and have been cut in every quarter of fiscal 2026.

Our take: The bear camp has the better description of fiscal 2027 and the bull camp has the better description of what happens after it. Our own estimate is roughly flat earnings on 6% to 11% revenue growth, with the back half doing almost all of the work, and we agree that is the central execution risk. What has changed is that the ramp is no longer a hope. The gross-margin schedule is in a written outlook statement, the data center revenue is in a quarterly guide with shipments already begun, and the automotive run rate has beaten a raised target twice. A flat year you can underwrite is worth more than a flat year you cannot, and the multiple has already contracted as though the year were worse than flat.

Debate: Is the Data Center Revenue High Quality?

Bull view: The constructive view is that $5 billion of revenue arriving within a year of the target being set, from binding purchase orders, with a named hyperscaler anchoring a framework worth up to $60 billion over ten years, supply rather than demand as the binding constraint above $5 billion, and silicon beating its own power-efficiency design target by 40%, is as de-risked as a first-year semiconductor ramp gets.

Bear view: The sceptical view is that the $5 billion is custom application-specific silicon, the lowest-margin form factor the company sells, concentrated in two customers, dilutive to gross margin by management's own quantification, and recorded net of a warrant that reduces revenue as it vests. The merchant products that carry the $15 billion fiscal 2029 target have exactly one named customer, and the entire category depends on hyperscaler capital expenditure that the market repriced violently on the very day of the print.

Our take: Both are right about different years, and the bear case is right about the part that matters least to the current valuation. Fiscal 2027 and 2028 data center revenue is contracted, concentrated and low-margin, and it is also real, growing and large enough to replace the Apple hole. Fiscal 2029's $15 billion is a merchant-product aspiration with one named customer and should carry no weight in a price today. The customer concentration is the risk we would underwrite least comfortably, because two hyperscalers deciding to slow a programme is a $5 billion revenue event, not a rounding error. We would rather own the contracted near years at 15x than pay for the uncontracted far ones.

Debate: Do the Licensing Renewals Deserve a Re-Rating?

Bull view: One camp argues this is the most underpriced development of the quarter. Roughly a third of earnings, generated at a 72% margin, has moved from a negotiation that could have gone badly to a contracted annuity running to at least fiscal 2030 and in places fiscal 2033, at rates management confirms were held. Licensing businesses with that duration and that margin are valued at multiples the whole company does not currently command, and the risk premium attached to the renewal should now come out of the stock.

Bear view: The counter is that removing a tail risk is not the same as adding value. The run rate is explicitly unchanged, so there are no incremental earnings, the royalty base remains capped by handset average selling prices in a flat-to-down unit market, and the renewals arguably just confirm what the market already assumed. A risk that does not materialise was never in the price to begin with.

Our take: The bull side is more right, and the reason is specific rather than general. We flagged in July that Apple and Huawei had gone unmentioned for two consecutive calls, and that the silence was growing more conspicuous because the Apple product relationship had just deteriorated. A licensee that has successfully insourced your chip is in the strongest negotiating position it will ever occupy, and the rate held anyway. That outcome tells you something about portfolio leverage that the pre-renewal price could not have known, which is exactly what a re-rating is supposed to respond to. The bear point about no incremental earnings is correct and is a reason this is worth a lower discount rate rather than a higher growth rate.

Model Update & Valuation Framework

Valuation impact: Fiscal 2026 closed at $10.54 of non-GAAP earnings per share, $0.03 above the ~$10.51 we published on August 1. We model fiscal 2027 at $10.00 to $10.75, with a central estimate of approximately $10.35, which is roughly flat with fiscal 2026 and heavily weighted to the second half: against the $2.25 first-quarter guide midpoint, the remaining three quarters must average about $2.70. At the November 5 close of $156.75 that is 14.9x the fiscal 2026 result and 15.1x our fiscal 2027 estimate. We raise the multiple to a range of 16.5x to roughly 19x from 15x to 17x, because the two variables that made the earnings base unknowable a quarter ago are now bounded: the licensing cash stream is contracted to at least fiscal 2030 at held rates, and the gross-margin recovery has a written date. That gives a fair-value range of $170 to $195, raised from $150 to $180, implying +8.5% to +24.4% from the current price with a midpoint of +16.4%. Adding the 2.35% dividend yield on the $3.68 annualised rate gives a midpoint total return near +19%. The upper half of the range carries the look-through to a fiscal 2028 that begins from a $7 billion data center run rate with the pricing actions fully in effect, which on our estimate earns roughly $13 and puts the stock at about 12x two years out.

Thesis Scorecard Post-Earnings

Commitments From Last Quarter, Graded

Overall: Thesis strengthened, and strengthened in the specific places where it was weakest. Of the ten commitments carried into this quarter, four were delivered or beaten, three were partly delivered, and three were missed or remain absent. The three that were missed are the Android disclosure, the fiscal 2027 earnings framework and the buyback pace, and none of them changes the earnings base. The ones delivered do: licensing is contracted, the margin trough is dated, data center is billing, and automotive beat a raised target for the second consecutive quarter. Fiscal 2027 remains a flat-to-down earnings year on our estimate, which is why this is an upgrade to Outperform rather than to high conviction.

Action: Buy. We upgrade to Outperform from Hold and raise fair value to $170-$195 from $150-$180. The position we have argued for four quarters was that this business was never bad, but that two unquantified variables made its earnings base unknowable and therefore not worth paying for. Both are now bounded. At $156.75 the stock trades at 14.9x a fiscal 2026 result the company delivered slightly above the midpoint of its own guide, 15.1x a fiscal 2027 we model as flat, and roughly 12x a fiscal 2028 that starts from a $7 billion data center run rate with pricing fully in effect. We would use the sector-driven weakness rather than wait for it to pass. Downgrade triggers from here: evidence that either of the two hyperscaler programmes is slipping, another quarter in which Android handset revenue is not separated from Apple, or gross margin failing to inflect in the fiscal second quarter as the pricing schedule now requires.