---
title: "QCOM: Handset base erodes — even AI can't offset it?"
type: "Topics"
locale: "en"
url: "https://longbridge.com/en/dolphin/post/44796961.md"
description: "Data center ramp is strong, but GPM is still compressing"
datetime: "2026-11-05T03:07:45.000Z"
locales:
  - [en](https://longbridge.com/en/dolphin/post/44796961.md)
  - [zh-CN](https://longbridge.com/zh-CN/dolphin/post/44796961.md)
  - [zh-HK](https://longbridge.com/zh-HK/dolphin/post/44796961.md)
author: "[Dolphin Research](https://longbridge.com/en/dolphin.md)"
generator: "portal-rs"
---

# QCOM: Handset base erodes — even AI can't offset it?

Qualcomm (QCOM.O) released its FY26 Q4 results (quarter ended Sep 2026) after the U.S. market close on Nov 5 (Beijing time). Key points follow.

**1) Next-quarter guide: management guides FY27 Q1 revenue of $10.4–11.2bn, broadly in line with the Street at $10.55bn.** The company guides Non-GAAP EPS of $2.15–2.35, with the midpoint slightly below consensus at $2.30.

**By segment, QCT revenue is guided at $9.0–9.6bn (now including data center starting this quarter; Street $9.01bn),** and QTL revenue at $1.3–1.5bn (Street $1.37bn).

Within QCT, Dolphin Research estimates **next-quarter handset revenue at $4.8–4.9bn (down mid-single-digit QoQ),** as Android growth is offset by a decline in Apple-related revenue. **Data center revenue will exceed $650mn in the first quarter (ahead of the Street at $430mn),** while auto + IoT combined should be roughly flat QoQ at about $3.9bn.

$Qualcomm(QCOM.US) guided **QCT EBT margin at 20–22%, down about 3ppt QoQ,** which Dolphin Research attributes to two drivers. **First, upfront investment and an early mix skewed to custom data center silicon dilute overall GPM, and second, the continued roll-off of high-margin Apple revenue.**

**2) Business mix:** Qualcomm’s business comprises Semiconductor (QCT) and Technology Licensing (QTL). **Semiconductors remain the largest revenue contributor, accounting for over 80%.**

**Within Semiconductors: ① Handset revenue was $5.14bn this quarter, down 26.2% YoY,** slightly below the Street at $5.25bn. The decline stems from two areas: **Apple baseband share stepped down faster, and memory price hikes pressured Android mid/low-end shipments, prompting OEM spec reductions and channel inventory digestion.**

**② Auto revenue was $1.99bn this quarter, +89.0% YoY, meaningfully ahead of the Street at $1.69bn,** driven by initial shipments of the 5th-gen Snapdragon Digital Chassis and broader adoption of integrated compute modules. Management noted the 5th-gen platform carries per-vehicle silicon content about 8x that of the 3rd-gen.

On an annualized run-rate basis, **auto revenue has reached about $8.0bn, roughly $1.0bn above last quarter’s $7.0bn outlook.**

**③ IoT revenue was $1.89bn this quarter,** +4.5% YoY and broadly in line with the Street at $1.85bn. Industrial networking and robotics grew about 20% YoY, but **consumer IoT remained constrained by memory shortages.**

**3) Operating metrics: GAAP revenue was $10.3bn, down 8.4% YoY and slightly above the Street at $10.12bn.** The outperformance was narrow.

**Almost all the beat came from Auto: this quarter’s $1.99bn exceeded expectations by roughly $300mn,** while **Handsets undershot by about $120mn,** reflecting last year’s high base and the accelerated exit of Apple baseband share.

GAAP revenue included a roughly $120mn offset related to a litigation settlement customer payment. The company added this back under Non-GAAP.

**GAAP GPM was 50.7%, down 460bps YoY and below the Street at 52.7%.** Excluding SBC, acquisition amortization, and the one-off charge above, Non-GAAP GPM was 52.4%, still lighter than the Street at 53.5%.

Costs rose broadly across wafers, packaging & test, advanced packaging, and memory. **Pricing increases will not begin to show until FY27 Q2 and will be fully reflected in the back half, leaving this quarter as the widest phase of the ‘costs up first, pricing later’ scissors effect for QCT margins.**

**4) Opex and profit:** GAAP opex was $3.88bn this quarter, including R&D of $2.68bn (+13.2% YoY) and S&M of $960mn (+5.8% YoY). There was also $235mn in restructuring charges, mostly severance.

**The Street had expected R&D to decline QoQ, but that runs counter to management’s continued data center push; reported R&D exceeded expectations at $2.42bn.** Excluding items such as SBC ($850mn), Non-GAAP opex (R&D + S&M) was $2.705bn, in line with company guidance of about $2.7bn.

GAAP operating profit was $1.36bn this quarter, down 53.5% YoY. **This was clearly below the Street at $1.71bn, primarily on lower GPM, higher R&D, and restructuring costs.**

**Ending inventory was $8.64bn, up 32% YoY,** with days increasing to about 155 vs. roughly 118 a year ago. There were also **$18.6bn of purchase commitments (of which $13.8bn due within 12 months), mainly multi-year capacity locks.**

**For the transcript, click**[**here**](https://longbridge.com/en/dolphin/post/44790051).

**Dolphin Research view: data center is scaling fast, but margins are still slipping.**

**Quarterly results were mixed; revenue modestly beat on Auto, but the YoY decline widened.** GPM continued to compress on broad-based cost inflation and falling high-margin Apple revenue. More importantly, **OP was roughly halved YoY, and the handset core remains in adjustment.**

Next-quarter revenue and EPS guidance are broadly in line with expectations, but QCT margins continue to trend lower, driven by lower-margin custom data center silicon. **The bright spot is data center scaling faster than expected, with mid-term targets intact and a plan for sequential ramp,** with **the upside constrained by supply rather than demand.** After Apple and Huawei renewals, QTL risk is largely removed.

**Dolphin Research believes management’s commentary supports the ‘data center baton’ as the mid-term thesis.** However, near-term stock catalysts are limited, as memory and other cost pressures plus Apple baseband share losses will keep weighing on results. Relief likely won’t come until price increases flow through from FY27 Q2 onward.

The market is focused on several areas beyond this print:

**1) Legacy core (base): Apple exits faster + memory pressure persists.**

Handsets are Qualcomm’s largest business at roughly half of revenue. With the broader handset market weak, the drag is evident. **Global handset shipments were 255mn units this quarter, -21% YoY.** Management expects the global handset market to decline by low-to-mid teens in 2026 and be flat to slightly down in 2027.

The market splits into Apple and Android. **Apple shipments fell 2% YoY this quarter, while Android fell 25% YoY.** For Qualcomm, **iPhone baseband share drops from about 70% in the Sep quarter to roughly 40% in the Dec quarter, and below 20% in iPhone 18.** Management said the timing and cadence of the Apple baseband switch are fully embedded in the operating plan.

**Management flagged several marginal positives:** **① China Android bottomed last quarter,** with sequential growth in this quarter and next. **② Channel inventory is thin,** and FY27 handset revenue should track end-market demand without the FY26 destock drag. **③ Share in Samsung flagships rises from a 50% baseline to above 70%,** extending into wearables and other peripherals. **④ Price hikes focus on new platforms and models,** limiting noise from early pull-ins.

DRAM/LPDDR contract prices are still rising, and allocation constraints remain. Memory now accounts for over half the BOM in entry-level phones, so pressure on the handset base likely persists near term. **Management assumes the 2027 global handset market will be ‘flat to slightly down’, which Dolphin Research views as optimistic with contract prices still moving higher.**

On margins, **management frames pricing to begin contributing from FY27 Q2 and fully reflect in H2,** with the baseline GPM returning to the lower end of the historical 48–50% range. On top of that, custom data center silicon will further dilute GPM, but it should be accretive to OPM.

For licensing, Qualcomm signed a new patent license with Apple in Sep (effective Apr 1, 2027; terms undisclosed), and a multi-year cross-license with Huawei in Oct. **The feared ‘Apr 2027 Apple license expiry’ risk is largely off the table.**

**Management said renewals preserve prior rates and per-device fees,** and that the next renewal cycle with major OEMs runs FY30–FY33. With 5G-era renewals complete, QTL uncertainty is largely resolved.

**2) AI: on-device AI and data center are potential growth vectors.**

With legacy markets soft, the company is investing across AI to drive new growth. Execution spans multiple end devices.

**a) On-device AI:** the company sees on-device AI as core to its Physical AI strategy, **covering smartphones, PCs, autos, XR, robotics, and industrial IoT.**

Management noted **the first wave of ‘AI agent’ phones launched on Snapdragon platforms,** and leading AI players aim to run 100bn-parameter models on-device with all-day battery by 2028. This requires a step-change in edge compute, increasing per-device silicon content in high-end Android.

The company won **SpaceX’s Starlink Mobile U.S. nationwide 5G terrestrial network order,** with revenue contribution starting by late FY27 and skewing to FY28.

**b) AI data center:** this is the most notable AI strategy. At Investor Day, management set targets of $5bn in FY27 and $15bn+ in FY29 for data center revenue. **This quarter, those targets were reiterated, with the biggest changes in customer progress and revenue cadence.**

**① Amazon warrants and customer progress:** the company issued up to 25mn warrants to Amazon (strike $161.26, 10-year term), tied to up to $60bn of potential purchases over 10 years. Fifteen percent (~3.75mn) vests immediately based on initial purchase commitments.

The grant-date fair value was about $1.8bn ($71.57 per warrant) and will be recognized as a contra-revenue item as related revenue is recognized. **This equates to roughly a 3% ‘rebate’ on the $60bn cap, modestly lowering reported data center revenue and GPM.**

Management emphasized **Amazon’s warrant structure is unique; other data center deals follow traditional commercial models.** Progress-wise, custom accelerators for AWS began shipping in Oct, **optical interconnect products will ramp in Nov, and a second optical interconnect customer has been secured.**

To date, multi-generation design wins include: **Connectivity with 2 customers (adding 1 beyond Amazon),** **custom chips/AI accelerators with 3 customers (combined, incl. Amazon),** and **CPU with 1 customer (unchanged, Meta).**

**② Revenue cadence:** management guides **data center revenue above $650mn next quarter,** then up sequentially, with **FY27 exit run-rate around $7.0bn (i.e., ~$1.75bn per quarter), and the $5.0bn full-year target unchanged.** This implies ~$2.5bn combined for FY27 Q2–Q3.

Dolphin Research expects most of the FY27 $5bn to come from two hyperscalers’ custom AI accelerators (one being Amazon), followed by connectivity. The third accelerator customer and a CPU customer beyond Meta should drive FY28–FY29 upside.

**Management specified the custom parts as ‘main AI accelerators’,** implying higher per-unit value but potentially lower in-house design content and lower GPM. FY28 starts from a ~$7.0bn run-rate, and management said the current order book supports strong FY28 growth, with high coverage for the FY29 $15bn target.

**③ Competitive landscape:** Nvidia is entering decoupled inference via Groq LPX, and Arm is directly selling AGI CPUs. Qualcomm faces direct competition in both inference and CPU, and data center’s lower GPM will weigh on consolidated margins.

On litigation, a jury on Oct 13 found Arm in breach of its Architecture License Agreement with Qualcomm. Management said this does not affect the CPU roadmap.

**Splitting FY27 into legacy (handsets, auto, etc.) and data center: assume legacy grows 2% YoY (handsets -15%, auto +50%+), while data center ramps from zero with a $5bn guide.** Total FY27 revenue growth would be about 13%, driven almost entirely by data center.

At a current market cap of $184.8bn, **FY27 core after-tax OP implies about 26x PE** (assuming 49% GPM and a 12.5% tax rate). **This is well above the company’s historical range of 10–25x, suggesting overall valuation is not cheap.**

**By sum-of-the-parts:** ① Legacy core FY27 after-tax core OP is about $6.0bn; at 15–20x PE, that implies roughly $90–120bn of value. ② For data center, the FY27 $5bn guide could yield about $0.9bn of after-tax core OP (assuming 20% OPM and a 12.5% tax rate); at 30–50x PE, that implies $27–45bn.

**Combined, fair value is roughly $120–165bn.** With handsets still weak and GPM clearly lower, the current $184.8bn market cap looks elevated. Management did deliver decent data center progress; the post-print sell-off mainly reflects softness in the legacy handset base and margin compression.

Qualcomm rode a ‘data center rollercoaster’ this year. FQ2 guidance called for handset revenue down 20%+ QoQ as memory shortages weighed on Android shipments, and shares fell to $127 in late Mar. Then the data center narrative took over, the stock rallied ~95% in two months to ~$249, but after Investor Day positives were priced in and FQ4 guidance missed, it slid back to $156.

Since Sep, Amazon warrants and Apple/Huawei license renewals landed, and the stock rebounded to $197 at one point. **Dolphin Research believes this recovery came from reduced QTL risk premia rather than earnings upgrades,** and **the sharp GPM drop in this print directly pressures FY27 EPS, driving the sizable after-hours decline.**

**The impact of this print differs across the two parts:**

**① Data center:** reiteration of the FY27 $5bn target and exit run-rate of ~$7bn, plus the ‘supply-constrained’ comment, improves certainty. **This portion can be valued toward the upper bound (about $40–50bn).**

**② Legacy core:** margins are still trending down, with pricing benefits only testable in H2 FY27. This is the main reason for the post-print drawdown. After cutting FY27, the legacy core is worth only about $100bn.

**Together that’s roughly $150bn, yet the stock still sits above $180bn, so further de-rating is likely.** Fortunately, management offered solid data center progress, which helps support the ‘future DC optionality’ narrative.

**The Apple and Huawei renewals were already priced in, while this print shows ongoing handset weakness and falling margins.** That defines the current earnings base—the trillion-yuan-equivalent cornerstone—but with no clear improvement likely before Q2 (pricing TBD). Near-term positives likely require announcements of new AI wins or customers, to re-rate the DC option value, though that pool remains relatively small.

\<End>

Related research from Dolphin Research

Sep 9, 2026 First Take: [Amazon as a key anchor: is Qualcomm’s AI second curve now ‘secured’?](https://longbridge.com/zh-CN/dolphin/post/43821429?channel=OWNN00210)

Jul 30, 2026 Trans: [Qualcomm (Trans): double-digit price hikes underway](https://longbridge.com/en/dolphin/post/43053884?channel=OWNN00210)

Jul 30, 2026 Earnings Review: [Qualcomm: AI fever fades—when do handsets exit the slump?](https://longbridge.com/en/dolphin/post/43058646?channel=OWNN00210)

Jun 25, 2026 Investor Day: [Qualcomm (Trans): shifting from chipmaker to platform solutions](https://longportapp.cn/en/topics/42178884)

Apr 30, 2026 Trans: [Qualcomm (Trans): Q3 is the bottom for China Android; new custom ASIC mega-win](https://longportapp.cn/en/topics/40312889)

Apr 30, 2026 Earnings Review: [Qualcomm: can this ‘AI stimulant’ offset handset weakness?](https://longportapp.cn/en/topics/40306582)

Feb 5, 2026 Trans: [Qualcomm (Trans): memory shortages push OEMs to cut production](https://longportapp.cn/en/topics/38482467)

Feb 5, 2026 Earnings Review: [Qualcomm: ‘memory crunch’ deepens—handset stocks in peril?](https://longportapp.cn/en/topics/38476249)

Nov 6, 2025 Trans: [Qualcomm (Trans): AI chip details and metrics to be disclosed early next year](https://longportapp.cn/en/topics/36056666)

Nov 6, 2025 Earnings Review: [Qualcomm: shedding the ‘handset stock’ label—AI compute as the new hope?](https://longportapp.cn/en/topics/36052139)

Jul 31, 2025 Trans: [Qualcomm (Trans): auto and IoT opportunities far exceed Apple-related revenue](https://longportapp.cn/en/topics/32443082)

Jul 31, 2025 Earnings Review: [Qualcomm: handset risks rising—when can AI stories feed real growth?](https://longportapp.cn/en/topics/32441078)

Risk disclosure and disclaimer: [Dolphin Research disclaimer and general disclosure](https://support.longbridge.global/topics/misc/dolphin-disclaimer)

### Related Stocks

- [QCOM.US](https://longbridge.com/en/quote/QCOM.US.md)
- [AAPL.US](https://longbridge.com/en/quote/AAPL.US.md)
- [AAPB.US](https://longbridge.com/en/quote/AAPB.US.md)
- [AAPD.US](https://longbridge.com/en/quote/AAPD.US.md)
- [AAPU.US](https://longbridge.com/en/quote/AAPU.US.md)
- [APLY.US](https://longbridge.com/en/quote/APLY.US.md)
- [AAPY.US](https://longbridge.com/en/quote/AAPY.US.md)
- [AAPX.US](https://longbridge.com/en/quote/AAPX.US.md)
- [AAPW.US](https://longbridge.com/en/quote/AAPW.US.md)
- [QCML.US](https://longbridge.com/en/quote/QCML.US.md)

## Comments (4)

- **YSLV · 2026-11-05T16:38:01.000Z**: Qualcomm is not as good as AMD
- **Dolphin fans · 2026-11-05T07:15:26.000Z**: margin bottoms this quarter. Price hikes kick in next year, so isn’t this dip actually the entry point?
- **NewUser_Leon · 2026-11-05T07:00:14.000Z**: Wall Street restricts the AI investment.
  - **Dolphin fans** (2026-11-05T07:16:51.000Z): Data center guidance beat expectations. The street isn’t punishing AI spend, it’s punishing the shrinking phone base.


---
> **Disclaimer: This article is for reference only and does not constitute any investment advice.**