QCOM (Trans): Initiated double-digit product price hikes
Dolphin Research Trans of $ Qualcomm.US FY26Q3 Earnings Call
I. Key Financials Recap
1. Shareholder returns: Returned $2.3bn in the quarter, including $1.4bn in buybacks and the remainder via cash dividends.
2. Results: FQ3 revenue was $9.9bn, at the high end of guidance, with non-GAAP EPS of $2.21. By segment, QCT revenue was $8.5bn with an EBT margin of 26%, both in line with guidance; QTL revenue was $1.3bn with an EBT margin of 69%, in line with expectations.Within QCT, Handsets delivered $5.1bn, reflecting the hit to global smartphone demand from memory inflation. IoT revenue was $1.8bn (+9% YoY), driven by Industrial, Networking and Robotics; Auto revenue was $1.6bn (+61% YoY), a new quarterly high.Combined non-handset QCT (Auto + IoT) grew 28% YoY.
3. Guidance: FQ4 revenue of $9.7–10.5bn and non-GAAP EPS of $2.05–2.25. QTL revenue of $1.2–1.4bn with an EBT margin of 68%–72%, consistent with seasonality; QCT revenue of $8.4–9.0bn with an EBT margin of 23%–25%.QCT Handsets at approx. $5.2bn, with Android QoQ growth partly offset by lower Apple; IoT roughly flat YoY as double-digit growth in Industrial/Networking/Robotics is offset by tablet and other consumer products constrained by memory; Auto up ~60% YoY to another record.Non-GAAP Opex at approx. $2.7bn, reflecting Modular consolidation and continued DC roadmap investments ahead of revenue ramp.
4. Gross margin and costs: Wafer, packaging, test, advanced packaging, memory and other input costs are all rising, and high-end mix has softened as OEMs trade down among premium SKUs and carry over prior-gen platforms to offset memory costs. Near term, QCT GPM sits slightly below its historical range.The baseline GPM range for the core biz is 48%–50%, and the company has implemented double-digit price increases. These will phase in over time due to existing contracts and product cycles and should flow through margins over the next few quarters.Initial DC revenue is weighted to custom silicon with meaningfully lower margins vs. baseline, creating a 150–200bps drag on QCT blended GPM.
5. Mid/long-term targets and Apple: At Investor Day, raised FY2029 QCT non-handset revenue target to $40bn (from $22bn), with Auto + IoT at >$24bn and DC at >$15bn. DC is expected to scale from $5bn in FY27 to $15bn in FY29.Non-handset share of QCT should exceed 50% in FY27 and reach roughly two-thirds by FY29. Non-handset YoY growth should accelerate from 24% in FY26 to >60% in FY27.Given supply constraints, Apple-related revenue will fall faster starting in FQ4; share in the next iPhone cycle will be well below the prior 20% estimate, and FY27 Apple revenue will be below the earlier 'just over $2bn' guide. The FY27 non-handset ramp is expected to fully cover all FY26 Apple revenue within the year.
II. Call Details
2.1 Management highlights
1. Strategy and industry backdrop
a. The 'next chapter' from Investor Day spans three pillars: entering DC with four distinct product lines; driving agentic AI and physical AI compute across all endpoints; and expanding from chips into full-stack software and platform solutions.
b. In the near term, the sector faces unprecedented inflation in memory and manufacturing inputs and DC-driven supply shortages, pressuring Handsets and consumer electronics. QCOM has initiated price increases; as these roll through, margins should re-align with the model.Despite these headwinds, overall revenue should still grow in FY27, led by an inflection in non-handset businesses across the full year.
2. Data center
a. Entry timing: Agentic workloads are redefining AI economics, making tokens-per-dollar and TCO the core of scale. Inference is decoupling and becoming distributed across DC, on-prem, network edge and devices, creating a continuum where DC entry is a natural extension of QCOM’s low-power compute assets.
b. Phased product cadence: Connectivity in FY26; custom silicon and AI accelerators in FY27; server-class CPUs in FY28. The full portfolio will land over the next two years, leveraging decades of low-power compute expertise and ecosystem ties.
c. Two recent custom ASIC wins will start contributing from the Dec. qtr., with wafers already in fab. Both are phase one of multi-year strategic relationships and are expected to expand.
d. HBC (high bandwidth compute) Gen 1 has taped out, moving into the next phase of customer sampling. By tightly integrating compute and high-density memory, HBC targets one of the toughest bottlenecks, improving perf/W, memory efficiency and TCO.Over the next few quarters, QCOM will demo silicon performance, and the first HBC solution is planned for mid-2027.
e. At the commercial platform layer (HBC-based AI accelerators, SerDes connectivity, CPUs), QCOM is in long-term discussions with nearly all top-tier DC operators.
f. The Modular acquisition has closed and integration is underway, strengthening the end-to-end software stack for DC and edge AI deployment. Hardware-agnostic, it simplifies cross-platform AI software and gives developers an open environment for heterogeneous compute.The goal is not just stronger AI software but pushing the industry from closed to open AI systems to foster competition, innovation and resilience. Modular will host ModCon in Aug.
3. Automotive
a. Signed a milestone expansion with BMW, winning a highly competitive selection to supply core compute for next-gen ADAS and digital cockpit. The programs extend into the next decade, materially expanding the Auto pipeline.
b. The newly announced collaboration with Stellantis extends the Auto pipeline into the 2030s.
c. Customers are shifting from socket-by-socket wins to multi-generation strategic partnerships, recognizing QCOM’s breadth, platform approach and long-term commitment.
d. Snapdragon Digital Chassis Gen 5 ramps in Sept., with meaningfully higher content per vehicle. QCOM is poised to become the No.1 auto semi vendor by reported revenue.
e. FY2026 Auto exit run-rate target raised from $6bn last qtr. to approx. $7bn.
4. Industrial & IoT
a. Investor Day set an $8bn FY2029 revenue target for Industrial, Networking and Robotics.
b. Design-win pipeline now exceeds $7bn, with >$3.5bn added this fiscal year, reflecting strong demand and new logos.
c. The category spans a broad portfolio of application-specific chips and full-stack software, serving >38k customers with a deep partner ecosystem. With Arduino and Edge Impulse, total users now exceed 30mn.
5. Handsets
a. Despite memory-driven market contraction, the agentic smartphone cycle is emerging and should broaden. Major China OEMs are preparing on-device agents and orchestrators, and as penetration rises, agentic experiences will matter more for premium demand.
b. Share at Samsung remains solid, with Snapdragon in ~70% of its flagships. Post-Galaxy Unpacked, the partnership is expanding across the Galaxy ecosystem, including the latest foldables, Galaxy Watch, and smart glasses co-developed with Google.
c. Beyond phones, PCs, smart glasses and other new personal AI endpoints are becoming agent entry points. Installed devices will need upgrades to support more personalized, context-aware and autonomous AI, creating a multi-year replacement cycle.
6. PC and emerging devices
a. QCOM remains ahead in Chromebook design wins, pairing Snapdragon with Gemini to deliver AI-first laptops.
b. Working with Microsoft on Project Solara, building a chip-to-cloud platform for agent-first enterprise devices.
c. The Snapdragon Start smart glasses program offers a full reference platform to enable brands to build devices. With ~600mn pairs shipped annually, the initiative aims to scale the ecosystem and accelerate the shift to smart glasses.More details will come at the Sept. Snapdragon Summit.
2.2 Q&A
Q: With rising input costs and price hikes in motion, how does QCT margin return to prior levels, and how long until ASP gains offset the cost inflation?
A: Two drivers are pressuring margins. First, high-end mix has softened slightly as customers trade among multiple premium chips and carry over prior-gen platforms to mitigate memory inflation. Second, input costs are rising across the supply chain.QCOM is raising prices to pass through higher costs, and the benefit should appear over the next few quarters. The increases phase in due to existing contracts and product cycles, and margins should return to historical ranges once this process plays out.
Q: DC ASIC revenue starts in the Dec. qtr.; what does the FY2027 ramp look like?
A: Revenue begins in the Dec. qtr. and climbs through the year. QCOM previously disclosed two custom chip programs with global hyperscalers, and both will start recognizing revenue in the Dec. qtr.With orders in hand and wafers in fab, QCOM has high confidence in both programs landing.
Q: What is the magnitude and scope of price hikes? How will the market absorb increases in price-sensitive consumer categories like Handsets?
A: The increases are broad-based across end markets. As noted, some areas will phase in over time due to active contracts or product transitions, similar to peers, and the magnitude is double digits, broadly in line with the industry.Crucially, the market downturn is fundamentally about memory-driven BOM inflation. Relative to memory’s step-up in BOM, a double-digit chip price increase is modest and simply passes through input and wafer cost inflation.QCOM does not expect a fundamental change in volumes at premium and above, and it maintains that China Handsets bottomed in FQ3.
Q: With Apple trending down and China OEMs recovering, how should we think about seasonality for Handsets into the Dec. qtr.?
A: While not guiding the Dec. qtr. yet, QCOM can frame the sequential trend from FQ4 to FQ1 for QCT. There are three key variables.First is Apple: share in the new iPhone will be well below the prior 20% estimate, and Apple-related revenue is expected to fall by ~50% from the Sept. to Dec. qtrs., accelerating the Apple exit path. Second is Android, which should grow and largely offset Apple’s decline. Third, DC revenue starts to ramp in FQ1.Netting these, Dec. qtr. revenue should see a slight sequential uptick. Structurally, the old pattern of 'FQ1 as peak quarter' no longer applies—as Apple’s historical Dec. spike exits the model, revenue is now expected to continue growing from Dec. to Mar.
Q: Does the faster Apple decline imply QCOM is proactively reallocating supply away from Apple to other customers?
A: No. Supply constraints are part of the picture, and the outcome of commercial discussions is a share well below 20%, implying FY27 Apple revenue below the prior 'just over $2bn' guide.The right way to view FY27 is that non-handset, targeted to grow >60% YoY, will fully replace Apple-related revenue within the year.
Q: At ~60% YoY non-handset growth, the ~$7.5bn increase roughly matches FY26 Apple revenue, most of which disappears in FY27. Is that the right read?
A: That’s a reasonable range estimate. Additional details are available in the supplemental materials on QCOM’s website.
Q: Margins should improve over coming quarters, but DC ramps are margin-dilutive per Investor Day. How much will DC offset the margin recovery?
A: Think of baseline GPM at 48%–50%, which QCOM expects to re-attain as price increases take effect. DC revenue, initially weighted to custom ASICs, carries materially lower margins vs. baseline and will dilute QCT blended GPM by about 150–200bps.
Q: Auto is set to ramp exiting this fiscal year. What is the current revenue mix, how large is ADAS, and will ADAS be a bigger driver next year? Is autonomy starting to contribute?
A: The Investor Day mix shows ADAS as a meaningful share of design wins, indicating how revenues should realize over the next several years. The key is a shift to platform-level wins—engagements with major OEMs increasingly span the full in-car compute platform rather than single sockets or point features.As customers move to next-gen chips, compute scales by orders of magnitude, driving a step-up in silicon content per vehicle. OEM adoption of next-gen platforms is pushing Auto revenue above prior expectations.
Q: Price hikes reflect higher input costs. How do you assess your own supply chain, especially potential wafer constraints?
A: The industry looks similar to the pandemic period, with 100% utilization across the chain and shortages and price hikes in wafers, packaging, test and test equipment. Scale matters, and having substantial volumes across nodes matters, both of which help.Inventory is also a strategic asset in shortage cycles, and QCOM’s actions here are visible. QCOM is confident current supply can support execution, even if everyone would like more leading-edge capacity.
Q: You doubled the DC target in June. Beyond the two named customers, what is the broader market feedback—could customer count rise, and how does the scope expand?
A: Two parts. First, for existing customers, there is clear visibility on expanding from custom ASICs to accelerators starting with HBC and then to CPUs.This maps to revenue growing from $5bn to $15bn, and QCOM will update these figures over time. At $15bn, the mix includes multi-product, multi-generational custom ASICs plus scaling accelerators and CPUs.Second, many prospects want to see silicon first—for accelerators (HBC-based) and CPUs alike. HBC is on plan and is a disruptive technology; as noted, silicon is expected over the next few quarters for demos and evaluation, which should open new opportunities.QCOM has additional engagements under discussion, but the next milestone is to deliver silicon, validate performance, and then decide on further expansion.
Q: On the China bottoming call, are OEMs signaling greater confidence in domestic memory supply, and how sustainable is the recovery?
A: Several pieces. China revenues bottomed in the Jun. qtr., and QCOM expects double-digit growth in the Sept. qtr., with high confidence, and to continue into the following qtr.Earlier this year, OEM purchases reflected both market demand and channel inventory digestion. Channel inventories are now compressed to levels that can no longer be worked down, so revenues are reverting to true market size—this structural shift underpins confidence.Domestic memory is indeed an important part of China OEM sourcing. That isn’t new, has been the case for years, and should continue to strengthen.
Q: How do U.S.–China controls affect DC, and how is this factored into FY27 ramp decisions?
A: For current programs, QCOM is not restricted and falls into the same category as others delivering solutions to China, so it is not a present concern.
Q: If memory remains a headwind, how does the 2027 Handset TAM evolve for QCOM—still capped, or can it return to the 5% CAGR from Investor Day?
A: The current view is low-teens YoY declines from FY25 to FY26, with the impact centered in mid/low tiers. For QCT Android, that implies ~20% YoY revenue decline and a >$1.50 EPS headwind.Over a longer cycle, as conditions stabilize and with Snapdragon’s product leadership and the agentic AI transition in phones, QCOM can rebuild end-to-end scale and growth. That >$1.50 EPS headwind should convert into upside leverage as the market normalizes.
Q: QTL is holding up despite market softness. Will client price increases add to QTL, and can this resilience continue?
A: QTL follows its existing model—royalties are tied to device ASPs with a cap. Below the cap, higher device prices add some revenue; above the cap, there is no incremental impact.This is fundamentally the normal operation of existing license agreements.
Q: You have stepped up investments and closed several deals reflected in the Sept. qtr. guide. How should we think about opex beyond that and next year?
A: Investor Day guidance stands. Use the current opex run-rate, which will include AlphaWave and the newly consolidated Modular.QCOM continues to invest in the DC CPU and accelerator roadmap, so there will be some growth, which will be the primary driver of opex trends ahead.
<End>
Risk disclosure and statement:Dolphin Research Disclaimer & General Disclosure
