Qualcomm – A Contrarian Pick With Huge Upside Potential (A Deep Dive)

This is my Qualcomm Deep Dive - A semiconductor giant completely misunderstood

Jul 04, 2026

Most investors look at Qualcomm and see a smartphone chip company with an Apple problem. That reading is not wrong; it is just catastrophically incomplete. Qualcomm is simultaneously the dominant silicon platform for premium Android smartphones, the leading automotive cockpit and ADAS chip supplier with a $65 billion design win pipeline, a monopolist in non-Apple smart glasses at the exact moment the category is inflecting, and the newest entrant into data center inference with Microsoft, Meta, and two hyperscaler custom silicon commitments already signed before a single chip has shipped. The market is pricing the Apple headwind. It is not pricing any of the rest.

What ties all of this together is a single engineering competency that Qualcomm has spent four decades compounding: designing silicon that delivers maximum AI compute performance within minimum power consumption, combined with world-class wireless connectivity, in a single integrated platform. Every market Qualcomm enters — smartphones, automotive, AI PCs, smart glasses, robotics, data centers — turns out to need exactly this combination, which is why it can arrive in new categories with a competitive product in years rather than the decade a true greenfield entrant would require. Apple spent six years and billions of dollars trying to replicate just the modem portion of this platform. That is the moat.

The investment case is straightforward even if the business is not. Qualcomm is a business in the middle of a profound quality transformation that the market refuses to see, distracted by a near-term Apple headwind that is visible, quantifiable, and time-bound. By 2028, that headwind is behind it, and what remains is a semiconductor company generating the majority of its revenue from automotive, IoT, and data center businesses with superior growth rates, lower customer concentration, more contractually predictable revenue streams, and exposure to some of the most exciting technology adoption curves of the decade. The market is currently valuing Qualcomm as the business it is leaving behind, not the business it is becoming. That gap between label and reality is the opportunity.

But don’t take my word for it. Let me show you exactly what I mean. In this Deep Dive, I will break down the business, its products/model, moat, competitive position, growth drivers and potential, the Apple headwind, its financials, recent performance, and ultimately the outlook and valuation so you can judge for yourself whether this one deserves a place in your portfolio or not.

Without further ado, let’s delve in!

Welcome to InvestInsights — an independent equity research publication rooted in long-term, buy-and-hold investing, publishing actionable stock/equity research reports weekly!

📈 You’re reading my latest [FREE] stock analysis. If you like this analysis, make sure to like & subscribe to receive much more like this, weekly!

This is Qualcomm!

Founded in 1985 and headquartered in San Diego, Qualcomm is one of the world’s leading fabless semiconductor companies, focusing on semiconductor design and wireless intellectual property. Today, the company generates over $44 billion in annual revenue and has a market cap of $200 billion.

Now, every fabless design business has its own expertise and focus market. AMD makes exceptional x86 CPUs and GPUs for the PC gaming and datacenter market; Nvidia dominates parallel GPU compute for AI training and high-performance computing; Broadcom designs best-in-class networking silicon, custom ASICs for hyperscalers, and enterprise infrastructure chips; Marvell focuses on custom silicon for cloud infrastructure, storage controllers, and high-speed networking; and MediaTek designs cost-competitive mobile SoCs and connectivity chips targeting the mid-range and entry-level smartphone market.

Qualcomm’s bread and butter is engineering excellence in wireless signal processing and power-constrained compute, making it a very dominant player in networking, high-end mobile devices, automotive, and the very broad IoT market – markets where connectivity and maximum compute performance with minimal power consumption are critical.

Every product Qualcomm makes is an expression of these same capabilities applied to different device categories. And with the world’s device ecosystem converging on the same fundamental requirements — intelligent compute, wireless connectivity, and power efficiency — Qualcomm’s core competency is becoming more rather than less relevant with every passing year, opening the door to successive new markets where these capabilities are the critical differentiator. Think datacenters, where demand is shifting to efficiency-first compute, or the explosion of connected devices across robotics, industrial edge AI, and personal AI hardware.

Qualcomm really is in a sweet spot, and it is well ahead of the competition technologically, thanks to decades of investment and a huge IP portfolio.

In my view, Qualcomm really is misunderstood and underappreciated. So, let’s take a very close look at this business and its operations.

Qualcomm operates through two distinct segments. The first is QCT, or Qualcomm CDMA Technologies, its chipset division responsible for designing the Snapdragon system-on-chip platforms that power premium smartphones, laptops, automobiles, and IoT devices, among many more. This segment generates nearly 90% of Qualcomm’s revenue.

Within QCT, handsets (mobile phones) is the largest revenue contributor, bringing in 57% of group revenue – this remains the gravitational center of the entire business today.

Qualcomm sells fully integrated system-on-chip (SoC) platforms under the Snapdragon brand that combine a CPU (its proprietary Oryon cores at the high end), GPU, NPU for on-device AI, cellular modem, Wi-Fi, Bluetooth, and GPS into a single piece of silicon. This is the core of Qualcomm’s handset business, but it also sells the cellular modem separately, mostly to Apple, which is a distinct product line from the full Snapdragon SoC.

The flagship here is the Snapdragon 8 Elite, which powers premium Android devices from Samsung, Xiaomi, OnePlus, and others – nearly the entire high-end Android market. Below that sits a tiered stack of mid-range and entry-level Snapdragon chips serving lower price points.

And Qualcomm isn’t just any player in smartphone SoC’s, but a pretty dominant force. On raw unit shipment share, MediaTek is the world’s largest smartphone chipset vendor with a 32% market share as of Q1 2026, followed by Qualcomm at 23%. However, Qualcomm is much more skewed toward the high end of the market – while MediaTek powers 69% of smartphones below $600, Qualcomm dominates the high end, which commands a much higher price per chip. Therefore, despite MediaTek’s volume leadership, Qualcomm led the competitive landscape in 2025 by revenue.

Why does Qualcomm mainly target the high-end market? The main reason is that this is where it can leverage its technological lead (including AI capabilities), it carries higher ASPs, and the refresh cycle is more reliable, which gives Qualcomm a degree of revenue predictability that the mid-range and budget tiers, which are far more sensitive to consumer spending cycles, do not.

It is, however, worth noting that Qualcomm isn’t as dominant in premium Android silicon as before. MediaTek has put a strong challenger on the market, securing key design wins and breaking part of Qualcomm’s undisputed monopoly – the Qualcomm Snapdragon 8 Elite Gen 5 is closely followed in specs by MediaTek’s Dimensity 9500. MediaTek is mostly winning some share in Asian markets, while Qualcomm maintains its monopoly in the North American market, and that is unlikely to shift.

In fact, while its dominance may be crumbling a bit, Qualcomm’s competitive position in this market remains very strong, not relying solely on SoC performance but also on the platform it offers, which remains unmatched. As pointed out, the Snapdragon SoC integrates CPU, GPU, NPU, cellular modem, Wi-Fi, Bluetooth, GPS, image signal processor, and audio DSP into a single piece of silicon, all co-designed and co-optimized to work together with shared power management. An OEM that chooses Snapdragon gets a known-good, fully integrated platform with years of software optimization, driver maturity, camera tuning libraries, and a global ecosystem of component partners, software developers, and carrier certifications already built around it.

That is hard to displace, especially with this backed by $9 billion in annual R&D, which no competitor comes close to matching.

Anyway, handsets is still Qualcomm’s core business today, and its position here remains strong, with a best-in-class offering.

But Qualcomm has also been rapidly innovating away from the handset market in recent years, in an effort to diversify from this mature, cyclical business. And with a lot of success, with 43% of revenue already coming from other sources, including automotive, IoT, AI PCs, and, most recently, data center CPUs. Qualcomm splits this into two reporting categories – Automotive and IoT (practically everything except handsets and automotive).

Starting with automotive, which generates roughly 13% of revenue, Qualcomm once again sells a platform rather than a single chip, branded as the Snapdragon Digital Chassis. Within that umbrella, there are three distinct product layers.

The first is the Snapdragon Cockpit platform, which powers digital dashboards, infotainment systems, driver displays, and in-vehicle AI assistants. This is where the bulk of current automotive revenue sits. The second is the Snapdragon Ride platform, which handles ADAS and automated driving, covering everything from basic lane-keep assist up to L2+ autonomous highway driving as demonstrated in BMW’s iX3 via Snapdragon Ride Pilot. The Snapdragon Ride Flex SoC supports mixed-criticality workloads, enabling the digital cockpit and ADAS to run on a single chip and reducing complexity and cost for automakers. The third layer is telematics and connectivity, covering the 5G and C-V2X (cellular vehicle-to-everything) modules that keep the car connected.

Taken together, Qualcomm is selling the compute, the perception, and the connectivity stack for the software-defined vehicle.

In this market, Qualcomm competes with Nvidia’s Drive platform and Mobileye, two strong, well-funded competitors, but Qualcomm’s edge sits in the fact that Mobileye is primarily a vision-based ADAS specialist and Nvidia is pushing centralized compute – Qualcomm offers a much broader platform spanning connectivity, cockpit, and autonomy on a single unified architecture. For automakers trying to simplify their supply chains as they transition to software-defined vehicles, that breadth is a meaningful advantage.

As a result, Qualcomm has emerged as a winner in this market in recent years. It already has 75 million vehicles out there equipped with the Snapdragon Cockpit Platform and is said to hold a 70% market share of the China cockpit chip market, which is particularly significant given that China accounts for roughly 44% of the global digital cockpit market and is where EV adoption and smart vehicle feature demands are most aggressive.

Furthermore, Qualcomm has a strong OEM roster, featuring BMW, Mercedes-Benz, Toyota, and Volkswagen Group, among many more.

In ADAS specifically, one of the most promising long-term opportunities in automotive, Qualcomm is one of the Big 5 (Mobileye, Nvidia, Qualcomm, Horizon Robotics, and Huawei), which hold an estimated 69% of the market and are growing. Qualcomm here remains well behind leader MobilEye in deployments, but it has a huge design pipeline which is expected to translate into strong share gains in the coming years. For reference, Snapdragon Ride Pilot, Qualcomm’s L2+ automated driving system, launched in BMW’s iX3 and has been validated across 60 countries, with the company targeting 100 countries by 2026. More than one million cars now operate ADAS and autonomous functions on Snapdragon Ride processors.

Ultimately, automotive remains a relatively early-stage market, and while competition is significant, Qualcomm’s position is very strong, driven by an excellent, integrated platform. This is a high-growth opportunity for Qualcomm

Next up is Qualcomm’s IoT segment, which accounts for roughly 16% of group revenue. This segment encompasses consumer devices (like smart cameras, drones, and wearables), industrial machines, edge networking equipment, extended reality headsets, AI PCs, and smart glasses, making it less a single vertical and more an umbrella for everything that is connected but is not a phone or a car.

The segment runs on two brand families. Snapdragon addresses the consumer-facing and personal computing categories, with the Snapdragon X series powering premium Windows ARM laptops, the AR1 family is the de facto silicon platform for smart glasses, including Meta Ray-Ban, Samsung Galaxy Glasses, and dozens of fashion brand partnerships; and the XR2 and Reality Elite chips power full VR and MR headsets.

Dragonwing, launched in 2025, covers the industrial and embedded tier, including edge gateways, industrial vision systems, AMRs, drones, and enterprise networking equipment, backed by a rapidly expanding software and developer ecosystem assembled through the acquisitions of Arduino, Edge Impulse, Foundries.io, Augentix, and Focus.AI, giving Qualcomm a 30-million-user developer community and a full-stack platform that no pure-play industrial chip vendor can currently match.

Today, we can safely say that Qualcomm is one of the leading suppliers of high-end IoT chips, helped by its dominant position in connectivity and power-constrained compute.

In recent years, Qualcomm has solved the problem of delivering meaningful performance within a battery-powered phone’s thermal envelope of a few watts sustained. Every other device category Qualcomm has entered turns out to need a less severe version of the same constraint. An AI PC has more thermal headroom than a phone but still needs all-day battery life. A smart glasses device has even less thermal headroom than a phone but needs to run for a full day on a battery a fraction of a phone’s size. An industrial sensor needs to run for years on minimal power. A robot needs sustained AI inference without overheating or draining a battery mid-task.

Add to that its specialty in connectivity, its ability to combine CPU, GPU, NPU, modem, and sensing capabilities onto a single chip with a unified software stack, and the fragmented nature of the market, and Qualcomm appears to be the perfect fit – Qualcomm’s Hexagon NPU architecture delivers best-in-class AI inference performance per watt at precisely the performance tier these devices require.

Let me break down some of the largest products that fall under the IoT umbrella, starting with AI PCs, a market Qualcomm entered just three years ago. Qualcomm’s Snapdragon X series has put it in genuine competition with Intel and AMD in the Windows laptop market for the first time and is expected to generate $4 billion in annual revenue by 2029, rapidly becoming a meaningful contributor.

Qualcomm sells the Snapdragon X series of ARM-based processors targeting Windows laptops. These chips integrate a high-performance CPU (Oryon), a capable GPU, a best-in-class NPU for local AI workloads, and built-in 5G/Wi-Fi connectivity – everything Intel and AMD CPUs require separate components for. The pitch to OEMs combines superior power efficiency (enabling thinner, lighter, longer-battery-life designs) with on-device AI performance via the Hexagon NPU.

Qualcomm’s early-mover advantage in on-device AI capabilities and best-in-class energy usage for Windows CPUs has allowed for rapid adoption of the X series, with 150 design wins to date and Qualcomm rapidly capturing a 9% market share in Windows laptops sold above $600 in key Western markets. That is huge for a product that didn’t exist three years ago, underscoring its quality – being the most power-efficient ARM-based processor for Windows laptops and the raw performance leader in dedicated AI processing.

Amid strong adoption of AI PCs and ARM-based share gains against x86 processors, this is a very compelling market for Qualcomm.

Another promising product category under the IoT umbrella is smartglasses or VR/XR. Here, Qualcomm effectively operates a monopoly in the non-Apple ecosystem. Qualcomm has 30 XR designs in production or development, with strong traction from Meta’s Snapdragon-powered smart glasses alongside Samsung’s Galaxy XR for Google’s Android XR. Every major non-Apple smart glasses and XR headset on the market or in development runs on Qualcomm silicon.

It sells two distinct chip families for XR. The AR1 series is purpose-built for lightweight, all-day-wearable smart glasses, with low power consumption, thermal efficiency, and the ability to run on-device AI models without a tethered phone. The AR1+ Gen 1 is 28% smaller than its predecessor and capable of running Meta’s Llama 3.2 AI model entirely on-device. The Ray-Ban Meta glasses run on the AR1 Gen 1, and Samsung’s upcoming Galaxy Glasses are expected to use the same platform.

Separately, the Snapdragon XR2+ Gen 2 and the newly announced Snapdragon Reality Elite, Qualcomm’s flagship XR chip delivering 60% higher GPU and 160% higher NPU performance over the prior generation, target full VR, AR, and MR headsets like the Samsung Galaxy XR.

So, Qualcomm practically operates a monopoly in an early-stage market with huge growth potential.

Finally, there are robotics and industrial edge AI. Under the Dragonwing brand, Qualcomm sells processors and full compute platforms targeting industrial and embedded applications: smart cameras, drones, industrial vision systems, edge gateways, AI TVs, video collaboration systems, and fixed wireless access equipment. The Dragonwing Q-8750 and Q-7790 processors are designed for secure on-device AI in drones, smart cameras, industrial vision, and video collaboration systems. This is also where Qualcomm’s networking infrastructure chips sit, including Wi-Fi access point silicon, cable modem chips, and enterprise networking equipment processors.

Meanwhile, the Dragonwing IQ10 Series is a new 18-core CPU and full robotics stack designed for industrial autonomous mobile robots and full-size humanoids, with partners like Figure using this platform to power general-purpose humanoid robots. The broader robotics platform roadmap includes the RB series for entry-level and mid-range robotic applications, through to the IQ10, which delivers up to 700 TOPS of AI compute for advanced humanoids.

This is another very early-stage market, not yet a real contributor to financials today, but a huge growth opportunity where Qualcomm is very well positioned, with its core competencies in on-device AI and real-time connectivity.

And then, not really in any category yet, there is Qualcomm’s recent push into data center CPUs. Following the June 2026 Investor Day, Qualcomm unveiled the Dragonfly portfolio of data center CPUs, AI accelerators, and custom silicon.

The first is the Dragonfly C1000 CPU, a data center server processor built on the same custom Oryon core architecture that powers the Snapdragon X PC chips. The C1000 is a chiplet design with over 250 cores, frequencies above 5 GHz, PCIe Gen 7, CXL support, and enterprise-grade reliability features. The target workload is explicitly agentic AI orchestration, the high-throughput sequential reasoning and context-switching that GPU parallelism handles poorly but fast, efficient ARM CPUs handle well.

This one is expected to ship in the second half of 2028.

The second is the Dragonfly AI Inference Accelerator family, with the AI200, AI250, and newly announced AI300, representing an annual cadence roadmap. The AI300 integrates Qualcomm’s High Bandwidth Compute Gen 2 technology, and Qualcomm claims 4-8x better performance per watt for memory bandwidth per card compared to existing GPU-based architectures, with commercial sampling of the AI250 expected in mid-2027 and the AI300 shipping in 2028.

The third and most technically interesting element is High Bandwidth Compute (HBC), Qualcomm’s novel memory architecture, the genuine engineering innovation underpinning the data center strategy. HBM is the current industry standard for high-performance AI accelerators, including Nvidia’s H100 and AMD’s MI300X, and works by stacking multiple DRAM dies vertically on top of each other and placing them in close proximity to the GPU compute die within a single package, delivering very high memory bandwidth at the cost of significant power consumption and extremely high manufacturing cost.

Qualcomm’s HBC takes a fundamentally different architectural approach: rather than stacking memory near the compute die, it stacks the compute units directly beneath the memory itself using 3D integration, and uses LPDDR, the low-power memory technology refined over decades in smartphones, rather than the power-hungry DRAM stacks HBM requires. With this approach, Qualcomm claims 6x higher bandwidth-per-watt than HBM and 200x higher capacity-per-watt than SRAM, so a huge improvement.

The core philosophical difference is that HBM optimizes for peak bandwidth regardless of power cost while HBC optimizes for bandwidth-per-watt and capacity-per-watt, which are the binding constraints for inference workloads where electricity cost, not peak speed, determines total cost of ownership. HBC is designed specifically to address this constraint. Commercial sampling of HBC Gen 1 with AI250 is expected to begin in mid-2027.

So, with this data center portfolio, Qualcomm is aiming at the inference market, which seems to be where the market is heading. Training happens once per model version; inference happens continuously, at scale, forever, and as hyperscalers run large language models around the clock to power AI assistants, coding tools, and agentic workflows, the cost of electricity per useful token generated becomes the dominant economic variable, which is precisely the constraint Qualcomm’s HBC memory architecture and Dragonfly inference accelerators are engineered to minimize. This positions Qualcomm’s inference-first architecture as a natural complement to rather than a replacement for Nvidia’s training dominance, and gives hyperscalers a compelling economic reason to diversify their silicon mix toward Qualcomm’s efficiency profile as inference workloads come to dominate their total compute spend.

Crucially, initial commercial success looks highly promising.

Microsoft confirmed the deployment of Qualcomm’s HBC architecture on Azure.
Meta has signed a multi-generation agreement to deploy the C1000 in its server infrastructure as part of a multi-generation partnership.
Two custom silicon deals are signed for $1+ billion, starting late 2026. Reportedly, ByteDance is one of these customers.

Hyperscalers at that scale do not sign multi-generation infrastructure agreements on architectures they consider unserious, so the commitments are meaningful validation even before a chip ships.

Qualcomm sees a clear runway for $15 billion in data center revenue by 2029, so the revenue potential here is huge, and that is a completely new revenue stream layered on top of the existing business, completely incremental.

That rounds up the QCT segment, hopefully giving a pretty good overview of Qualcomm’s chipset product portfolio and competitive position.

What is left is the second engine, the QTL (Qualcomm Technology Licensing) segment, one of the most durable and capital-light revenue streams in all of technology.

This one is pretty straightforward – this is a royalty machine that collects fees from virtually every manufacturer selling devices over a 3G, 4G, or 5G network, by virtue of Qualcomm’s foundational patent portfolio in wireless communications standards.

You see, back in the 1980s and 1990s, Qualcomm’s engineers made foundational contributions to CDMA, the underlying transmission technology that made 3G possible and whose architectural DNA carried forward into 4G and 5G. Rather than simply selling chips, Qualcomm made a strategic decision to actively participate in global wireless standards bodies and ensure its inventions became embedded in the agreed-upon technical standards that every manufacturer must implement to build a device capable of connecting to a cellular network.

Once a patent becomes “standard-essential”, meaning it is impossible to build a standards-compliant device without using it, the patent holder holds structural leverage over the entire industry. That is the foundation on which QTL was built.

QTL grants licenses to use portions of Qualcomm’s intellectual property portfolio, which includes patent rights essential to and useful in the manufacture, sale, and use of wireless products implementing LTE and OFDMA-based 5G standards and their derivatives, on a worldwide basis. In practical terms, this means that virtually every smartphone manufacturer on earth, whether they use a Qualcomm chip or a MediaTek chip or a Samsung Exynos chip, must pay Qualcomm a royalty to legally sell that device.

QTL licensing revenues consist primarily of per-unit royalties, with licensees paying quarterly royalties based on their sales of products that incorporate or use Qualcomm’s licensed intellectual property. These royalties are generally calculated as a percentage of the wholesale selling price of complete licensed products, net of certain permissible deductions, with certain products subject to per-unit minimums and caps.

Qualcomm has achieved over 90% penetration in global 5G smartphone licensing, meaning the vast majority of 5G devices sold worldwide generate a royalty stream. The revenue is largely locked in by multi-year licensing agreements with OEMs, making QTL far less cyclical than QCT.

Also, this is incredibly high margin, with EBT margins in the mid-70s range, simply because it carries almost no cost of goods sold, no manufacturing, no supply chain, and minimal incremental cost to add a new licensee. As a result, 10-12% of revenue brings in nearly 50% of profits.

However, QTL’s revenue trajectory is stable rather than exciting, as the global smartphone market is mature, royalties are mostly contracted, and there is no imminent step-change in the base from 5G adoption since that wave has largely already run.

The longer-term wildcard is 6G. Qualcomm is already embedding itself in the 6G standards process in exactly the same way it did with 3G, 4G, and 5G, spending R&D capital today to generate standard-essential patents tomorrow. If history rhymes, QTL will enter the next generation with the same structural toll-collection position it holds today, just on a larger addressable base.

For now, the correct way to think about QTL is as the engine that funds R&D, subsidizes chip margins, and ensures the business remains highly profitable even during semiconductor downturns.

That then mostly covers the business, which is fairly complex due to its broad product portfolio and reach across many end-markets.

Let’s move on to one of the biggest questions – what can we expect from Qualcomm in terms of growth over the next 5-10 years?

Growth potential

So, what can we expect from Qualcomm in terms of growth over the next 5-10 years? Generally, Wall Street doesn’t give it much credit due to its mature, cyclical, and slow-growing smartphone core, but I believe that view is outdated, especially as Qualcomm rapidly diversifies and as AI opens up more exciting opportunities for the handset business. Really, Qualcomm’s growth opportunity looks good.

Starting with the core handset business, indeed, this isn’t the most exciting business, and in terms of unit shipments, growth is more likely to be negative than positive through 2030 given high penetration. However, Qualcomm doesn’t need unit growth or market share gains for positive growth. Take 2026: smartphone SoC shipments are projected to decline by 7%, yet the market will likely grow in the double digits, driven by ASP expansion from premiumization, rising semiconductor content per device, and the adoption of AI features.

ASP expansion is the key dynamic to understand in the handset story. The overall smartphone market is not growing dramatically in unit terms, but it is premiumizing.

For one, consumers are trading up, holding devices longer, and demanding more silicon capability per phone, and Qualcomm is the primary beneficiary of that shift, selling more higher-end chipsets. Consumers in developed markets are holding phones longer but trading up when they do replace them, while emerging-market consumers are graduating from entry-level to mid-range and from mid-range to premium at an accelerating rate as disposable incomes rise. The global smartphone ASP crossed $400 for the first time in Q4 2025, a milestone that directly benefits Qualcomm because its revenue per device scales with device price. As long as the mix continues shifting upward, driven by AI features, camera improvements, and consumers valuing longevity, Qualcomm’s revenue per unit grows even if total market units do not.

Additionally, the content per phone is increasing over successive generations. The most direct channel is the growing complexity and integration of the Snapdragon SoC itself. Each new generation adds capabilities, such as a more powerful NPU, a more advanced modem, more sophisticated image processing, and better Wi-Fi generation, which justifies a higher chip price and a larger proportion of the device’s bill of materials.

A shift to on-device AI can also be a significant contributor to this higher ASP dynamic.

You see, for the past several years, the dominant assumption in the technology industry has been that AI is a cloud business. You ask ChatGPT a question, your request travels to a data center, a massive GPU cluster processes it, and the answer comes back to your device. The device itself is essentially a dumb terminal. The intelligence lives in the cloud. Under that model, the critical hardware is Nvidia’s H100s in hyperscaler data centers, and the critical infrastructure companies are the cloud providers.

However, this might shift over time. As AI models become smaller and more efficient through techniques like quantization and distillation, and as on-device silicon becomes more powerful, an increasing share of AI inference will migrate from the cloud to the edge. Why? There are four main reasons:

The first is latency. A cloud round-trip takes hundreds of milliseconds. An on-device inference takes milliseconds. Not important today, but it will be as adoption grows.
The second is privacy. Processing data locally means sensitive information never leaves the device. This is increasingly important to both consumers and regulators, and it is a structural advantage that cloud AI cannot replicate by definition.
The third is cost. Every inference query sent to a cloud data center incurs costs for compute, bandwidth, and energy. As AI becomes embedded in billions of daily micro-interactions, the economics of routing everything through the cloud becomes untenable. On-device inference is effectively free at marginal cost once the chip is sold.
The fourth is connectivity independence. A device that runs AI locally works on a plane, in a tunnel, in a rural area, or anywhere else with poor connectivity. Cloud AI requires a reliable internet connection. For certain applications, that is critical, like ADAS.

Obviously, this creates a huge opportunity for Qualcomm. This means smartphones (and other connected devices) are potentially becoming the primary compute platform for agentic AI tasks, making the Hexagon NPU inside each Snapdragon platform increasingly central to the purchase decision. We already see this with ByteDance Doubao-powered phones and Xiaomi’s Miclaw agent framework already shipping on Snapdragon silicon.

And each additional leap in on-device AI capability justifies a new design cycle and, importantly, a higher chip price, hence potentially significantly contributing to higher ASPs.

These mechanisms can collectively still drive mid- to high-single-digit revenue growth from a flat unit base. Market Research firms point to a 7-9% CAGR for the smartphone SoC market, which should, at the very least, allow Qualcomm to grow its smartphone business at a mid-single-digit CAGR, taking into account growing competition in the high-end segment.

Indeed, no exciting growth, but a pretty solid cash-generating core.

The more exciting growth is likely to come from IoT, with a strong outlook for the AI PC, XR, and robotics – early-stage markets with huge runways.

Starting with AI PCs, market research firms point to a 19% CAGR, while Tech Insights expects a 42% CAGR for the AI PC processor market through 2029. More importantly for Qualcomm, Gartner expects AI PCs to become the standard form factor by 2029, accounting for more than half of global PC shipments, while IDC projects the installed base of AI PCs jumping from 5% in 2023 to 94% worldwide by 2028. Within that market, ARM-based laptops are anticipated to capture a larger share of the consumer segment, in which Qualcomm is a leading player.

We should add, however, that Qualcomm’s Windows on ARM exclusivity has expired after 8 years of being the exclusive provider for Windows on ARM. This has opened the door for Nvidia, AMD, and MediaTek to enter the market with ARM-based processors, so competition will ramp up significantly in the next few years, making further market share gains harder to achieve.

Nonetheless, Qualcomm aims for $4 billion in PC revenue by 2029, which doesn’t look like a long shot, implying only a 10-12% market share up from its current 9%.

Looking at the XR market, the setup looks even more promising, though growing from a small base. This market is expected to compound at a 40%+ CAGR through 2030, with the smart glasses market growing at a 24% CAGR through 2033. With its monopoly in this market, Qualcomm is poised to significantly benefit from rapidly ramping volumes, expected to take off from 2027 onward as more models are launched.

And then there is robotics, which is the earliest-stage market but also the largest in potential. Qualcomm’s early positioning with humanoid developers like Figure, combined with the Dragonwing IQ10’s purpose-built compute capability for humanoid applications, mirrors the strategy it used in automotive — establish the platform relationship early, before volumes justify the design investment from the developer’s perspective, and then capture the revenue as production scales. The automotive analogy matters here: Qualcomm entered the market in 2002 with telematics, spent 15 years building relationships, and now has a $65 billion pipeline. The robotics relationship timeline is likely similar.

The revenue contribution from robotics will likely be near zero through 2030, but might ramp up thereafter.

The aggregate picture is of an IoT segment that could realistically double or triple from its current $6.6 billion annual revenue to the minimal $14 billion management targets by fiscal 2029, without issues. I believe management’s 20% CAGR through 2029 is on the conservative side, given current market forecasts.

Moving to automotive, I am even more bullish.

The automotive SoC market, the one most directly relevant to Qualcomm, is forecasted to grow at an 8-10% CAGR, driven entirely by exploding content per vehicle. For reference, S&P Mobility forecasts that average semiconductor content per vehicle could grow from around $1,000 today to over $1,400 by 2030, with ADAS semiconductor content per vehicle alone growing from $160 today to over $260 by 2030. The EV transition amplifies this further: the average semiconductor cost per ICE car is $500-600, while for EVs it could surpass $1,500 by 2025 and $2,000 by 2030, due to power modules, sensors, and AI processors.

As a result, a world that sells the same number of cars but transitions from ICE to EV needs dramatically more silicon per vehicle, and that secular shift has a decade or more to run.

Meanwhile, Qualcomm is in an excellent position to outpace the market, with exposure to faster-growing verticals – high-end SoCs – and Qualcomm’s integrated solution approach seeing strong adoption, allowing it to grow its content per vehicle even faster than the competition. This alone should lift its CAGR to over 15%, according to management.

And then we have yet to factor in market share gains and a huge design-win pipeline, giving management visibility into a CAGR closer to 26% through 2029, which is absolutely impressive.

That then leaves me to address one of Qualcomm’s proven, most underappreciated, yet most powerful growth levers, which comes down to continuous growth in its TAM through new market entries and brilliant use of its R&D, leveraged across multiple markets, better framed as its multi-market cross-subsidization advantage.

The elegance of Qualcomm’s diversification and growth strategy is that it is not really diversification at all; it is the same core competency being applied to successive waves of device categories that require intelligent wireless compute, rapidly expanding the TAM.

Crucial in this is Qualcomm’s favorable positioning in power-constrained compute and connectivity. Smartphones required it first: a platform where processing power, cellular connectivity, and battery life had to coexist. Then came automotive: vehicles are becoming wireless-connected AI compute platforms on wheels, requiring powerful ADAS compute, 5G connectivity, and V2X communication within the thermal and reliability constraints of an automotive environment. Next up are smart glasses and XR: the smallest and most power-constrained form factor yet — always-on AI compute, computer vision, and wireless connectivity in a device that must weigh under 50 grams and run all day on a battery the size of a pencil eraser – this once again plays into Qualcomm’s core strengths. The same goes for robotics and industrial IoT: robots, autonomous vehicles, and industrial machines all need local AI compute for perception and decision-making, 5G connectivity for fleet management and remote operation, and power efficiency for battery-operated mobile platforms. And now, increasingly, datacenters: as the market shifts to inference, memory bandwidth per watt and compute efficiency per watt matter more than raw TOPS for the inference workloads that dominate real-world AI deployment.

In other words, Qualcomm’s foundational competency, the thing it does better than anyone else on earth and that underpins every market it operates in, is designing silicon that delivers maximum AI compute performance within minimum power consumption, combined with best-in-class wireless connectivity, in a single integrated platform. Every product Qualcomm makes is an expression of this same capability applied to a different device category.

So, when Qualcomm enters a new market, it is not starting from scratch in engineering. It is taking an already mature core technology platform and applying a comparatively modest amount of incremental engineering to adapt that platform’s form factor, power envelope, and feature set to the new market’s specific requirements. This is fundamentally different from a true greenfield entrant, who must build the core compute architecture, the connectivity stack, and the software ecosystem all from scratch before they can even begin competing on market-specific features.

The PC market entry illustrates this well. When Qualcomm launched the Snapdragon X in 2023, it was not inventing a new CPU architecture or an NPU from scratch. Instead, it was taking the Oryon CPU core (acquired through the Nuvia team), the Hexagon NPU lineage already mature from years of smartphone deployment, and re-engineering the power management, I/O, and software compatibility layer for the Windows laptop form factor. The result was a chip that immediately competed with Intel and AMD’s NPU performance despite Qualcomm having effectively zero prior Windows PC market share, because the hardest engineering problems — efficient AI compute architecture and power management discipline — were already solved.

This also means the $9 billion in annual R&D investment in cellular modem technology, NPU architecture, and RF front-end components is shared across smartphones, automotive, IoT, PCs, and data centers simultaneously. A modem breakthrough developed for the smartphone market gets deployed into automotive connectivity at near-zero incremental R&D cost. NPU efficiency improvements developed for handsets flow directly into the IoT and PC platforms. This cross-market leverage means Qualcomm’s effective R&D return per dollar spent is structurally higher than that of more focused competitors, and it is one reason the company can pursue automotive, IoT, PC, XR, robotics, and data center simultaneously without the R&D budget spiraling out of control.

Also, this makes it loads easier to remain technologically dominant – Qualcomm’s effective return per dollar of R&D spent is structurally higher than that of more focused competitors, and no competitor entering any single one of those markets brings the same breadth of foundational technology.

It will likely maintain this trajectory, eagerly entering highly promising markets with its superior technologies, taking robotics as a recent example of such an early entry. This will continue to be a growth driver in the future, but one that is impossible to quantify, though something to account for in perpetual growth assumptions.

The best recent example of this dynamic and its effects is Qualcomm’s datacenter market entry today, which uses the exact same logic – it is not starting from zero; it is redeploying capabilities built for entirely different purposes that happen to map unusually well onto what modern AI data centers need, creating huge incremental revenue potential.

An immediate competitive product allows Qualcomm to rapidly gain market share in a huge, expanding market, where Qualcomm’s core specialties are becoming increasingly important. As a result, management believes it can capture 5% of this TAM by 2029, which is a $15+ billion incremental revenue opportunity.

That is huge.

All things considered, Qualcomm seems well-positioned to deliver fairly strong growth through 2030, with a double-digit growth rate absolutely possible, considering slower handset revenue growth should be more than offset by strong growth in automotive and IoT verticals, especially as this grows to a higher percentage of the total and considering ramping datacenter revenues toward the end of the decade.

Furthermore, there is a clear runway to maintain this well into the next decade, given the growing importance of its core competences and the early-stage nature of some of the markets where it is becoming a dominant force.

Really, I am quite optimistic.

But there is a critical near-term headwind that needs to be addressed, which is likely to be a significant drag on the result in 2026 and 2027 and needs to be considered when discussing growth prospects.

The Apple headwind

As noted earlier, a major revenue source for Qualcomm is supplying modems to the iPhone. How big is this revenue source? As of 2025, this is estimated to generate roughly $5.7- $ 5.9 billion annually, with an additional $1.6-$1.9 billion from RF components and other subsystems. That puts Apple’s total chip-related revenue at roughly $7.5 billion annually at its peak, representing close to 17% of Qualcomm’s fiscal 2025 revenue of $44.3 billion.

This makes Apple Qualcomm’s largest customer. However, the big issue here is that Apple has been working on its own in-house modem for years, and it is currently winding down its Qualcomm modem sales. For reference, while 70% of iPhones used a Qualcomm modem in 2025, this will drop to just 20% in 2026, moving to zero by 2027. CEO Amon has been admirably transparent about this: “We’re planning our business assuming that they are going to use their own modem.”

In other words, starting in 2026 and 2027, Qualcomm will gradually lose 17% of its revenue as Apple moves its modem business in-house, a significant loss that will create a nearly $7 billion revenue hole. And given that this business, by estimates, carries above-average margins, the earnings impact is proportionally even larger than the revenue impact suggests.

This means Qualcomm’s results in 2026 and 2027 will be significantly dragged down.

Obviously, that is not great for Qualcomm in the near-term. However, this Apple headwind has been a huge overhang on the stock for years, depressing the multiple that investors are willing to pay. Positively, with the overhang now finally removed from 2028 onward, I actually think Qualcomm is more favorably positioned, poised to rerate, with investors able to value its other business lines more realistically. So, while a headwind in the near term, I am actually positive about this development in the long run. Simply put, investors can quantify the headwind now and estimate cleaner financials by 2028.

What is also worth pointing out is that Qualcomm collects royalties from Apple under a six-year licensing agreement (with a two-year renewal option) struck in 2019 following their legal settlement, estimated at just over $2 billion annually. What happens after that is one of the genuine open questions in the Qualcomm story. Apple cannot simply stop paying QTL royalties by switching to its own modem, as it still sells iPhones that operate on 5G networks that rely on Qualcomm’s standard-essential patents, which are inescapable regardless of whose chip is inside. Any renegotiation or litigation would be a replay of the bruising 2017-2019 dispute, which Apple ultimately lost. Most analysts assume a new licensing agreement will be reached, though potentially at a renegotiated rate.

In general, even beyond 2027, customer concentration remains a focus point for Qualcomm investors. Apple, Samsung, and Xiaomi each individually contributed more than 10% of Qualcomm’s consolidated revenue in fiscal 2025. Taken together, these three customers account for a very large proportion of total revenue, likely in the 40-50% range when their combined chip and licensing payments are aggregated, which is a direct result of the concentration of the smartphone industry.

Growth in automotive and IoT should reduce this concentration significantly by 2029, but it will remain relatively high, with the impact of Xiaomi or Samsung moving away from Qualcomm still significant.

There is also above-average China exposure for Qualcomm, at roughly 63% of revenue, mainly due to its strong position in the Chinese smartphone and automotive markets, plus this is where a lot of manufacturing of devices for Western markets occurs, automatically raising Qualcomm’s China exposure, even if the product go to Europe and the U.S. Nonetheless, it does create heightened geopolitical sensitivity that is structural rather than discretionary. This is unlikely to go anywhere.

With all of that in mind, let’s jump into Qualcomm’s recent performance and financials!

Financial & Performance Review

Qualcomm released its latest financial results in late April – its fiscal Q2 2026 – and delivered a decent report that exceeded expectations, even as the business faces significant headwinds that have led to an objectively somewhat lackluster performance. Let’s break down the numbers.

Qualcomm reported total fiscal Q2 revenue of $10.6 billion, in line with consensus estimates and down 2% YoY, amid expected headwinds. You see, while the business has strongly recovered from a cyclical downturn in 2023 and has shown strong numbers throughout 2024 and 2025, momentum has decelerated sharply so far in 2026, turning negative again in Q2.

Crucially, this isn’t company-specific, but a broader weakness in the handset market related to explosive memory prices.

Q2 handset revenue was $6 billion, down 13% YoY, as expected, with growth turning negative after strong results in recent years. You see, the explosive demand for high-end memory in datacenters and the subsequent memory shortage have led to insane price hikes by memory producers. This then forces smartphone manufacturers to raise their prices to offset the cost of expensive memory, leading to uncertainty in the sector, as higher prices generally go hand in hand with lower volumes, as consumers are less eager to buy and postpone purchases.

Therefore, handset OEMs, particularly in China, are taking a cautious approach by reducing build plans and drawing down channel inventory to avoid a buildup if demand crashes. Simply put, higher memory prices lead to higher smartphone costs (remember, Apple recently raised its prices for this reason), which tends to reduce volumes, prompting smartphone manufacturers to reduce inventory, which means cutting output and undershipping current demand. And this leads to fewer orders for Qualcomm SoCs.

So, actual smartphone sales are holding up well, with volumes flat YoY, but manufacturers are already anticipating falling demand, and this is hurting Qualcomm.

This isn’t unexpected, but it is a serious near-term headwind that could persist for quite a long time, with this memory shortage indicated to last through 2027 and into 2028.

Positively, inventories were already quite low, and with these further reduced right now, Qualcomm is already taking the pain today, but should see a swift recovery, with management already pointing to a bottom in fiscal Q3, and a gradual sequential recovery after that.

What should further help offset this headwind in the second half of the year and into 2027 is growing agentic demand pushing consumers to the premium tier. Take the ByteDance/Doubao-powered agentic AI phone from ZTE Nubia, Xiaomi’s recent announcement of a Miclaw agent framework, and other agentic systems now in development across the Android ecosystem. These kinds of AI advancements all need Qualcomm chips and are driving an AI upgrade cycle in the coming years, which should offset memory headwinds.

So, already depleted inventory and an AI-driven upgrade cycle should allow Qualcomm to quickly recover and return to positive growth in fiscal 2027. Sure, high memory prices are a serious headwind, but Qualcomm is well positioned to navigate them. Also, the company’s current deal with Samsung means it takes a share greater than 70% vs Samsung’s in-house silicon, up from a historical 50%, which is a tailwind as well.

Ultimately, short-term handset revenues will be depressed by external headwinds, but this should not be prolonged, at least not to this degree.

Moving to automotive, the story is the complete opposite, with revenue accelerating strongly from recent quarters. Automotive revenue in Q2 hit $1.4 billion, up 38% YoY, more than double the 15% it reported in Q1 or the 17% in Q4, despite lapping a huge 59% growth last year, showing very strong underlying momentum. Also, this means automotive now has a $5 billion annual run rate, expected to hit $6 billion by the end of the year.

Growth here is driven by accelerating demand and increasing content per vehicle. Specifically, the transition to Qualcomm’s fourth-generation digital cockpit and ADAS systems is driving growth. It is the largest generation-to-generation content increase in Qualcomm’s history, delivering 3x higher CPU throughput, a threefold increase in GPU capability, and 12x higher NPU performance while supporting in-vehicle agents and processing for Level 3 and Level 4 autonomous driving, making it significantly more future-proof.

This is driving very strong momentum in this business, and Qualcomm expects to continue gaining share with its best-in-class, most complete solution, as evidenced by recent design wins. Meanwhile, these new generations of SoCs significantly increase content per vehicle, driving growth, especially amid strong ADAS adoption.

This segment is really firing on all cylinders, with 23 consecutive quarters of double-digit growth and a $65 billion design win pipeline.

At the same time, IoT is also performing well, delivering $1.7 billion in Q2 revenue, up 9% YoY and fairly stable, driven by growth across consumer and industrial products. Within IoT, Qualcomm is seeing AI drive a promising product renewal cycle and a healthy product pipeline, helping sustain growth. For example, Qualcomm expects a wave of smartglasses to be released in the second half of the year, all running on Qualcomm solutions, which should boost consumer demand.

IoT and automotive, combined, delivered 20% YoY growth, underscoring continued diversification of the business and strong overall performance. This is in line with long-term targets.

Ultimately, the strong performance in IoT and automotive, offset by the temporary decline in handset revenue, led to a total QCT revenue of $9.1 billion, down 4% YoY due to the handset business’s continued weight.

Rounding out the top-line performance, QTL (licensing) revenues hit $1.4 billion in Q2, up 5% YoY and at the high end of guidance, driven by a favorable mix, with global handset units approximately flat YoY.

Moving to the P&L, Qualcomm has been delivering solid numbers since 2020, despite a dip in 2023. OpEx has consistently grown more slowly than revenue in recent years, registering a 6% CAGR between 2020 and 2025, compared to a 15% revenue CAGR, driving solid operating leverage.

However, with investments in growth and innovation steady and revenue slightly down in Q2 due to the memory headwinds, Qualcomm’s margins also dipped. For reference, while revenue dipped 2%, both R&D expenses and SG&A were up 11% YoY, putting some pressure on margins.

The EBT margin, Qualcomm’s preferred metric, was down 350 bps YoY to 30.6%, with a 200 bps improvement in QTL more than offset by a 300 bps decline in QCT, resulting in EBT dollars down 12% YoY to $3.25 billion.

Positively, despite near-term headwinds, Qualcomm’s long-term trajectory is strong, with the company consistently realizing solid operating leverage through the cycles, translating into higher margins. Looking ahead, management believes it can reduce OpEx as a percentage of revenue from roughly 23% in 2026 to sub-20% by 2029. So, we can expect Qualcomm to keep delivering expanding margins.

Further down the line, Qualcomm reported a Q2 net income margin of 26.8%, also down 250 bps YoY, with a lower share count offsetting some of the lower EBT margin. This translated into an EPS of $2.65, at the high end of guidance and beating the consensus by $0.08, though still down 7% YoY amid a decline in revenue and some margin pressure.

Finally, Qualcomm delivered a Q2 FCF of $1.9 billion, down 18% YoY and reflecting an 18% FCF margin. Looking at the more representative TTM numbers, Qualcomm generated a solid $12.5 billion at a 28.1% margin, slightly down from 28.9% in fiscal FY25, but solid nonetheless.

These excellent cash flows allowed Qualcomm to also maintain a solid balance sheet, ending the quarter with total cash and short-term investments worth $9.8 billion and a total debt of $15.3 billion, leaving the company with a net debt of roughly $5.5 billion, which is healthy enough given the nature of the business and its strong cash flows.

This also allows Qualcomm to keep opportunistically returning cash to shareholders, which has actually accelerated in recent quarters. In Q2 alone, it returned $3.7 billion, including $2.8 billion in buybacks and $945 million in dividends. Regarding the dividend, management raised this by 3.4% recently (along with a new $20 billion buyback program), its 23rd consecutive increase, with shares now yielding just over 2% based on a sub-30% payout ratio, which is excellent – investors get a very solid yield that remains extremely well covered and has loads of room for growth.

To round off this section, it is worth highlighting Qualcomm’s excellent reinvestment metrics, which are a clear proof of management’s ability to generate shareholder value through strong returns on deployed cash. For reference, Return on Invested Capital (ROIC) and Return on Equity (ROE) are among the most important metrics an investor can use to evaluate a business’s true quality, because unlike revenue growth or margin figures — which can be manufactured through accounting choices or one-time events — ROIC and ROE measure whether management is actually creating value with the capital shareholders have entrusted to them. ROIC specifically answers the most fundamental question in business: for every dollar deployed in the company’s operations, how many dollars of operating profit does it generate? A business consistently earning ROIC above its weighted average cost of capital is compounding value, and that includes Qualcomm.

Qualcomm’s TTM ROIC is an excellent 27%, and its ROE is 36% - those are excellent numbers well ahead of the cost of capital, clearly indicating Qualcomm knows how to consistently generate shareholder value.

Outlook & Valuation

As always, let’s start with management’s guidance. For Q3, management now expects to deliver revenue of $9.2 billion to $10 billion, down 8% YoY. This includes a QTL revenue of $1.15 to $1.35, reflecting a sequential decline primarily due to the operating assumption of weaker low-tier handset units. Meanwhile, QCT revenue is expected to be $7.9 billion to $8.5 billion, including handset revenue of $4.9 billion (reflecting the industry-wide headwind from high memory prices), IoT revenue growing in the high single digits, and automotive up roughly 50% YoY, accelerating further.

On the bottom line, management guides for an EPS of $2.10 to $2.30, down roughly 21% at the midpoint, reflecting the revenue loss and continued investments, though this should already be the bottom.

Looking at longer-term guidance, management updated its 2029 targets during the 2026 Analyst Day.

Management now expects automotive revenue to grow at a 26% CAGR to $10 billion by 2029, up from $8 billion previously, backed by a $65 billion design win pipeline.
IoT revenue is targeted to hit $14+ billion by fiscal FY29, reflecting a 20% CAGR.
Data center revenue to grow from zero today to $15+ billion by fiscal FY29

This means Qualcomm expects non-handset revenue to exceed $40 billion by fiscal 2029, reflecting a 40% CAGR, which is stellar. At the same time, management expects its handset revenue to grow at a 5%+ CAGR from a 2026 base, likely accounting for less than a third of fiscal 2029 revenue. Above all, this indicates a much improved revenue mix, leaning toward higher-growth verticals and with far less customer concentration than inherently came with the handset business.

Finally, management sees ample room for margin expansion, especially as growth is expected to be strong. Driven by this, management expects a minimal fiscal 2029 EPS of $18, reflecting an 11% CAGR from its 2025 base.

Overall, that is a very impressive outlook.

With that, let me move to my own financial framework. Starting with fiscal 2026, I am cautious given the memory crunch going on, which will continue to pressure handset volumes, dragging on Qualcomm’s performance. Yes, we should see a sequential improvement in Q4, but this is also where the Apple step-down will happen with the launch of the iPhone 18, so I think handset performance will remain pressured. Strength in automotive and IoT will offset some of this, but overall growth will still be negative in my view. And with the high-margin licensing business also facing lower volumes and investments unbothered, margins will be hit even harder. In my view, a 4% drop in revenue and a 10% drop in EPS are realistic for 2026.

For 2027, the Apple modem phase-out will remain a headwind, as the last 20% will be lost, with revenue estimated at roughly $2 billion. At the same time, I expect handset ex-Apple revenue momentum to improve, driven by an AI-driven upgrade cycle taking hold and continued premiumization, likely enough to drive mid-to-high single-digit growth, so I expect total handset revenue to be about flat. Licensing will likely be slightly lower YoY, but I expect automotive momentum to remain very strong and IoT to grow in the double digits as well, resulting in an estimated 5% YoY growth. EPS is likely to be similar.

Looking further ahead, I expect a sharp improvement in performance from 2028 onward, with the Apple headwind gone, handset momentum improving, and the faster-growing IoT and automotive segments becoming a larger share of total revenue. Especially with the Apple drag gone and tailwinds kicking in, I expect Qualcomm to deliver strongly accelerating growth into the high teens to mid-twenties through 2030. On the bottom line, I expect a slight margin improvement, driven by operating leverage from accelerating growth, offset by a larger share of lower-margin revenue. Nonetheless, this should allow for EPS to outpace revenue by a few percentage points.

These expectations are reflected in the financial model below. With this, I am taking a slightly cautious approach to management’s guidance, which points to FY29 revenue of over $70 billion.

That brings me to valuation, and with Qualcomm shares having sharply pulled back from recent highs, losing 30% since late May, these look quite attractively valued. At a current price of $176, shares trade at:

16x 2026 earnings
A 0.9x PEG
16x 2026 FCF (consensus)

There is just no way around the fact that these are highly compelling multiples for a business this well-positioned, this dominant, with excellent financials, and a seriously compelling outlook. Yes, current multiples reflect a roughly 7% premium to historical averages, but Qualcomm is looking probably the best it has in over a decade.

In my view, the key here is that Qualcomm will be fundamentally better positioned by 2028/29 with the Apple headwind gone and non-handset revenue accounting for the majority, translating into a structurally faster-growing and more promising business with excellent, superior technology, now dominating much more attractive end markets.

Yet, I don’t think the market is currently pricing this in at all, being very shortsighted. Qualcomm isn’t a stock for the next two years, where it’s losing the Apple modem business and faces headwinds from elevated memory prices; the Qualcomm appeal sits in the post-2027 period. The thesis here is the re-pricing once the business structurally improves, layered on top of a much more attractive growth outlook given Qualcomm’s dominance in early-stage, high-growth markets.

Qualcomm becomes significantly more attractive once dependence on Apple is gone and handsets are no longer the main business – the factors that have historically pressured the multiple investors were willing to pay.

Automotive at $10 billion carries entirely different quality characteristics: contractually anchored design wins, five- to seven-year production revenue visibility, no single customer above 10%, and a market growing at 8-11% annually, driven by non-discretionary regulatory mandates. IoT at $14 billion spans a dozen device categories across consumer, industrial, and enterprise segments, with no meaningful customer concentration, and growth is driven by the secular on-device AI adoption wave. Data center, even at a conservative fraction of the $15 billion target, adds a high-margin, high-growth revenue layer with hyperscaler anchor customers.

Plus, Qualcomm has a proven ability to enter promising markets with superior designs and compete within a few years, giving even more potential upside to the current TAM.

And all of this comes with excellent financials, significant room to improve margins, strong cash generation, strong reinvestment metrics, and a healthy balance sheet.

The re-rating potential is huge. But let’s be conservative (on top of my conservative revenue forecast) and assume a re-rating to 20x earnings by fiscal 2028, which still leaves loads of upside while adding a sufficient margin of safety, justified given uncertainty around the ramp of the data center business. Using this multiple and my current 2028 projection, I calculate a $277 end-of-fiscal-year 2028 price target, which implies annualized returns of roughly 22% (including dividends)

That easily clears my 15-20% threshold – in my book, that is an excellent risk-reward ratio, especially given the cautious approach to the calculation. In other words, I believe Qualcomm shares are excellent value right here, really anywhere below $190, with huge potential upside and plenty of downside protection.

It is a more complicated story than others out there, but that is exactly the opportunity at hand.

Rating + fair value: Buy - Accumulate below $190
2028 Target Price: $277
Implied CAGR from current price: ~22%

Please leave a like if you enjoyed this post and found it valuable!

Disclosure: I/we may or may not have a beneficial long position in any of the securities discussed in this post, either through stock ownership, options, or other derivatives. This article expresses our own opinions, and we are not receiving any sort of compensation for it.

No recommendation or advice is being given as to whether any investment is suitable for a particular investor. The information provided in this analysis is for educational and informational purposes only. It is not intended as and should not be considered investment advice or a recommendation to buy or sell any security.

Investing in stocks and securities involves risks, and past performance is not indicative of future results. Readers are advised to conduct their own research before making any investment decisions.

The Soji Brief

Jul 5

28% free cash flow margin next to basically 0% recent revenue growth is the part that jumps out. A business that mediocre-sounding on the top line but still throwing off that much cash usually means the market is pricing the old story (phone chips, Apple walking away) and ignoring what’s actually showing up in the cash flow statement. The Meta CPU deal and the Hugging Face partnership are the first real proof that the diversification thesis isn’t just a slide in an investor deck. Our screener still has it 44% below fair value even with the growth concern baked in.