Loading analysis...
DD on QCOM Being Undervalued and Its Automotive/Data Center Transformation
Show original post
Hi, this is my 8th grade research project on Qualcomm. All figures based on FY2025 financials, Q2 FY2026 earnings/transcript, and TTM data. I know, I know "WSB is a casino - put the fries in the bag". But I need someone to rip thesis to shreds if I'm off.
Qualcomm runs two segments:
- QCT (Qualcomm CDMA Technologies) - the chip division. Designs and sells Snapdragon SoCs for smartphones, automotive, IoT, and increasingly data center. 87% of revenue ($38.4B in FY2025).
- QTL (Qualcomm Technology Licensing) - licenses QCOM's patent portfolio to every manufacturer selling a 3G/4G/5G device on the planet. 13% of revenue ($5.6B) but prints ~72% EBIT margins with minimal capital requirements. It's essentially a toll booth on the global handset market.
The Setup
QCOM trades at 17x forward earnings in a semiconductor peer group with a median closer to 35x. That discount exists for two reasons:
- China exposure. Market is worried about tomfoolery around export restrictions and tariffs while China represents ~46% of revenue.
| Region | Revenue FY2025 | % Total | YoY |
|---|---|---|---|
| China | $20.3B | 46% | +14% |
| US | $10.5B | 24% | +9% |
| Korea | $9.5B | 22% | +19% |
- Apple manufacturing and using its own modem chips for iPhone after using QCOM's since iPhone's release over disputes and lawsuits for the last decade that QCOM charged too much. Link.
What the market has underpriced is that both headwinds are well-understood, the near-term pain is timing not structure, and two genuine growth vectors - automotive and data center - are accelerating simultaneously.
Financial History: Recovery From a Brutal Cycle
| FY | Revenue | Net Income | FCF | EPS (GAAP) |
|---|---|---|---|---|
| 2022 | $44.2B | $12.9B | $6.8B | $11.37 |
| 2023 | $35.8B | $7.2B | $9.8B | $6.42 |
| 2024 | $39.0B | $10.1B | $11.2B | $8.97 |
| 2025 | $44.3B | $5.5B | $12.8B | $5.01 |
FY2023 was a post-COVID semiconductor hangover - smartphone demand collapsed, revenue fell 19%. The recovery has been clean: FY2025 revenue matched the FY2022 peak at $44B+, and FCF hit a record $12.8B.
The asterisk on FY2025 earnings is important. Reported net income of $5.5B dramatically understates the business. Operating income was $12.4B - the gap is a $6.1B one-time tax charge in Q4 FY2025 from IRS treatment of capitalized R&D expenses. Q2 FY2026 saw a mirror-image $5.7B non-cash tax benefit for the same reason. Both are excluded from non-GAAP. The operational business runs at roughly $12B annual operating profit and $12.8B FCF. Judge it on those.
Margins Tell the Real Story
On a TTM basis:
| Metric | Value |
|---|---|
| Gross Margin | 54.8% |
| Operating Margin | 25.5% |
| Net Margin (GAAP) | 22.3% |
| FCF Margin | 18.0% |
| ROE | 36.4% |
| ROA | 17.4% |
55% gross margins and 36% ROE reflect a business with genuine pricing power - primarily from the licensing business and Snapdragon's dominant position in premium Android.
The Cheapest Quality Name in Semis
| Metric | QCOM (TTM) | QCOM (Fwd) | Peer Median (Fwd) |
|---|---|---|---|
| PE | 19.8x | 17x | ~35x |
| EV/EBITDA | 18.6x | - | ~39x (TTM) |
| P/FCF | 24.4x | - | ~118x (TTM) |
| Div. Yield | 1.0% | - | ~0.3% (TTM) |
The forward PE of 17x uses consensus FY2026 EPS of $10.73 (non-GAAP, adjusted) against $182 share price. For context, NVDA trades at 28x forward on 75% expected revenue growth. ADI at 35x, TXN at 37x, AVGO at 38x - all growing modestly. AMD at 52x. MPWR at 66x
QCOM at 17x is being priced for a structurally impaired business. The data doesn't support that.
The Two Known Headwinds (And Why They're Bounded)
1. Apple Modem Transition
Apple launched the iPhone 16e in early 2025 with its in-house modem, ending QCOM's monopoly on Apple silicon (and launched iPhone Air with new gen C1X modem). The company has a supply agreement through the current year at ~20% share of new iPhones. Beyond that, sell-side models put QCT product revenue from Apple at roughly $2B in FY2027 - down from a higher base but already widely reflected in consensus estimates. The QTL royalty stream (Apple pays to use QCOM's wireless patents regardless of whose modem is in the phone) is a separate negotiation and remains intact at a similar scale pending renegotiation.
The bottom line: the headwind is real, it's roughly $2-3B of QCT revenue at risk, and it's already in the estimate models.
2. China / Memory Dynamics
China is 46% of revenue - down from 62% in FY2023 but still the single biggest risk factor. The near-term pain, however, is more nuanced than simple tariff or share-loss fears.
AI data center demand for HBM memory is squeezing memory supply and raising prices. Chinese handset OEMs, facing higher component costs, are deliberately slowing builds and draining channel inventory rather than paying elevated memory prices. QCOM's chip shipments to China are significantly below actual consumer sell-through demand - the phones are still selling, OEMs are just not restocking.
Qualcomm has real-time visibility into this through its QTL licensing data (they see every phone that activates globally). Management during most recent earnings call think Q3 FY2026 as the inventory bottom with sequential growth returning in Q4. So what looks like Chinese demand dwindling very well could be a timing story and not a structural share-loss story.
What's Actually Growing
Automotive Is Underappreciated Compounding Machine
| Quarter | Auto Revenue | YoY Growth |
|---|---|---|
| Q2 FY2025 | $959M | +59% |
| Q3 FY2025 | $899M | +68% |
| Q4 FY2025 | $961M | +61% |
| Q1 FY2026 | $1.12B | +61% |
| Q2 FY2026 | $1.3B | +38% |
Annualized run rate crossed $5B in Q2 FY2026 - management guided to exit FY2026 at $6B+. Q3 FY2026 automotive is guided to grow ~50% YoY, an acceleration despite the overall revenue headwinds.
The product transition from cockpit to full digital chassis (cockpit + connectivity + ADAS + autonomy) is what's driving this. Each generation-over-generation upgrade is the largest content-per-vehicle increase in QCOM's history - 3x CPU, 3x GPU, 12x NPU performance in Gen 5 vs Gen 4. BMW ADAS is in production. Bosch and Wave just announced partnerships. The automotive design win pipeline converts to revenue 2-4 years out, which means the orders being won today show up in FY2027-2028 revenue.
At $6B+ and growing 40-50%, automotive is approaching the size of QCOM's entire licensing business.
IoT Is Getting an AI Tailwind
IoT grew 9% in Q2 FY2026, with industrial and consumer both contributing. The more interesting development: Qualcomm's IQ 10 platform (700 TOPS on-device AI, 18-core CPU) is generating design wins in robotics (Figure AI, Nura), industrial automation, and physical AI applications.
The New Catalyst Is Data Center
This is what the market isn't pricing yet. From the Q2 FY2026 earnings call:
- Custom silicon engagement with a leading hyperscaler, initial shipments December 2026
- Management described it as margin accretive and a multi-generation engagement
- Strategy is both merchant silicon (selling to all comers) and custom ASIC (bespoke chips for specific hyperscalers)
- AlphaWave acquisition adds connectivity IP and custom ASIC execution capability
- Full roadmap reveal at Investor Day, June 24
The thesis is as AI inference scales, the data center disaggregates from monolithic GPU clusters into specialized compute like Google's TPUs or Amazon's Gravitron. Qualcomm's CPU architecture (which already leads on performance/watt in mobile, PC, and auto) translates directly to data center workloads with tight energy requirements. The company has spent years building this quietly. The December shipment is the first public proof point.
None of this is in consensus forward estimates. Analysts are modeling a furthering contracting QCOM (like -10% EPS and revenue growth over the next year). Any credible data center revenue is pure upside.
Quietly Aggressive Share Buyback
In FY2025, Qualcomm returned $12.6B to shareholders on $12.8B of FCF - essentially all of it:
- $8.8B in buybacks (reducing share count from 1.14B toward ~1.07B)
- $3.8B in dividends (~1% yield)
Q2 FY2026 alone saw $3.7B returned ($2.8B buybacks + $945M dividends), described as an "acceleration" of the capital return program. The Samsung multi-year deal (>70% Snapdragon share, reaffirmed for this year and next) gives management the revenue visibility to sustain this pace.
Monte Carlo DCF: Scenario Analysis
Starting from $12.8B base FCF, 1.072B shares, $195B Market Cap ($182 share price):
| Scenario | Assumptions | P10 Mkt Cap | Median Mkt Cap | P90 Mkt Cap | P(Undervalued) |
|---|---|---|---|---|---|
| Bear | 2% FCF growth, 11% WACC - China structural loss, no data center, Apple gone | $46B | $152B | $430B | 40% |
| Base | 8% growth, 10% WACC - inventory normalizes, auto grows, data center emerging | $76B | $223B | $565B | 56% |
| Bull | 15% growth, 9.5% WACC - data center contributes, agentic upgrade cycle, auto $10B+ | $113B | $317B | $787B | 72% |
| Transformative | 22% growth, 9% WACC - platform company across auto + DC + edge AI + 6G | $133B | $455B | $1.29T | 81% |
Two things stand out. First, the bear case downside is bounded - even in the worst modeled outcome, the median intrinsic value ($152B) is only 22% below today. A company producing $12.8B in FCF annually doesn't go to zero; the licensing business alone is worth $30-40B in a downside case. Second, the distribution is asymmetric - upside scenarios produce median outcomes 1.6x to 2.3x the current market cap, driven by FCF compounding in automotive and data center.
The bear scenario (40% probability it's undervalued) is the honest admission that risks are real of sustained China tariff escalation, memory-driven demand destruction that outlasts the inventory cycle, or data center execution failure and would all push toward that left tail.
TL;DR
QCOM is a $195B market cap generating $12.8B in annual FCF - a 6.6% FCF yield - with its two largest headwinds (Apple, China inventory) well-understood, sized, and priced in. The business that remains after those headwinds is growing: automotive at $6B+ and accelerating, IoT expanding into physical AI, and a data center entry that isn't in anyone's model yet. 17x forward earnings against a peer group at 35x, you're being compensated to take on a headline risk that the management says is peaking. The June 24 Investor Day is the catalyst that closes the information gap on the data center opportunity. If QCOM is still trading at a 50% discount to peers in a year, I guess I'm wrong. Price Target $300-400 by end of 2027.
Positions
$40K in shares @$190 and single 21Aug 220C for investor day
The OP is bullish on Qualcomm (QCOM). The core thesis is that the market still prices it as a legacy handset chip company weighed down by Apple’s in-house modem efforts and China exposure, while underestimating its high cash flow, licensing business, automotive chips, and potential re-acceleration from data center custom silicon. The post divides Qualcomm into the QCT chip business and the QTL licensing business, arguing that QCT accounts for the vast majority of revenue, while QTL, despite being a smaller revenue contributor, has an extremely high EBIT margin and represents high-quality cash flow similar to a “patent toll booth.” The OP uses FY2025, Q2 FY2026, and TTM data to show that the company had FY2025 revenue of about $44.3 billion and FCF of about $12.8 billion, while headline GAAP net income and EPS were distorted by one-time tax items; operating income and free cash flow better reflect the company’s true earnings power. On valuation, the OP argues that QCOM trades at about 17x forward PE, significantly below the semiconductor peer median, and that the discount mainly comes from high China revenue exposure and Apple in-house modem substitution risk, both of which are already well understood by the market. Regarding Apple, the OP believes around $2-3 billion of QCT product revenue is at risk, but QTL patent licensing does not immediately disappear just because Apple uses an in-house modem. Regarding China, the OP attributes recent pressure to handset OEM destocking caused by HBM/memory tightness rather than structural deterioration in end demand or share. The growth evidence mainly centers on automotive, IoT/physical AI, and data centers: automotive revenue has maintained high double-digit growth for multiple quarters, Q2 FY2026 annualized run-rate exceeded $5 billion, and management guided for an FY2026 exit run-rate above $6 billion; on data centers, the OP emphasizes that the Q2 FY2026 call mentioned a custom silicon engagement with a leading hyperscaler, with initial shipments expected to begin around December 2026, and views the June 24, 2026 Investor Day as a catalyst for revealing the data center and Physical AI roadmap. For valuation modeling, the OP starts from $12.8 billion of FCF, about 1.072 billion shares, and a stock price of about $182 to run a Monte Carlo DCF, producing median outcomes above the current market cap under base/bull/transformative scenarios and setting a year-end 2027 price target of $300-400. The image attachment is a position screenshot showing QCOM at about $194.57 at the time, up 4.30% on the day, with about 215 shares of QCOM and 1 220C expiring August 21, 2026, matching the disclosure in the text of about a $40,000 stock position and one 21Aug 220C. The most important weaknesses are: the peer valuation comparison may be selective; QCOM’s handset business cyclicality and Apple substitution risk are not necessarily fully “priced in”; data center custom silicon currently only has initial shipment timing and a general customer description, with no details on scale, pricing, margin, or competitiveness; the DCF is highly sensitive to long-term FCF growth rates and terminal value, and the claim that “bear case downside is bounded” depends on model assumptions rather than any market guarantee.
There was relatively little substantive discussion in the comments, and the sample leaned heavily meme-like. The top comment and several high-scoring comments did not mainly rebut QCOM’s fundamentals, but instead questioned the OP’s “8th grade research project” persona, pointing out that the account had a 10-year history and looked like an adult pretending to be an eighth-grade student. Others said the post was obviously generated by Claude/ChatGPT, mocking the originality and credibility of the research. Although these comments scored highly, they mainly attacked the writing style and identity framing and did not systematically dismantle the financial model or business assumptions. One more substantive comment argued that QCOM’s discount mainly comes from Apple modem substitution risk and said the commenter owned QCOM shares but had closed long calls after the stock jumped; that comment agreed QCOM is reducing its dependence on AAPL and recognized its financial quality, management, and FCF, but also warned that the moat is weakening. Another comment mentioned that QCOM had a high weighting in a certain Democratic House member’s portfolio, implying a political-trade-following angle, but this is not a verified investment argument. Another comment compressed the main post into a TLDR of “Ticker QCOM, direction up, catalyst June 24 Investor Day, shares and 21Aug 220 calls,” which was a supportive restatement. There were no observed replies from [OP], so it is not possible to see how the OP handled pushback. Overall, the community’s sarcasm about the post format and signs of AI generation far outweighed deeper verification of the QCOM thesis; the few substantive responses acknowledged that QCOM’s valuation and cash flow are attractive, but remained cautious about declining Apple dependence, a weakening moat, and option timing risk.
Loading Truth Social...
Loading live updates...