QCOM — The cheapest chip stock on every multiple, carrying a $7B time bomb

What the company actually does

• QCT (Qualcomm CDMA Technologies) — the chip business, generating the vast majority of revenue. Includes Snapdragon processors for smartphones, automotive SoCs (system-on-chip), and IoT/edge computing chips.
• QTL (Qualcomm Technology Licensing) — licenses Qualcomm's foundational cellular patents (3G, 4G, 5G) to device manufacturers globally. High-margin, capital-light.
• QSI (Qualcomm Strategic Initiatives) — venture investments and early-stage technology development.

Screening criteria: the three gates

Gate 1 — ROE track record

Gate 2 — Free cash flow

Gate 3 — Valuation

Revenue and earnings

Balance sheet

Competitive position and moat

Risk factors

Near-term catalysts and market context

• June 24, 2026 — Investor Day: Qualcomm will detail its data center and Physical AI roadmap. This is the single most important near-term event for the bull case. If Qualcomm can show credible contracted revenue and a product pipeline for AI inference chips, the "unproven diversification" narrative changes. If the Day produces slide decks rather than design wins, the bear case strengthens.
• July 29, 2026 — Q3 FY2026 earnings (after market close): The first quarterly read that will show Q3 results without the tax reversal noise. Non-GAAP EPS consensus stands at roughly $2.65–$2.70. Automotive and IoT segment trajectory will be watched closely.
• September 2026 — iPhone 18 launch: The formal start of the Apple modem transition. Qualcomm expects to supply modems for roughly one-fifth of the iPhone 18 lineup before going to zero in the 2027 cycle.
• ByteDance AI chip ramp: Qualcomm announced a data center AI chip engagement with a leading hyperscaler expected to produce initial shipments later in calendar 2026. CEO Amon said at Q2 earnings: "We are equally excited by our entry into the data center, where a leading hyperscaler custom silicon engagement is on track for initial shipments later this calendar year." 16 The ByteDance ASIC deal announcement in late May 2026 drove a sharp one-day rally, then was completely unwound over the following two trading days — a "sell the news" pattern suggesting the market already had the AI diversification story priced in at a higher level.
• 52-week range: $140.87 – $259.92. Current price $191.20, sitting 27% above the 52-week low and 26% below the 52-week high. RSI (14-day): 43.31 — neutral, not oversold. Short interest: 4.27% of float (45.04M shares), declining from a prior month high of 50.44M, suggesting some short covering. 18
• Dividend: annual $3.68/share (1.93% yield), 23 consecutive years of growth. Ex-date and next payment information available at investor.qualcomm.com.

Bull / bear framework

Bull case

1. FCF yield of 6.20% is the honest multiple, not the P/E. The P/E is distorted by GAAP tax accounting in both directions. P/FCF of 16.12× is exactly where it has averaged over five years — and the underlying business generated $12.5B in free cash on $44.5B of revenue last year. At 6.20% FCF yield with growing automotive and IoT segments, the market is not obviously overcharging for the cash stream.
2. The Apple modem loss may already be priced in. Wall Street has been modeling the Apple transition for years. A $7–8B annual revenue loss sounds alarming, but it represents approximately 16–18% of Qualcomm's $44.5B TTM revenue — and the modem business carries lower margins than QTL's licensing stream. The stock has already corrected 26% from its 52-week high. Priced-in risks tend to produce smaller sell-offs when they materialize than when they are still uncertain.
3. Automotive is a credible replacement cycle. Q2 FY2026 automotive revenue of $1.33B at +38% YoY, with design wins extending through 2031, represents a segment in its early ramp. Unlike smartphones, automotive design cycles are 3–7 years — once Snapdragon Digital Chassis is designed into a vehicle platform, it is effectively locked in for the production lifetime of that platform. The royalty economics are structurally attractive.
4. Diversification has real contracted revenue, not just slides. The data center AI chip engagement is confirmed (CEO on the record that initial shipments are planned for calendar 2026). Qualcomm's custom silicon work is not speculative in the same way a startup's product roadmap is.
5. Capital return is a floor. At a 5.44% combined shareholder yield, QCOM is returning more than its FCF yield net of re-investment — the dividend is safe at a 39% payout ratio, and the buyback is reducing the share count. If earnings growth is flat but buybacks reduce shares by 3.65%/year, EPS grows by approximately the same amount mechanically.

Bear case

1. The Apple revenue hole may be underestimated in timing. Qualcomm supplies modem chips that generate an estimated $7.3–7.8B annually from Apple. 2 Even if automotive and IoT are growing at 30–38%, they generated a combined ~$2.6B last quarter — still well below the Apple modem revenue that disappears within twelve months. The growth rate on diversification does not close the gap fast enough at current scale.
2. Zero consensus EPS growth is the market's base case, not a pessimistic scenario. The 3-year forward EPS growth forecast of 0.07% is consensus — it already incorporates Qualcomm's own guidance. If the Apple modem transition runs faster than expected, or smartphone unit volumes disappoint again, this consensus number goes negative. Flat-earnings businesses that don't grow eventually see multiple compression.
3. China concentration has two-sided risk. Export controls could restrict chip sales; Chinese OEMs building out domestic alternatives (like HiSilicon, funded by Huawei) could gradually reduce dependence on Qualcomm chips. The combination of regulatory risk and customer self-sufficiency in a single geography representing 50%+ of chip revenue is not a benign setup.
4. Analyst consensus is below the current price. Thirty-nine analysts, average target $180–$186. When the professional consensus says the stock is 3–6% overvalued right now, the burden of proof falls on the counter-thesis. The bear end of the analyst range (Seaport at $100, Morgan Stanley at $132, Barclays at $130) reflects models that treat the Apple transition as an earnings inflection, not a priced-in event. 19
5. The data center AI thesis requires execution, not just announcement. Qualcomm's hyperscaler custom silicon engagement will produce initial shipments "later this calendar year" — but ByteDance is one customer, and the announcement caused a one-day rally that was fully erased within 48 hours. The market is treating data center AI as a story to be proven, not a done deal. Barchart's analysis of the short interest ($11.8B at a decade-high) summarized the skeptic view: "The chipmaker needs to quickly deliver on its AI pivot for bulls to win." 19