QCOM — Qualcomm
TL;DR
Qualcomm is a secular bear story converging on a single timeline: Apple’s licensing deal expires March 2027, Apple’s C1 modem shipped in 2025, and Apple expects to eliminate Qualcomm from its lineup by 2027–2028. That cliff — Qualcomm’s largest customer walking out the door — arrives while Qualcomm faces simultaneous pressure from China (MediaTek/Huawei share gains, customers downgrading to cheaper SKU variants), rising DRAM costs it can’t hedge with bulk contracts (unlike Nvidia/Broadcom/AMD), and the departure of key Nuvia/Oryon CPU architects.1
Business
Fabless semiconductor company. Primary revenue: mobile SoCs (Snapdragon) with integrated modem + CPU/GPU/NPU, sold into Android OEMs and Apple (modem only). Secondary revenue: IoT, automotive (Snapdragon Digital Chassis), and PC (Snapdragon X Elite via Nuvia’s Oryon CPU cores). Royalty-based licensing business (QTL) on top of QCT (chips). Largest single customer is Apple (modem supply). Manufactured at TSMC (primary) and Samsung (SF2/SF4 considered for 2027 Snapdragon derivatives).123
Thesis
There is no corpus bull thesis on QCOM. The corpus frame is uniformly bearish on the medium-term:
- Apple modem cliff, March 2027. Apple’s C1 modem shipped in iPhone 16e (2025); Apple expects to eliminate Qualcomm from its lineup “by 2027–2028.” The Apple-TSMC corpus estimates Apple’s annual chip savings from Qualcomm displacement at $1.2B. Apple’s modem team is “literally down the street from Qualcomm HQ,” staffed largely with ex-Qualcomm and ex-Intel engineers.31
- China share erosion. Qualcomm reportedly has the highest (or near-highest) revenue exposure to China in the S&P 500. MediaTek is holding or gaining premium-tier share. Huawei is “fully modem self-sufficient and has no incentive to sign any license deal with Qualcomm.” Chinese OEMs are downgrading from Snapdragon flagship SKUs to cheaper binned “lite” variants — the 8G6 was priced 20% higher than 8G5 but many OEMs chose the 8G6 lite at 8G5 pricing, and Qualcomm is reportedly considering a 5–10% price cut on 8G6.1
- DRAM squeeze. Qualcomm doesn’t buy DRAM in bulk and likely has no long-term contracts, putting it at a pricing disadvantage versus Nvidia/Broadcom/AMD/hyperscalers who have locked in supply. As DRAM prices rise driven by HBM allocation, “economic value is shifting from Qualcomm/MediaTek to Samsung/SK/Micron.”2
- Nuvia architect attrition. Key Oryon CPU architects departed post-vest. Oryon v4 design is ~95% complete, v5 only ~20–50% complete.1
The sole dated bull angle in the corpus (Citrini, Jan 2024) frames QCOM as an “edge AI picks-and-shovels” beneficiary via on-device inference with Snapdragon — predates the bear catalysts above and is not actionable at the current timeline.4
Risks
- Apple cliff execution. If Apple’s C1/C2 modem ramp hits yield or performance issues and Apple keeps Qualcomm longer, the bear thesis timeline extends — but doesn’t change the direction.
- Samsung DX sourcing dynamics. Samsung DX annually pits its own Exynos against Qualcomm; after two full Qualcomm years (2024–2025 when Samsung SF3 couldn’t hit yield/power targets), Exynos returned to 20–30% of Galaxy S in 2026. Qualcomm is reportedly considering Samsung SF2 for 2027 Snapdragon derivatives. Using Samsung Foundry creates annual uncertainty in the most visible Snapdragon design win.3
- Automotive and PC diversification timeline. Snapdragon Digital Chassis (automotive) and Snapdragon X Elite (PC/Copilot+) are the stated diversification plays, but neither has reached a scale that offsets smartphone exposure.1
Recent catalysts
- 2026-02-06 — Irrational Analysis earnings roundup: Q4’25 results; QCOM stock tanked on earnings night alongside ARM (ARM recovered next day, Qualcomm did not). Author frames as “diversification was not fast enough.”1
- 2026-01-16 — Irrational Analysis “Memory Madness”: confirms 8G5 Elite vs. lite SKU binning, DRAM pricing squeeze on SoC vendors; author declares self “Qualcomm perma-bear.”2
- 2026-01-08 — SemiAnalysis Apple-TSMC: $1.2B annual Apple savings figure from modem displacement; Apple modem team staffed with ex-Qualcomm engineers.3
- 2026-01-04 — Irrational Analysis “Fabulous Failures”: Samsung DX 2026 Exynos return to 20–30% Galaxy share; 2027 Snapdragon-SF2 sourcing rumor.5
Second-order reads
- Samsung Foundry health: Qualcomm’s annual sourcing decision is a read-through to Samsung Foundry yield/competitiveness — Exynos returning to Galaxy S in 2026 implies SF3e improved. Watch for SF2 qualification updates → relevant for TSM share at premium nodes.
- MediaTek: Qualcomm’s China share loss is MediaTek’s gain at the premium Android tier; any QCOM China weakness corroborates the MTSI-adjacent narrative of value shifting to component vendors.
- Memory: DRAM pricing impact on Snapdragon ASP and OEM SKU-mix is the same dynamic discussed in MU and SNDK dossiers — rising memory prices squeeze SoC vendors who lack bulk contracts.
Valuation & positioning
No valuation figures in the corpus. Bear thesis anchors to March 2027 Apple licensing expiry and 2027–2028 modem elimination — the catalyst window is 12–18 months out. Irrational Analysis author is an explicit perma-bear on QCOM.12
Sources
Related
TSM — primary foundry; Apple modem displacement reduces a major TSMC-via-Qualcomm revenue stream for QCOM ARM — licensing partner; Oryon CPU cores are Nuvia/ARM-derived architecture INTC — indirect: ARM Neoverse encroachment on datacenter CPU is the same structural shift threatening x86 MU SNDK — DRAM price inflation is the shared cost input squeezing QCOM margins
Footnotes
-
Irrational Analysis — Earnings Roundup: COHR QCOM $ARM — 2026-02-06 ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8
-
Irrational Analysis — Memory Madness — 2026-01-16 ↩ ↩2 ↩3 ↩4
-
SemiAnalysis — Apple-TSMC: The Partnership That Built Modern Semiconductors — 2026-01-08 ↩ ↩2 ↩3 ↩4
-
Citrini Research — Thematic Equity Update — 2024-01-22 (dated; edge AI angle predates current bear catalysts) ↩
-
Irrational Analysis — Fabulous Failures — 2026-01-04 ↩