Jensen Huang Just Told Investors to Buy This Under-the-Radar Artificial Intelligence (AI) Chip Stock (Hint: It's Not Marvell)

Jensen Huang publicly endorsed Qualcomm, saying Nvidia is not well-suited to mobile AI and pointing investors toward Qualcomm's edge-computing strengths in smartphones, autos, and IoT. Qualcomm's Q2 automotive revenue grew 38% year over year, while IoT posted high-single-digit growth, and the company is expanding into AI inference with AI200 and AI250 accelerators. The article frames Qualcomm as a lower-power, reasonably valued beneficiary of on-device AI adoption, which is modestly supportive for the stock.

Analysis

The key signal is not the endorsement itself, but the market segmentation it implies: edge AI is becoming a structurally different profit pool than data-center AI. If the AI stack bifurcates into high-power training and low-power inference, NVDA keeps the crown on training, while QCOM gets a longer-duration monetization curve in devices where battery, thermals, and BOM discipline matter more than raw FLOPS. That makes QCOM less of a direct competitor and more of a pick-and-shovel enabler for the next wave of AI adoption in consumer and automotive endpoints.

The second-order winner set likely extends beyond QCOM to the ecosystem that benefits from inference silicon proliferation: handset OEMs, autos, and OEM-linked software services that can localize workloads and reduce cloud spend. The risk is that market enthusiasm over “AI everywhere” overstates near-term revenue conversion; edge AI adoption is real, but ASP uplift per device is typically slower and more fragmented than data-center capex cycles. If enterprise budgets tighten, the first thing cut is experimental edge deployments that do not show immediate productivity payback.

The contrarian read is that QCOM may be underappreciated not because it wins the headline AI race, but because it avoids the capital-intensity trap. A modest multiple is meaningful if the company can compound earnings through multiple end-markets without needing hyperscaler-scale capex; that creates a better downside floor than higher-flying AI names priced for perfect execution. For NVDA, the bigger implication is that its addressable market is still enormous, but it may face more coexistence than replacement—especially as inference migrates out of the cloud over the next 12-24 months.

Key Decisions for Investors

• Long QCOM vs. short a basket of premium-multiple AI hardware names over 3-6 months; thesis: edge inference monetization is earlier but more durable, while valuation compression is more likely elsewhere if AI capex decelerates.
• Buy QCOM on pullbacks with a 6-12 month horizon; target a rerating if automotive and IoT continue compounding and AI PC adoption stabilizes, with downside limited by valuation support.
• Hold NVDA but do not add aggressively on this headline; the message is coexistence, not displacement. Use QCOM strength as a signal to trim some beta in the AI semiconductor basket rather than chase the entire group.
• Optionality trade: QCOM call spreads 6-9 months out to express upside from edge-AI adoption with defined risk; favor strikes that assume only modest multiple expansion, not a full AI re-rate.
• If you want a cleaner relative-value expression, long QCOM / short INTC on a 6-12 month view: QCOM has clearer AI endpoint monetization, while INTC remains more execution-dependent and less directly leveraged to mobile inference.