Qualcomm Is the Most Underrated AI Chipmaker to Buy

Qualcomm raised its fiscal 2029 guidance target to $40B (nearly 2x prior target) and aims for >$15B in AI infrastructure/data-center revenue by fiscal 2029, anchored in agentic AI and hyperscaler custom-silicon deals. Near term, it projects fiscal 2026 Q3 revenue of $9.2B-$10.0B ($9.6B midpoint), implying a 7.4% YoY decline, but investors are framing 2027+ as the inflection point. The article highlights ongoing buybacks totaling $5.4B in H1 FY2026 plus a new $20B authorization and notes a 1.8% dividend yield, with the stock valued below peers (P/E 21.2).

Analysis

The core setup is a timing mismatch: QCOM may be building a real AI option, but the equity will only re-rate if that option becomes visible in reported revenue before the market loses patience. Over the next 1-2 quarters, handset weakness and execution skepticism still dominate the tape, so the stock can trade like a mature cyclical even if the 2027-29 narrative is intact. The buyback/dividend support lowers left-tail risk, but it does not create upside unless non-handset revenue starts compounding.

Relative winners are more likely the proven AI infrastructure names than the new entrant. AVGO has the cleaner monetization path because it already has the design-win ecosystem, while QCOM is a higher-beta call option on inference and custom silicon; NVDA is less threatened in training, but inference share shifts toward power-efficient ASICs could create narrative pressure over time. The second-order winner may actually be foundry/advanced-packaging capacity, which usually captures the earliest dollars when new silicon programs ramp.

Contrarian view: the market may be underestimating QCOM's long-duration optionality, but it is probably overestimating the speed of monetization. If 2026 guidance still looks handset-led and management does not disclose at least one credible hyperscaler ramp, the AI story can get marked down as marketing rather than earnings power. The clean falsifier is simple: no sequential acceleration in non-handset revenue, no customer disclosure, or any further near-term guide cut.

Key Decisions for Investors

• Initiate only a starter long in QCOM on a 5-7% pullback; treat it as a 12-18 month rerating trade, not a near-term earnings trade. Trim if the next two quarters fail to show sequential improvement in non-handset revenue.
• Express the thesis with a Jan-2027 QCOM call spread struck roughly 15-20% above spot; this captures the 2027+ inflection without paying full theta for the dead-time between now and commercialization.
• Do not chase AVGO on the same headline; it is the cleaner execution story and likely the better AI infrastructure long if you want current earnings power rather than optionality.
• Set a watch item, not a trade: if management still has not named a meaningful hyperscaler or disclosed data-center revenue momentum by the next two earnings prints, assume the multiple can compress back toward mature-chip levels.