Does Billionaire David Tepper Know Something Wall Street Doesn't? He's Selling Alphabet and Amazon and Piling Into This AI Stock Instead.

Billionaire David Tepper trimmed Appaloosa’s Alphabet and Amazon stakes in Q3 2025—selling roughly 7.5% of Alphabet and 7.4% of Amazon—while increasing his Qualcomm position by 255.7%. Despite broad Wall Street bullishness on Alphabet (57 of 66 analysts buy/strong buy) and Amazon (66 of 67 buy/strong buy), Tepper is aggressively buying Qualcomm amid its push into edge AI and upcoming AI200/AI250 data‑center chips; Qualcomm has mixed analyst support (16 of 36 buy/strong buy, 19 hold) and a forward P/E of about 13.8. Appaloosa still holds sizable positions (Amazon remains its second‑largest, Alphabet fifth), so the trades signal tactical repositioning toward AI exposure rather than outright exits from mega‑cap names.

Analysis

Market structure: Tepper's 255.7% increase in QCOM signals a tactical tilt from cloud-centric AI (AMZN, GOOG) toward edge and diversified accelerator supply. Winners: QCOM, semicap vendors, handset OEMs that monetize edge AI; losers over time could be pure cloud-margin plays if edge offloads volume. Supply/demand suggests stronger near-term demand for AI chips and fab capacity (TSMC/SMIC), implying higher equipment/order visibility for 6–18 months and upward pressure on semi capex flows.

Risk assessment: Tail risks include US/China export controls, QCOM execution delays (AI200 push in 2026, AI250 in 2027), and TSMC capacity shortages; each could wipe out >30% of expected upside if realized. Immediate (days) moves will track positioning and options vol; short-term (3–12 months) hinges on QCOM product reveals and 2025/26 guidance; long-term (2–5 years) depends on adoption of edge AI vs cloud accelerators. Hidden dependency: QCOM’s licensing revenue and handset cycles still drive cash flow — a smartphone downturn would materially impair funding for data-center pushes.

Trade implications: Tactical allocation to QCOM (valuation forward P/E ~13.8) is justified but must be risk-managed: prefer staged entries and volatility-financed LEAPs ahead of 1–2 major product/capacity catalysts. Relative-value: long QCOM vs short AMZN/GOOG expresses confidence in edge monetization (size 1–2% net exposure) while hedging cloud cyclicality. Cross-asset: expect upward skew in semiconductor equity vols; consider buying focused semicap names and selling Treasury duration if capex ramps accelerate inflation expectations.

Contrarian angles: Consensus underprices QCOM’s licensing + new-accelerator optionality and overprices time-to-revenue for competitors; downside scenario is a pricing war with NVDA/AMD compressing ASPs — historical parallel: Intel’s failed server AI push where late entry lost share. Actionable trigger points: cut exposure by half if QCOM slips >20% on execution news or if AI200 slips beyond H2 2026; add if confirmed design wins within 6–9 months.

Key Decisions for Investors

• Establish a staged 2–3% portfolio long in QCOM over 90 days: initial 1% equity position now, add 0.5% on any >8% pullback, another 0.5% on >15% pullback; target +30–50% upside over 12–24 months, hard stop-loss at -20% from average entry.
• Buy QCOM 12–18 month LEAP calls (~delta 0.30–0.40) sized to 1% portfolio exposure and finance by selling 30–60 day calls (~1% notional) to collect premium; roll LEAPs if design-win headlines occur within 6–9 months.
• Deploy a pair trade: long QCOM 2% vs short AMZN 1.5% (or short GOOG 1.5%) to express edge-over-cloud thesis; unwind if QCOM underperforms the short leg by >15% within 3 months or after QCOM posts confirmed AI200 wins.
• Trim Mag7 overweight by 1–3% and reallocate proceeds to semiconductor suppliers and semicap suppliers (e.g., equipment/TSMC-exposure) over 30 days; increase duration exposure to 0.25–0.5% if capex commentary implies sustained demand to hedge secular inflation in tech capex.
• Monitor three hard catalysts in next 6–12 months and act accordingly: 1) QCOM product roadmap cadence (confirm AI200 taped/validated by Q4 2025); 2) TSMC capacity allocation reports (if QCOM share <10% of AI wafer allocation, reduce position by 50%); 3) any US/China export-control action—exit or hedge immediately if new controls restrict QCOM’s China licensing or wafer access.