Tech analyst Dan Ives flags 10 stocks to own, insists there's no AI bubble

Wedbush analyst Dan Ives argues the AI rally is early-stage, noting only ~3% of U.S. companies and <1% globally have adopted AI, and that enterprise spending, government demand and chip shortages — particularly for Nvidia — are driving valuations rather than speculation. He highlights structurally indispensable names across chips, hyperscalers, devices, cybersecurity and autonomous tech (top picks: Microsoft, Palantir, Nvidia, AMD, Tesla, Apple, Meta, Alphabet, CrowdStrike, Palo Alto Networks) while placing Amazon, Salesforce, IBM and Intel in a broader 'AI 30' universe as supportive but not category-defining. The report frames AI adoption as far from late-cycle froth and points to supply-demand imbalances (chip shortages) underpinning upside for key incumbents.

Analysis

Market structure: AI adoption creates clear winners (NVDA, AMD, MSFT, GOOGL/GOOG, AAPL, META, PLTR, CRWD, PANW) because GPU/accelerator scarcity and hyperscaler lock‑in increase pricing power and recurring revenue; expect NVDA to command >50% gross margin on datacenter GPUs and maintain >30% YoY revenue growth through FY+1 if supply holds. Losers are legacy CPU vendors (INTC), slow‑moving enterprise software (CRM, IBM) and any hardware vendors unable to secure 2025‑26 foundry slots. Cross‑asset: sustained tech outperformance will steepen real yield paths, lift USD (pressure EM FX), increase implied volatility in single‑name tech options, and boost copper/specialty gases demand (+5–10% next 12 months).

Risk assessment: Tail risks include export controls to China (10–25% probability over 12 months), a corporate capex pullback in a recession (15–30% hit to FY+1 guidance), or a major NVDA yield problem (5–10%). Immediate risk window: next 30–90 days (earnings/guidance, TSMC capacity updates); medium term: 3–12 months (share shifts, product launches); long term: 2–5 years (robotaxi adoption, enterprise AI saturation). Hidden dependencies: hyperscaler procurement cycles, TSMC node capacity, and data‑center power constraints; catalysts: NVDA/AMD earnings, TSMC capacity roadmap, and EU/US AI regulation announcements.

Trade implications: Tactical: overweight NVDA and AMD for 3–12 months (scale on ≤‑10% pullbacks), underweight INTC/IBM/CRM. Pair trades: long AMD vs short INTC (size 1–2% net); long CRWD/PANW (security moat) 1–2% each. Options: prefer 3–6 month call spreads on NVDA to cap premium, sell OTM puts to accumulate on dips; buy protective puts on long positions around macro events.

Contrarian angles: Consensus underestimates integration costs and margin pressure as competition (custom accelerators, Habana, Graphcore) scales; NVDA’s pricing power may compress if AMD/own silicon lift share to >25% in 18 months. Historical parallel: 1999 tech froth differs materially—this cycle has real cash flows but also real regulatory fragmentation risk. Watch for consolidation and talent‑cost inflation that could erode net margins unexpectedly.

Key Decisions for Investors

• Establish a 2–3% portfolio long in NVDA using 3–6 month call spreads (buy nearer‑ATM, sell 15–25% OTM) to capture upside while capping premium; scale add on any pullback ≥8–12%; take profits on a 25–35% move or if guidance falls >10% sequentially.
• Initiate a 1–2% long position in AMD for 12–24 months to play share gains in datacenter GPUs, hedge with a 1% short in INTC (or buy 6–12 month INTC puts) and add on any AMD dip >15%; reduce if AMD quarterly server design wins miss consensus by >300bps.
• Reduce exposure to CRM, IBM, INTC by ~50% vs benchmark over the next 3 months and redeploy into AI infrastructure and cybersecurity; consider selling 6–12 month covered calls on remaining CRM/IBM to generate income until clear AI leverage appears.
• Allocate 1–2% each to CRWD and PANW via outright or 6–12 month calls (buy or call spreads); set stop/exit if reported ARR growth drops below 20% YoY for CRWD or 15% YoY for PANW, and watch for margin expansion as a trigger to add.