Dan Ives Says AI Stocks Aren’t in a Bubble and Names His Top 10 Picks

Wedbush analyst Dan Ives argues the AI boom is in its early innings, citing only ~3% of U.S. companies and <1% of global firms having adopted AI and projecting AI-related spending of $550–$600 billion by 2026; he points to tight AI-chip supply and large real revenues at incumbents as evidence the rally is demand-driven rather than speculative. Ives names 10 category-defining AI plays — including NVDA, MSFT, AMD, AAPL, GOOGL, META, PLTR, CRWD, PANW and TSLA — and notes Nvidia’s chips are in acute demand; Wall Street average targets cited imply NVDA has ~42.8% upside (avg target $257.26) while TSLA’s average target ($383.04) implies ~10.2% downside.

Analysis

Market structure: The immediate winners are NVDA, MSFT, GOOGL and enterprise AI software/security names (CRWD, PANW) because they capture both chip/IP and recurring software spend; losers include late-stage valuation-dependent AI pure‑plays and discretionary/auto (TSLA) if robotaxi timelines slip. Tight supply of HBM and GPUs keeps pricing power for NVDA/AMD/TSMC for 6–18 months and supports gross‑margin tailwinds of +200–800bps versus peers that lack proprietary silicon. Cross‑asset: stronger tech earnings should tighten high‑grade credit spreads but push 2–5y Treasury yields +10–30bps on higher capex; USD likely to stay firm on US tech outperformance, and copper/energy demand rises modestly as data‑center buildouts accelerate.

Risk assessment: Tail risks include US/China export controls or EU antitrust actions (10–25% probability next 12–24 months), a sudden supply glut if TSMC/Intel scale capacity faster than expected (low-probability 5–15% within 18 months), and execution failure in autonomous driving (higher for TSLA). Short term (days–weeks) focus on NVDA supply commentary and MSFT commercial AI uptake; medium (3–12 months) on chip capacity and customer case studies; long term (2024–2026) on realized AI spend hitting the $550–600B range Ives cites. Hidden dependencies: power grid and colocation capacity; higher power costs are a second‑order margin tax on data‑center operators.

Trade implications: Allocate concentrated, time‑limited exposure to NVDA (core) and MSFT (enterprise AI) while using options to control risk. Consider 2–4% portfolio long NVDA, 2–3% long MSFT, 1–2% long CRWD/PANW for security decarbonization and SOC automation; short 1–2% TSLA or use short-dated puts on TSLA to express skepticism on robotaxi timelines. Use pair trade long NVDA vs short TSLA or PLTR to hedge sentiment. Options: buy NVDA 12–18 month 25% OTM LEAPS (limit size to 1–2% cash) and sell covered calls on core MSFT after 10–15% rallies.

Contrarian angles: The consensus underestimates integration friction—50–70% of enterprises may need 12–36 months to extract meaningful productivity gains—so short-term multiples are vulnerable to execution slippage. Valuation concentration in NVDA (analyst implied +43% upside) is a single‑point failure; set automatic trims if NVDA outperforms the market by >25% in 30 days or if semiconductor inventory days rise >15% quarter‑over‑quarter. Historic parallel: unlike 1999, revenues exist, but concentrated winners + rapid capex cycles create regime risk where few equities carry portfolio performance—manage sizing and correlation risk accordingly.

Key Decisions for Investors

• Establish a core long NVDA position equal to 3% of portfolio capital via 12–18 month 25% OTM LEAPS (limit exposure to 1–2% cash for option premium) and layer 1–2% additional delta via near‑term calls ahead of next earnings if supply commentary is constructive.
• Initiate a 2–3% long position in MSFT (cash equity) to capture enterprise AI adoption; if MSFT rallies >15% in 60 days, sell 50% of the position and write 6–9 month covered calls to monetize premium.
• Open a paired trade: long 2% NVDA vs short 1–1.5% TSLA (cash short or buy put spread, 3–6 month expiries) to express conviction in semiconductor-driven AI demand and skepticism on near‑term autonomous revenue for TSLA.
• Add a 1–2% tactical long across CRWD and PANW (split) for AI‑driven cybersecurity exposure; set a stop‑loss to trim if combined position falls >12% or if subscription churn tick‑up >100bps in next two quarters.
• Reduce discretionary/auto exposure (TSLA and suppliers) by 25% within 30 days and reallocate to semis/security if NVDA/AMD gross margins exceed consensus by >200bps or if NVDA sells through >90% of guided GPU units for the quarter.