1 AI Stock I Wouldn't Touch, and 1 I Absolutely Love

Cloudflare reported Q4 revenue up 33.6% YoY to $614.5M but a GAAP net loss of $12.1M and operating loss widened to $49.2M; stock-based compensation totaled $451.5M in 2025 (≈33% YoY and >20% of revenue), while market cap exceeds $76B despite ~<$2.2B annual revenue. Apple posted fiscal Q1 2026 revenue of $143.8B (+16% YoY) and EPS +19%, and struck a multi-year deal to embed Google Gemini into upcoming Siri/Apple Intelligence features. Implication: Apple is positioned as a profitable, lower-risk AI beneficiary with a hardware upgrade cycle and services upside; Cloudflare shows strong top-line momentum but high equity-compensation-driven losses and a valuation that elevates execution risk.

Analysis

A large consumer-hardware/software ecosystem that layers advanced LLM capabilities onto devices indirectly monetizes AI through hardware upgrade cycles and higher services ARPU; the real leverage comes from distribution velocity, not model IP. The primary beneficiary is the party selling inference at scale — every incremental on-device or cloud-hosted query can be monetized via CPU/GPU time, subscription uplifts, or per-query fees, creating a recurring annuity that compounds faster than one-off hardware revenues.

For specialist connectivity/security vendors, the competitive hazard is twofold: hyperscalers can internalize edge-routing and inference hosting into their broader cloud stacks, and aggressive stock-based pay structures mask true per-share economics until normalized. If compensation-driven dilution continues, investor returns will hinge on demonstrable per-customer ARPU lifts and margin expansion; absent that, multiple compression is the likely outcome even with healthy top-line growth.

Key catalysts to watch over 3–18 months are product launch cadence (driving upgrade waves), quarterly disclosures on ARPU/ACV conversion and SBC run-rate guidance, and any regulatory scrutiny of exclusive LLM arrangements. For positioning, prefer convex exposure to large, cash-generative ecosystems that can monetize AI incrementally, hedge exposure to high-growth-but-loss-making infrastructure names, and size options to reflect binary execution outcomes rather than linear beta exposure.

Key Decisions for Investors

• Pair trade (12–18 months): Long AAPL (core equity or buy a 12-month call spread 10% OTM) + Long GOOG/GOOGL (buy 12-month LEAPS or 6–12 month calls) vs Short NET (buy Sep-2026 put spread ~25–35% OTM). R/R: asymmetric upside (20–40% on longs if upgrade+services monetization accelerates) vs concentrated downside on NET (potential 40–60% downside if dilution/multiple compression materializes); size net exposure to 2–4% of portfolio.
• Tactical hedged short on NET (3–6 months): Buy a protective put spread (limit max loss to premium + spread) after earnings if SBC guidance stays elevated. R/R: limited defined loss (premium) vs potential large gain if market re-rates; keep position small (1–2%).
• Convex long on NVDA (6–12 months): Small allocation (1–2%) via out-of-the-money call options to capture upside from accelerated inference demand; expect >2x upside if adoption surprises, premium loss if semiconductor cycle slows.
• Risk trigger/stop: if a target hardware/software launch misses consumer adoption signals or if quarterly ARPU/ACV conversion underperforms expectations, reduce long AAPL/GOOG exposure by 30% and reallocate to cash within 1–3 weeks to preserve capital.