1 Beaten-Down AI Stock to Buy and 1 to Avoid

Alphabet (GOOG/GOOGL) is down ~13% YTD but remains a wide‑moat leader leveraging AI (Gemini, AI search modes) to boost engagement and create multiple growth avenues while retaining the ability to cut costs if needed. Recursion Pharmaceuticals (RXRX) is an AI-driven biotech with no marketed products and no phase‑3 assets, consistently unprofitable and high‑risk — a potential high-reward but speculative investment unsuitable for average risk tolerance.

Analysis

Winners will be firms that capture the incremental compute and monetization that AI introduces rather than those that merely hype it. Expect sustained revenue mix shifts toward enterprise AI subscriptions, fine-tuning services, and higher-margin cloud AI offerings over the next 12–36 months; that benefits hyperscalers with differentiated models and GPU capture (NVDA) and penalizes ad-adjacent publishers and SEO-reliant traffic if generative overviews shorten user journeys. A second-order supply-side effect: sustained model training demand is shifting capex from general-purpose data-center expansion to specialized accelerator capacity and pre-trained model IP — this widens the moat for vendors that control silicon and software stacks and raises bar for legacy fabs and commodity cloud resellers (INTC faces strategic choice points).

Key risk horizons differ by name. For large cap tech the primary near-term risks are execution/PR around product launches and QoQ margins (days–quarters), while the fundamental upside unfolds over years as ARPU from enterprise AI products materializes; cost flexibility is a credible downside hedge for incumbents. For AI-enabled biotech platforms the timeline to de-risk is multi-year and binary: IND/phase 2 signals or partner licensing are the only reliable value inflection points — absent those, dilution and negative sentiment dominate. Clinical failure or model commoditization can quickly reverse narratives in either bucket.

Tactically, prefer quality exposure funded by selling optionality around speculative names. Size thematic long positions in market leaders modestly (2–5% of portfolio) with 6–12 month put protection, and express asymmetric short or limited-risk bearish exposure to unproven platform biotechs through put spreads or small outright shorts. Monitor three near-term catalysts for re-rating: major enterprise AI contract wins, GPU supply/delivery surprises, and initial randomized clinical readouts or pharma licensing deals; use those events to add/pass on conviction.

Contrarian read: the market underweights incumbent optionality to reprice legacy products into higher-margin AI bundles — that optionality is real, monetizable, and defensible via data network effects. Conversely, the market often prices AI-biotech as a pure tech story without discounting clinical, regulatory, and partner-adoption execution risk; that mismatch creates asymmetric trade opportunities where limited-cost option structures can pay off if either outcome crystallizes.

Key Decisions for Investors

• Long GOOGL (2–4% portfolio) with a 12-month horizon; hedge with 6–9 month 10% OTM puts sized to cover 50% of the position. Rationale: capture enterprise AI ARPU upside while limiting drawdown risk; target asymmetric return of 20–40% vs a downside capped by the hedge.
• Pair trade: Long GOOG (notional X) / Short RXRX (notional X) for 6–18 months. Implement short via RXRX 6–12 month put spreads (limit max loss) or a small outright short (max 0.5–1% portfolio) to express premium-on-platform vs unproven pipeline. R/R: asymmetry favors long incumbent quality vs binary biotech downside.
• Buy NVDA call spreads (3–6 month) to play continued accelerator tightness and training demand; size at 1–2% portfolio. Use call spreads to keep cost low while capturing near-term supply-driven re-rating; unwind on any sign of easing GPU deliveries or ASP compression.
• Avoid unhedged long positions in RXRX; if seeking binary upside, allocate <0.5% portfolio to LEAP call options (12–24 months) rather than stock to limit dilution/clinical-failure exposure. Exit or trim upon negative phase2/IND news or before capital raises to avoid dilution.
• Event-based alert: add to long GOOG or NVDA on pullbacks of 10–15% that are not accompanied by model/clinical negatives, and remove hedge on confirmed enterprise contract announcements or material partnerships to harvest optionality.