2 Stocks to Buy for AI's Next Stage

The piece argues that a recent selloff in AI “infrastructure” names creates buying opportunities in experience-focused, “Stage 2” AI companies such as Thomson Reuters and ServiceNow. Thomson Reuters, which has endured a ~60% selloff since mid-last year, has acquired AI capabilities (Additive, SafeSend, Casetext) and the author’s model assigns it ~105% upside, citing upper-single-digit revenue growth and faster profit gains. ServiceNow serves >85% of the Fortune 500 with a 98% retention rate, posted 21% revenue growth in 2025, and analysts forecast ~20% revenue growth and a ~49% EPS increase this year, positioning it as a scalable AI-experience provider that should benefit as enterprises operationalize LLMs.

Analysis

Market structure: The selloff is bifurcating “Stage 1” infrastructure (chips, data centers, utilities) from “Stage 2” experience-layer firms (enterprise AI SaaS, verticalized AI tools). Winners—ServiceNow (NOW), Thomson Reuters (TRI), and other human-in-the-loop vendors—gain pricing power and higher gross margins as customers pay for accuracy, integration and retention (NOW: 98% retention, 75% multi-product attach). Losers are capital-intensive suppliers (e.g., SWKS, CRUS) facing margin compression if data-center demand growth slows or capex is re-staged.

Risk assessment: Tail risks include regulatory intervention on LLMs or liability suits for legal-AI products (6–24 months), a rapid drop in enterprise IT budgets in a recession (near-term), or vertical integration by hyperscalers that accelerates within 12–36 months. Hidden dependencies: Stage-2 wins depend on sustained enterprise spend and renewal metrics (e.g., NOW retention >95%, TRI contract renewals); compute-price deflation or oversupply could amplify Stage-1 defaults and wider credit stress (50–150bps wider HY spreads).

Trade implications: Tactical posture is long durable experience-layer software and short select Stage-1 hardware/developer plays. Use 9–24 month horizons: size core longs 2–3% of portfolio for NOW and 1–2% for TRI, layered (50% now, 50% on 10% pullback). Hedge with 3–6 month put spreads on a SWKS/CRUS basket to limit downside and consider long-dated LEAPS or cash-secured puts to bias positive on NOW/TRI while keeping defined risk.

Contrarian angles: The market underestimates switching costs and human-in-the-loop value—this is the iPhone-app-store parallel where platforms, not suppliers, captured long-term value. Current sentiment likely overstates near-term AI infracapacity risk by ~15–40%, creating entry windows; unintended consequence: a consolidation wave among infrastructure providers could further consolidate pricing power into software platforms over 12–36 months.

Key Decisions for Investors

• Establish a 2–3% portfolio long position in ServiceNow (NOW) with a 12–24 month horizon; DCA 50% immediately and 50% on a further 10% pullback. Alternatively buy 12–18 month ATM call options equal to 25–50% of the notional stock exposure to lever upside while capping capital at risk.
• Add a 1–2% long in Thomson Reuters (TRI) for 12–24 months targeting material rerating; prefer a 12–24 month call spread or buy stock on >15% pullback. Set a tactical stop-loss at -30% from entry or exit if TRI misses AI-related revenue growth two consecutive quarters.
• Establish a 1% short (or buy 3–6 month put spreads) on a small basket of Stage-1 suppliers (equal-weight SWKS and CRUS) to express downside from potential capex rephasing; increase hedge size if implied vol >30% or if data-center capex guidance is cut by >20% on an earnings call.
• Reduce telecom exposure: trim AT&T (T) and Verizon (VZ) weight in comms sleeve by 50–75% over next 30 days and reallocate proceeds to enterprise software (NOW/TRI). Reassess by Q4 2026 or earlier if telco free cash flow improves by >15% y/y.
• Monitor three high-signal catalysts over next 30–90 days before scaling: GPT-5 public release/partnerships, NOW quarterly retention/attach metrics (red flag if retention <95% or attach <70%), and TRI large-customer renewals or AI product revenue beating/missing guidance; use these triggers to add/trim positions.