3 Artificial Intelligence (AI) Software Stocks I'd Buy Today if I Were Starting From Scratch

Palantir reported Q4 FY2025 revenue up 70% YoY to $1.41B with U.S. commercial revenue +137% to $507M and U.S. government revenue +66% to $570M; management expects U.S. commercial to exceed $3.14B in FY2026 (implying ≥115% YoY). Salesforce's Agentforce reached ~$800M ARR in FY2026 (up 169% YoY) with 29,000 deals in 15 months and increases in large deals (> $1M +26% YoY; > $10M +33% YoY). Microsoft exited Q2 FY2026 with ~15M paid Microsoft 365 Copilot seats (up >160% YoY) and 4.7M GitHub Copilot subscribers, and is rolling out an E7 tier priced ~65% above E5 to drive ARPU and consumption-based AI monetization.

Analysis

The move from pilots to production creates a new class of platform winners where data-model-workflow integration — not model quality alone — becomes the primary switching cost. Palantir’s ontology approach and short-cycle AIP workshops accelerate a feedback loop: faster deployments -> more labeled operational context -> lower model drift and higher marginal ROI per seat, which compounds gross retention and expands TAM for adjacent tooling (data governance, connectors, vertical apps).

Second-order beneficiaries include data plumbing and governance vendors that sit between legacy ERPs and LLMs (Informatica-style plays), and hybrid-cloud infra that supports on-prem inference and sensitive data handling (Azure + identity stacks). Conversely, pure-play inference hardware vendors face a paradox: they remain indispensable for training and latency-sensitive inference, but much of the economic value is migrating to platform software via subscription and usage pricing, compressing hardware capture of long-term AI spend.

Principal risks are non-linear: an operational hallucination or high-profile data breach could produce immediate procurement freezes and multi-quarter churn; a renewed macro capex pullback would cause customers to suspend new agent rollouts even as existing subscriptions persist. Time horizons separate signals: deal-count and ARR composition inflections show up in quarterly cadence, but true margin and TAM capture play out over 12–36 months as agents move from assistants to mission-critical workflows.

Key Decisions for Investors

• Long PLTR equity (6–24 month horizon), position size 2–3% portfolio. Hedge with a 6–9 month put ~25% OTM to cap downside during early commercialization volatility. Rationale: asymmetric upside if platform embeds into ops across two or more verticals; downside limited vs outright long.
• Bull call spread on CRM (buy 12-month near‑ATM call, sell 12‑month ~35–45% OTM) sized 2% portfolio. This captures continued Agentforce monetization and Informatica-driven data moat with defined cost; reward capped but 3–4x on premium if adoption accelerates, loss limited to premium paid.
• Directional pair: long MSFT via cash-secured short 3–6 month puts (collect premium) and hedge by buying 6–12 month INTC puts (or short INTC stock) in 1:1 vega-adjusted notional. Timeframe 3–12 months; thesis: software/consumption capture > cyclical silicon rerating. Risk: NVDA‑led hardware rally lifts INTC; keep position small (1–2%).
• Tactical long INFA (6–12 month) — buy shares or long calls sized 1–2% as a play on secular demand for trusted data plumbing. Rationale: improves adoption velocity for enterprise agents; downside is execution/integration risk with large CRM customers.