2 AI Stocks That Are Cheaper Today Than They Were on Jan. 1 -- for No Good Reason

Both Amazon and Meta are ~10% down year-to-date, but Wells Fargo strategist Ohsung Kwon now expects free cash flow to come in above consensus and strong revenue growth as AI spending drives demand. Amazon’s heavy AI data-center buildout and partnerships with Anthropic/OpenAI should accelerate AWS growth and improve e-commerce operating profits; the stock trades at a forward P/E below 27 and a forward 1-year P/E of ~22 (2027 estimates). Meta’s AI-driven ad recommendation improvements and advertiser tools are creating a powerful revenue flywheel, with the stock trading at a forward P/E below 20 and a forward 1-year P/E of ~16.5 (2027 estimates).

Analysis

The direct lever in play is accelerator capacity and the software layer that monetizes it; firms that own model deployment, fine‑tuning marketplaces and billing/observability will capture >2x the margin of pure raw compute providers because they convert one‑time training spend into recurring inference fees and services. Expect regional cloud and colocation vendors to experience margin compression first — likely driving 2–4 meaningful M&A transactions in the next 12–24 months as scale becomes the dominant defense against rising unit costs.

Two high‑impact, short‑to‑medium term catalysts will determine realized returns: (1) hardware supply and pricing (spot GPU price curves through H2 2026) and (2) rack‑level utilization efficiency (sustained utilization above ~60–70% for two consecutive quarters is the practical breakpoint for operating leverage to show up in free cash flow). A sudden improvement in FLOPs/$ from model compression or novel inference chips could shift the FCF inflection forward by 6–12 months; conversely, supply shocks or an advertiser demand pullback would push payback horizons beyond 18 months.

The market is binary about capex: either “waste” or “instant FCF.” The nuance is cadence — capex secures optionality that monetizes via high‑margin enterprise contracts and proprietary model services, a conversion that historically takes 18–36 months and is lumpy. Tactical positions should therefore be time‑boxed to that window and explicitly hedged to hardware cycles and CPM/ARPU volatility rather than pure headline sentiment about “AI spend.”

Key Decisions for Investors

• Enter a 12–24 month bullish, financed-leverage on AMZN (size 2–3% NAV): buy Jan‑2028 LEAP calls ~25% OTM and sell Jan‑2027 calls ~30% OTM to pay down premium. Rationale: asymmetric upside if rack utilization/enterprise contracts accelerate; max loss = net premium (~2–3% NAV), target = 2x–3x premium if FCF inflection becomes visible within 18 months, stop-loss if two consecutive quarters miss on cloud utilization metrics.
• Construct a 9–12 month call spread on META (size 1–2% NAV): buy 12‑month ATM call and sell 12‑month call ~30% above to fund. Rationale: captures ad monetization upside while capping cost; target 30–50% return if sequential CPMs and ARPU rebound, max loss = premium paid, cut exposure if advertiser counts fall for two straight quarters or regulatory ad‑targeting constraints escalate.
• Small, asymmetric contrarian on INTEL to play hardware re‑rating (size 0.5–1% NAV): buy 12‑18 month call spread (ATM to +40% strike) to limit premium. Rationale: if market needs diversify beyond GPU dominance or Intel shows credible datacenter roadmap, upside is >3x; downside limited to premium if NVDA retains exclusivity on accelerator economics.
• Macro hedge: buy 6–12 month WFC puts ~10% OTM (size 0.5–1% NAV) to protect against a macro shock that would hit ad budgets and enterprise cloud spending simultaneously. Rationale: preserves capital in severe downside scenarios where both monetization and FCF timelines slip beyond 18 months; cost is small insurance versus equity positions.