2 AI Stocks Down 20% or More That Are Screaming Buys

Microsoft is trading about 30% below its all-time high despite fiscal 2026 Q2 revenue rising 17% year-over-year and Azure revenue climbing 39%; operating P/E is near decade lows and the author recommends buying. Meta is down ~23% from its peak but reported Q4 revenue up 24% y/y; the market is worried that its $115–$135 billion data-center capex plan this year (funded by operating cash flow) could be excessive, though the author argues underspending risk is greater and remains bullish.

Analysis

MSFT’s pullback has created optionality asymmetry rather than a pure fundamentals story — its earnings moat plus embedded OpenAI linkage is a multi-year convexity to enterprise AI spend. If Microsoft converts even a low-single-digit percentage of global IT budgets into premium AI services over 12–24 months, operating leverage will outpace where multiples have re-set, producing >25–35% total return from current levels while downside is cushioned by recurring software annuities.

Meta’s capital-intensive sprint to own the AI stack amplifies demand for datacenter compute and interconnect hardware, tightening supply for GPUs, networking ASICs, and power delivery over the next 6–12 months. That creates a staggered benefit cycle: NVDA and systems suppliers see near-term revenue upside, while colocators and energy-intensive infrastructure players capture longer-duration cash flows as buildouts mature.

Key near-term catalysts to watch are concrete monetization steps (enterprise contracts/licensing), public product rollouts, and forward guidance on hardware demand; these will move prices in weeks to quarters. Tail risks include AI regulation, a material market-wide technology drawdown from macro/rates, or an execution failure on high-commitment projects — any of which can reprice multiples rapidly within 3–6 months.

Consensus underestimates two second-order outcomes: (1) crowding of GPU demand creating transient pricing power for suppliers over the next 6–9 months, and (2) the asymmetric payback on diversified platform owners (MSFT) versus single-outcome bets (META). That argues for convex, time-limited exposure to MSFT and targeted, hedged exposure to infrastructure beneficiaries rather than undifferentiated long risk in single capex-heavy names.

Key Decisions for Investors

• Buy MSFT stock (size: 3–5% portfolio) with a 12-month horizon — target 30–40% upside if AI monetization ramps; place a tactical stop-loss at -12% from entry to protect against multiple compression from macro shock.
• Replace a straight NVDA long with a 3–6 month call spread (buy 1x ATM call, sell 1x OTM call ~20–25% above) to capture GPU-driven upside while financing premium — allocate 1–2% portfolio; max loss = premium (~100% of allocation), payoff capped but >2x if NVDA prints material upside on guidance.
• Initiate a paired trade: long MSFT (2% portfolio) / buy META 6–9 month put spread (sell nearer-dated put to reduce cost; net debit ~0.5% portfolio) — this creates long-convexity to platform execution while capping downside exposure should Meta’s capex fail to produce near-term returns; target pair payoff >2:1 if MSFT re-rates and META underperforms.
• Tactical value play in INTC: buy shares or 12-month LEAP calls (delta ~0.35) sized 1–2% portfolio to capture potential re-leveraging from industry capex and foundry upside; expect 40–60% upside on successful execution, with execution risk concentrated over 12–24 months — pare down position at 30–40% realized gain or if guidance slips materially.