3 Artificial Intelligence (AI) Stocks to Buy at a Discount

Micron: shares are up ~300% over the past year but still trade at ~6.8x forward EPS; revenue was $23.9B last quarter (prior quarter $13.6B, year-ago $8B) with management expecting $33.5B next quarter and able to fill only ~50%–66% of orders, implying sustained pricing upside if shortages persist. Nvidia: Q4 revenue rose 73% YoY and management projects ~77% growth next quarter, with the stock trading ~21.1x forward earnings versus the S&P 500 at ~20.6x—suggesting a valuation-supported growth opportunity. Microsoft: revenue grew 17% YoY and EPS rose ~60% (boosted by OpenAI-related gains); using operating P/E the stock is at multi-year lows, making it a recommended buy for long-term AI exposure.

Analysis

The memory shortage is creating asymmetric outcomes across the hardware stack: hyperscalers and GPU vendors win immediate economics but OEMs with fixed BOMs and lower pricing power will see margin pressure, accelerating component substitution and software-level memory optimization projects. That creates a transient pricing umbrella for incumbent DRAM suppliers, but also a predictable incentive for aggressive capex from competitors and foundries within 12–24 months — precisely the horizon where the current pain trade can reverse into oversupply.

For AI accelerators, entrenched architecture advantages (software ecosystem, model certification, tooling) are as important as raw silicon. That makes market share stickier for incumbents than on pure-performance metrics, but it also raises regulatory and geopolitical fragility: export controls or localized cloud stacks could bifurcate growth and leave priced-in global assumptions exposed. For Microsoft, OpenAI exposure is a demand amplifier but also a margin & execution risk — monetization cadence, model-hosting economics, and customer procurement cycles will determine whether the investment converts to durable FCF or transient re-rating.

Trade implementation should therefore be duration-mixed: capture the near-term squeeze in memory with stock + limited-risk options, anchor AI-platform exposure with multi-year LEAPS, and harvest yield on large-cap software positions via covered calls to mitigate downside. Size positions to reflect binary outcomes — memory can mean-revert quickly once capex responds, while platform winners reprice over years.

Contrarian view: consensus treats these names as a single ‘AI’ bucket; the material difference is cyclicality (memory) versus structural moat (accelerators, platforms). If capex and policy shocks arrive together, you could see a rapid de-risking in MU and a delayed but sharp re-pricing of NVDA/MSFT depending on export/regulatory paths — prepare trigger-based exits rather than time-based holds.

Key Decisions for Investors

• MU — Tactical long: Buy MU shares to 1.5% portfolio weight on a two-tranche basis (50% now, 50% on a 10–15% pullback). Time horizon 12–24 months. Rationale: capture ongoing supply tightness; downside capped by cyclical risk. Risk management: stop-loss at -25% from entry; target +60–80% if shortage persists past next two quarters.
• MU — Defined-risk options: Buy 9–15 month call spread (long nearer-the-money call, short higher strike) sized at 0.25% portfolio risk capital. This structure captures a sharp rally while limiting premium loss if memory pricing normalizes. Expected payoff 3–6x on premium if squeeze continues, max loss = premium.
• NVDA — Strategic LEAPS: Buy 18–36 month call LEAPS equal to 1–2% portfolio notional (or purchase shares if preferred), financed by selling short-dated calls (calendar financing). Time horizon 24–36 months. R/R: asymmetric upside (>100% if secular AI compute growth continues) with capped near-term financing. Monitor: export-control headlines and hyperscaler capex cadence as primary catalysts.
• MSFT — Income + optional upside: Accumulate MSFT to 2% portfolio and sell 6–12 month covered calls at a modest premium to current levels to harvest yield while retaining upside. Time horizon 12–24 months. Rationale: hedge valuation risk from event-driven OpenAI earnings while participating in platform monetization; unwind covered calls on positive AI monetization beats.