I Own Nvidia, Microsoft, and Meta. Here's What I'm Doing With All 3 Right Now.

Stocks in focus: Microsoft is down ~21% in 2026 through Mar. 20 after announcing $37.5B capex in fiscal Q2 (66% YoY) while reporting Q2 sales of $81.3B (+17% YoY) and a trailing P/E of 23; the author plans to add to MSFT. Meta posted Q4 revenue of $59.9B (+24% YoY) and is guiding 2026 capex of $115–$135B (from $72.2B in 2025). Nvidia recorded FY2026 revenue of $215.9B (vs. $130.5B prior) and CEO Jensen Huang projects GPU orders of ~$1 trillion by end-2027; overall recommendation is buy-and-hold, with emphasis on adding Microsoft.

Analysis

The market is moving from a broad AI beta to a more discriminating structure: winners will be those with durable pricing power in inference hardware, scale advantages in cloud footprint, and proven unit economics from capex investments; losers will be vendors relying on recurring software multiples or weaker channel pricing. That creates a divergence between capital-intensive infra suppliers (foundries, power/cooling, datacenter REITs) and many high-multiple enterprise software/security names that face churn if customers reallocate budgets to raw compute.

Key near-term risks are demand rephasing and inventory cycles: hyperscaler capex can flip from boom to pause within two quarters if ROI signals are cloudy, creating sharp ASP declines for GPUs and pushing customers to second-source silicon. Over 6–24 months watch three binary catalysts that can reverse the narrative—quarterly cloud utilization trends, Nvidia order cadence/ASP commentary, and geopolitical export/regulatory steps that change addressable markets.

Consensus is underweight the stickiness of cloud-captured AI revenue streams (benefitting large cloud-integrators) but overestimates how quickly capex translates into profitable end-user monetization (risking a multi-quarter re-acceleration lag). That asymmetry argues for convex exposures to Nvidia’s long-term structural story while hedging theme-risk by shorting stretched software/security assets that lack similar balance-sheet optionality; tactically prefer capped-cost option structures and pair trades to control downside and capture large upside if inference adoption continues to scale.

Key Decisions for Investors

• Long NVDA convexity: Buy 18–30 month LEAP call spreads (e.g., long 2027/28 LEAP call, sell a higher strike) sized 1–2% of NAV to capture asymmetric upside while capping premium decay. Target horizon 12–36 months; expect multi-bagger upside if inference volume and ASPs hold, max loss = premium (limited).
• Accumulate MSFT on tactical weakness: Add incremental long exposure on a 10–15% drawdown from today or on two consecutive quarters of cloud utilization/margin beats; position size 2–4% of NAV. Hedge with a small allocation to short-dated downside protection (~3–6 month puts) sized to limit drawdown to ~5% of NAV.
• Pair trade — long MSFT / short cybersecurity beta (e.g., CRWD or ZS): Go long MSFT 2% NAV vs short CRWD 1.5% NAV (dollar-neutral-ish) over 3–12 months to capture rotation from software multiple compression into scale cloud names. Risk: if broad software multiple rebounds, keep stop-loss at 8–10% on the short leg or convert to a put spread to cap losses.
• Short stretched SaaS/security via defined-risk option spreads: Buy 3–6 month 15–25% OTM put spreads on names showing weakest unit economics (e.g., CRWD/ZS) sized 0.5–1% NAV per trade. Reward: >3x potential return if re-rating continues; downside capped to premium paid. Monitor early signs of capex re-acceleration or rapid M&A support which would invalidate the short.