Microsoft Is Now the Cheapest "Magnificent Seven" Stock. Does That Make It a Buy?

Microsoft trades at a 25 P/E, the lowest among the Magnificent Seven and its lowest since 2022, while the stock has been flat over the past year. About 45% of a $625B backlog is tied to OpenAI, and Microsoft spent $49B in capex in H1 FY26 (on track for ~$100B FY), but it holds $89B in liquidity and generated over $97B in free cash flow trailing 12 months; H1 revenue was $159B (+18% YoY) and net income was $66B (+36% YoY). Given the valuation trough and strong cash generation the article views Microsoft as a reasonable buy, but flags material risks from OpenAI concentration and heavy AI capex.

Analysis

Microsoft’s OpenAI linkage functions like a concentrated embedded-call option: it materially leverages Microsoft to any realization of large LLM-driven cloud economics while simultaneously concentrating counterparty and reputational risk. That structure amplifies demand for high-end accelerators and datacenter services (NVIDIA, select ODMs, and colo providers), but it also creates a two-speed market where hyperscalers that can vertically integrate model hosting capture a widening share of long‑duration AI economic rents, squeezing smaller cloud hosts and legacy middleware vendors.

Near-term catalysts will be product cadence, per‑token pricing experiments, and enterprise adoption velocity; all can swing margins and guidance within quarters. Tail risks include licensing/franchise disputes, regulatory interventions around model access or customer data, and supply-chain squeezes for accelerators which could compress gross margins across the stack; conversely, a clear enterprise billing model or large multi‑year hosting contracts would re-rate hyperscaler multiples quickly.

The market appears to be bifurcating between franchise-play and partnership‑risk narratives. That creates specific, asymmetric opportunities: favor capital‑intensive beneficiaries of hyperscaler buildouts and selective long-dated optionality on Microsoft’s service‑monetization pathway, while expressing skepticism toward vendors whose OpenAI ties are valuation-relevant but operationally marginalized. Execution should emphasize time‑staggered entries to capture near‑term dislocation and longer‑dated payoff if AI monetization follows the stronger adoption curve.

Key Decisions for Investors

• Pair trade (3–9 months): Long NVDA equity (or 3–6 month call spread) / Short ORCL equity — rationale: NVDA captures accelerator scarcity and price inelasticity, ORCL carries partnership execution risk and sentiment hole. Target R/R ~2.5:1 (expect 30–50% upside on NVDA leg vs 15–25% downside on ORCL); size 2:1 notional to the long leg. Hard stop: 18% adverse move on NVDA or 12% adverse move on ORCL, reassess on earnings.
• Long MSFT asymmetric option (12–24 months): Buy long-dated LEAP call or a debit call spread 12–24 months out to own convexity to AI monetization while limiting capital at risk — consider funding by selling a nearer-term call (3–6 months) to finance cost. Target payoff: >2x if enterprise AI monetization accelerates; max loss limited to premium. Exit/trim triggers: material negative OpenAI litigation outcome or 20% drawdown in Azure subscription momentum.
• Tactical hedge (0–6 months): Buy ORCL near-term puts or sell small size of ORCL near-term calls to express asymmetric downside view on partnership contagion while keeping overall tech exposure constructive. Use as a low-cost insurance to the long hyperscaler/AI positions; position size should be <=25% of total long AI exposure, with profit target 100% if ORCL guidance misses materially.