Microsoft Hasn't Been This Cheap Since 2018 - And History Says This Comes Next

Microsoft is rated Strong Buy and trades at 21.88x TTM P/E, its lowest multiple since 2018, despite Azure growing 40% and commercial RPO reaching $627B. About 25% of that backlog converts within 12 months, supporting $190B of CapEx and multi-year revenue visibility. A meaningful share of the backlog is tied to OpenAI, which adds some earnings volatility as Microsoft absorbs part of OpenAI's ongoing losses.

Analysis

The market is treating MSFT like a duration proxy, but the real setup is closer to a balance-sheet-backed infrastructure monopolist with embedded option value on AI demand. The backlog provides unusually visible conversion into revenue, which should support a high CapEx cadence without the usual penalty of “growth spending before monetization”; that reduces the probability of a near-term multiple compression even if free cash flow lags. The more important second-order effect is that AI winners across the supply chain likely remain under-discounted relative to MSFT itself: model training, inference, and data-center buildout create persistent demand for power, networking, storage, and GPU-adjacent infrastructure that should keep capital flowing even if software growth normalizes.

The hidden risk is not demand, but earnings quality. If a meaningful portion of the backlog is tied to OpenAI economics, MSFT has effectively written a call option on another company’s burn rate, which can inject quarter-to-quarter noise into gross margin and operating income. That matters because the current multiple is being justified by visibility; any sign that AI-linked revenue is accreting slower than infrastructure spend, or that loss-sharing expands, could force a de-rating before the backlog itself becomes a problem.

Consensus is probably underestimating how much of the upside is already in the backlog and how much of the downside sits in the cost base. This is not a straight-line earnings comp story for the next 2-3 quarters; it is a 12-24 month capital allocation story where the market may temporarily reward revenue durability but eventually demand proof of incremental ROI on the $190B spend. The setup is constructive, but the asymmetry is better in adjacent beneficiaries and in hedged exposure to MSFT’s AI capex execution risk than in an outright chase at current levels.

Key Decisions for Investors

• Maintain a core long MSFT position, but size it as a quality compounder rather than a high-beta AI expression; use pullbacks after earnings or capex headlines to add over a 3-6 month horizon.
• Prefer a pair trade: long MSFT / short a basket of AI infrastructure spenders with weaker balance sheets or inferior backlog visibility, to isolate execution quality while reducing multiple risk.
• Buy call spreads on semiconductor/power/networking beneficiaries over the next 6-12 months, since the second-order capex wave is more levered there than in MSFT itself; risk/reward is better if AI buildout persists.
• Hedge MSFT event risk around earnings with short-dated puts or put spreads if management commentary implies rising OpenAI-related loss absorption; the catalyst window is 1-2 quarters, not years.
• Avoid aggressive chasing above the current multiple until there is evidence that AI-related CapEx is translating into margin preservation, not just backlog growth; otherwise the stock can drift sideways despite strong fundamentals.