Microsoft Earns 31% More Than It Did a Year Ago and Is Worth Less

Microsoft fiscal 2026 results were strong: net income +31% to $133.7B on revenue +18% to $331.8B, with Azure revenue topping $100B (+41%). However, the article flags a cash-flow mismatch: planned calendar 2026 capex/finance leases of ~$175B exceeds fiscal 2026 operating income of $155.2B, pulling free cash flow to about $67B (down from ~$72B). Despite a large $678B RPO backlog (+84% YoY) and growth expectations, the market appears to be pricing in higher uncertainty tied to generative AI spending and data-center payoff, leaving the view closer to “hold than buy.”

Analysis

The market is no longer rewarding earnings growth at the same multiple when it comes with a very large incremental capital burden. That creates a subtle but important hierarchy: capital-light cash generators such as AAPL should continue to screen as “quality compounders,” while MSFT gets treated more like an infrastructure utility whose returns depend on utilization staying high enough to amortize the buildout. The second-order winner is the AI hardware stack: every quarter of elevated cloud capex is a direct demand tailwind for NVDA and, by extension, the broader semiconductor capex complex.

The near-term risk is not that Microsoft’s business weakens; it is that free cash flow stays visually unimpressive versus headline earnings for another 1-2 quarters, which keeps the multiple compressed. The real reversal catalyst is evidence that capex intensity peaks while remaining performance obligations convert into revenue cleanly over the next 2-4 quarters; if that happens, the market will likely re-rate the stock before the sell side fully updates estimates. A failure case is simple: if spending stays elevated but Azure growth normalizes into the high-20s/low-30s, the stock can de-rate further even with strong reported EPS.

Contrarian view: consensus is probably over-focusing on the absolute size of the spend and underweighting the fact that MSFT still converts a huge amount of operating income and has unusually visible demand. The better question is not whether capex is high, but whether the incremental return on that capex stays above the cost of capital; if yes, the market is likely underappreciating the durability of the AI infrastructure buildout. If no, this becomes a years-long multiple problem rather than a one-quarter headline issue.

Key Decisions for Investors

• Long AAPL / short MSFT as a 1-3 month pair trade: own the capital-light compounder versus the capital-intensive AI platform. Base case is continued valuation divergence; invalidate if MSFT guides capex meaningfully below the current run-rate or free cash flow re-accelerates above ~$80B annualized.
• Long NVDA against MSFT on any pullback in the next 2-6 weeks: elevated hyperscaler capex is the cleanest upstream demand signal for AI accelerators. Risk is that MSFT shifts toward more custom silicon or slows data-center additions, which would compress NVDA’s relative upside.
• Avoid initiating fresh outright MSFT longs until the market sees at least one quarter of capex stabilization and conversion of backlog into visibly stronger FCF. If the stock re-tests the prior year’s relative-low valuation band without a downward revision to Azure growth, that would be the more attractive entry.
• Use IGV/XLK weakness as a hedge against the “AI capex crowding out software multiple” theme: if enterprise software keeps trading on margin fear, the basket remains vulnerable even when reported growth holds. Cover if software revenue durability improves or if AI monetization starts to show through in net retention/expansion metrics.
• Watch the next capex guide as the key falsifier: a step-down in 2027 spending would be bullish MSFT and bearish NVDA relative, while another upward revision would reinforce the current relative-value gap.