Microsoft shares had their biggest one-day jump in 18 years on July 30, 2026. The rally capped a year in which the company spent roughly $145 billion building out its infrastructure and its AI business crossed $37 billion in yearly recurring revenue. Wall Street finished sharply higher that Thursday, with chip stocks rising and Microsoft shares soaring, as the company's fiscal Q4 results and its outlook for the months ahead offset broader economic worries (Reuters. The rally came after a rough stretch: Microsoft shares were down roughly 17% for the year as of late July, trailing the broader market (Yahoo Finance).

The Spending Story

Microsoft's spending on equipment and infrastructure climbed steeply through fiscal year 2026. Q1 came in at $34.9 billion, driven by demand for cloud and AI offerings, with roughly half going to short-lived assets such as GPUs and CPUs (Microsoft Investor). Q2 rose to $37.5 billion, with about two-thirds directed at short-lived hardware (Microsoft Investor). Q3 dipped to $31.9 billion, which Microsoft attributed to normal variability from cloud buildouts and the timing of equipment delivery (Microsoft Investor). Q4 then spiked to $41 billion, reflecting higher component pricing (Microsoft Investor).

The Q4 detail about higher component prices is worth pausing on. When a company as large as Microsoft says parts are getting more expensive, it suggests that chip and server makers have gained some pricing power. Whether that pressure continues into fiscal 2027 will depend on competition among chip suppliers and how quickly next-generation chips arrive.

The AI Revenue Engine

Microsoft's AI business surpassed $37 billion in annual recurring revenue in Q3 of fiscal 2026, up 123% year-over-year (Microsoft Investor). Growing that fast on a base already in the tens of billions is unusual. The "recurring revenue" label matters: it means subscription-like, contracted revenue rather than one-off spikes, which tends to be more dependable when projecting future earnings.

OpenAI and Earnings Mechanics

Microsoft's stake in OpenAI remained a small drag on reported profits. Net losses from that stake reduced net income by $14 million in Q3 (Microsoft Investor). Q3 diluted EPS — earnings per share, a common way to gauge profitability — came in at $4.27 (Microsoft Investor). In Q4, Microsoft provided guidance on April 29, 2026 that resulted in a $0.27 benefit to diluted EPS, and the quarter included a $3.2 billion gain (Microsoft Investor).

That $3.2 billion gain and the $0.27 EPS benefit materially shaped the Q4 headline numbers that drove the July 30 rally. Take those items away, and the underlying story — strong AI revenue growth offset by accelerating cost growth — is more complicated.

Shareholder Returns and Forward Guidance

In Q2 alone, Microsoft returned $12.7 billion to shareholders through dividends and share buybacks (Microsoft Investor). That pace of cash return, sustained alongside spending of more than $35 billion per quarter, means Microsoft was funding the largest infrastructure build in its history while returning cash at a rate exceeding $50 billion annually if the Q2 pace held.

Microsoft's Q1 fiscal 2026 guidance, issued October 29, 2025, projected revenue of $79.5 to $80.6 billion — 14% to 16% growth. The company also projected cost of goods sold, or COGS, of $26.35 to $26.55 billion — 21% to 22% growth (Microsoft Investor).

The cost growth rate outpacing revenue growth by roughly 6 to 8 percentage points points directly to the margin squeeze that rising AI infrastructure costs impose. When costs grow nearly twice as fast as sales, shrinking profit margins follow as simple math, not speculation.

What the Rally Resets

The July 30 surge matters less as a valuation event and more as a shift in mood. A stock down 17% for the year that then posts its largest single-day gain in nearly two decades is a market repricing expectations with unusual force. The trigger was Q4 results that included a $3.2 billion gain and a $0.27 EPS benefit from guidance issued in April.

The central question is whether the $37 billion AI revenue run-rate, growing at triple digits, can eventually absorb the cost of replacing all that short-lived hardware. GPUs and CPUs typically wear out on a three-to-five-year schedule, creating a structural mismatch: AI revenue has to grow fast enough to outpace the rate at which those assets lose their value on the books. Microsoft's own guidance showed costs growing 21% to 22% against revenue growth of 14% to 16%. That gap is the number to watch.