Microsoft Q4 Results: $15B Capex Drop Is Accounting Shift
Microsoft’s $15 billion capex reduction is an accounting artifact, not a spending cut. Lease reclassification drives the drop, while actual AI infrastructure spend remains robust with Q1 FY27 capex expected to exceed $50 billion.

*this image is generated using AI for illustrative purposes only.
Microsoft Corp. revised its calendar-2026 capital expenditure forecast downward to roughly $175 billion, a decrease of approximately $15 billion from the $190 billion projected in April. The reduction does not reflect a slowdown in artificial intelligence infrastructure spending but rather an accounting adjustment regarding lease classifications. By extending the estimated useful life of its data centers and office buildings from 15 years to 25 years, effective at the start of fiscal 2027, Microsoft has shifted future data center leases from finance leases to operating leases. Since finance leases are included in reported capex while operating leases are not, this reclassification accounts for the entire $15 billion variance without any change in actual cash outflows for servers, chips, or buildings.
CFO Amy Hood clarified the distinction during the earnings call, stating that outside of this useful-life impact, the company’s calendar year 2026 capex investment expectations remain unchanged. The underlying spending trajectory continues to point sharply upward, driven by demand signals across the portfolio. Microsoft expects fiscal first-quarter 2027 capex, including finance leases, to exceed $50 billion. This follows a robust performance in the fourth quarter of fiscal 2026, where capex and finance leases jumped 69% year-over-year to $41 billion. Hood confirmed that fiscal 2027 capital expenditures will grow year-over-year, reinforcing the company’s commitment to its data-center buildout despite the lower headline number.
What the Numbers Show
The divergence between reported capex and actual investment intent highlights the importance of looking beyond headline figures for hyperscalers. While the $175 billion forecast appears lower than the prior $190 billion estimate, the operational reality is one of continued aggressive expansion. The 69% year-over-year jump in fourth-quarter fiscal 2026 capex to $41 billion demonstrates that execution remains strong. The upcoming fiscal first-quarter 2027 figure exceeding $50 billion further validates that the accounting change is purely structural, not strategic. Investors tracking AI infrastructure spend should note that the conviction in Microsoft’s buildout remains intact, with the bookkeeping shift serving as the sole driver of the apparent cut.
| Metric | Value | Context |
|---|---|---|
| Calendar-2026 Capex Forecast | $175 billion | Down from $190 billion |
| Previous Forecast (April) | $190 billion | Higher due to lease classification |
| Q4 FY26 Capex & Finance Leases | $41 billion | Up 69% year-over-year |
| Q1 FY27 Capex Expectation | >$50 billion | Including finance leases |
| Useful Life Extension | 15 to 25 years | Effective start of FY27 |
Stock Performance
Microsoft stock rose 15.66% to $451.71 on Thursday, reflecting positive sentiment following the earnings report. Over the past month, the stock gained approximately 21.7%, significantly outperforming the S&P 500, which declined by 0.5% over the same period. Year-to-date, however, Microsoft is down roughly 7%, compared to a 7.6% gain for the broader index. The stock trades well above its 50-day moving average of $397.90 and its 100-day moving average of $398.48, indicating bullish momentum. Key resistance is noted near the 52-week high of $555.45, while support may be found around the 200-day moving average at $434.12. Trading volume stood at 51,198,810 shares, surpassing the average volume of 39,212,409, suggesting heightened investor interest in the name.
Will other major hyperscalers like Amazon, Google, or Meta adopt similar lease classification adjustments to manage their reported capex figures?
How might this accounting shift impact analyst valuation models that heavily weight capex-to-revenue ratios for AI infrastructure spending?
Could the extension of useful life estimates from 15 to 25 years signal a change in Microsoft's hardware refresh cycles or data center longevity expectations?

































