Finance

Microsoft Spent $190 Billion on AI This Year — Here's What to Watch in Tomorrow's Earnings Report

Marcus SterlingPublished 3d ago6 min readBased on 16 sources
Reading level
Microsoft Spent $190 Billion on AI This Year — Here's What to Watch in Tomorrow's Earnings Report

Microsoft will release its fiscal year 2026 fourth-quarter financial results on Wednesday, July 29, 2026, after U.S. markets close, followed by its earnings conference call that evening (Microsoft Investor Relations). The report closes out a year in which Microsoft projected $190 billion in capital expenditure, or capex — the money a company spends on physical assets like data centers, servers, and computer chips. That figure exceeded what analysts expected and was driven by AI investments (Bloomberg, Reuters). It matches Alphabet's projected spending for the same period, placing the two tech giants neck-and-neck on the single largest spending item shaping their financial results (Bloomberg).

The spending path through the fiscal year has been steep and uneven. Microsoft reported record capital expenditure of nearly $35 billion in its fiscal first quarter, the three months ended September 30, 2025 (Reuters). Spending then jumped to $37.5 billion in the October–December 2025 quarter, a year-over-year increase of nearly 66% (Reuters). Management guided to a decline in the January–March 2026 quarter off that $37.5 billion peak (Reuters). Net income for that January–March quarter came in at $31,778 million (Microsoft Investor Relations).

What investors will scrutinize tomorrow is whether the Q4 results confirm that the quarterly spending decline signaled a temporary plateau rather than a peak. The composition of spending matters as much as the headline number. Microsoft disclosed that roughly two-thirds of its $37.5 billion in Q2 capital expenditures went toward short-lived assets, primarily GPUs and CPUs (Microsoft Investor Relations). Short-lived assets lose their value faster than long-lived infrastructure like buildings, so they eat into profits more quickly. Think of it like a delivery company buying trucks that wear out in three years versus building a warehouse that lasts thirty — the trucks hit the budget harder and sooner. With about two-thirds of Q2 spending allocated to hardware that depreciates on that faster schedule, the gap between cash going out the door and the profit pressure it creates will be a central focus of the call.

On the revenue side, Microsoft 365 Copilot added 5 million users to its $30-per-month offering through Q3 (Reuters). At an annualized rate, 5 million users at $30 per month translates to roughly $1.8 billion in potential subscription revenue, though how many users actually convert to paying customers and how much companies negotiate discounts will determine the real revenue. Tomorrow's report should clarify whether Copilot adoption is tracking ahead of, in line with, or behind the path management implied at the start of the fiscal year.

The geographic footprint of Microsoft's AI infrastructure buildout expanded materially over the fiscal year. In September 2025, Microsoft announced a $30 billion investment in the UK, including $15 billion in capital expenditures (Microsoft Blog). In December 2025, the company committed $17.5 billion to India for AI diffusion, building on a prior $3 billion investment announced in January 2025 (Microsoft News). In April 2026, Microsoft announced a A$25 billion capital and operational expenditure commitment in Australia by the end of 2029 to expand Azure AI infrastructure (Microsoft News). Also in April 2026, a $10 billion investment in Japan from 2026 through 2029 was announced, covering AI infrastructure, cybersecurity, and workforce (Microsoft News). Earlier, in February 2025, Microsoft announced a PLN 2.8 billion investment in Poland across cloud and AI infrastructure, skilling, and cybersecurity (Microsoft News Poland). Domestically, Microsoft's datacenter investments will contribute $16.5 billion in capital expenditure in Wisconsin (Microsoft Local).

The broader context here is that Big Tech collectively ratcheted AI spending past $700 billion for the year, according to Bloomberg (Bloomberg). Microsoft's $190 billion FY2026 capex projection accounts for roughly 27% of that total, and the Q4 results will be the first full-year accounting of whether the revenue from all this AI spending is growing fast enough to justify the cost of the hardware now dragging on profits.

For markets, the key numbers to watch in tomorrow's release are Azure's growth rate, the cloud segment's operating margin (which is how much profit is left after covering the costs of running the business), and any updated spending guidance for next year. Because so much of Microsoft's spending went toward hardware that loses value quickly, even a modest drop in quarterly capex does not necessarily mean the company is pulling back; it may simply reflect the timing of chip purchases. But if Azure's growth slows while spending stays high, the gap between what Microsoft is investing and what it's earning back will shrink, and that is the metric most likely to move the stock.