The biggest reason is still Azure. In Q3, Azure and other cloud services revenue grew 40% year over year, and Microsoft said it expects Q4 Azure growth of 39% to 40% in constant currency (Microsoft Q3 earnings call, April 29, 2026). That matters because it suggests the pace is not fading just as the market is trying to decide whether the AI infrastructure cycle has room left.

The broader cloud business was strong too. Microsoft Cloud revenue reached $54.5 billion in Q3, up 29% from a year earlier (Microsoft Q3 earnings call, April 29, 2026). Back in Q2, Microsoft Cloud revenue was $51.5 billion, up 26%, and commercial remaining performance obligation had reached $625 billion, up 110% year over year (Microsoft Q2 press release, January 28, 2026). That backlog is not a perfect forecast, but it does show a pipeline that stretches well beyond one quarter’s enthusiasm.

This is the part of the story that tends to get reduced to “AI hype,” which is a lazy way to miss what the numbers are actually saying. Cloud growth at this scale, with guidance staying hot into the next quarter, suggests Microsoft is not merely riding a mood. It is monetizing the thing investors have been hoping would show up in the income statement.

The earnings call also showed that AI at Microsoft is no longer a single-product bet. More than 10,000 customers have used more than one model on Foundry, 5,000 have used open source models, and the number using both Anthropic and OpenAI models doubled quarter over quarter (Microsoft Q3 earnings call, April 29, 2026). Microsoft said over 300 customers are on track to process more than one trillion tokens on Foundry this year, up 30% quarter over quarter (Microsoft Q3 earnings call, April 29, 2026).

The adoption data runs through the rest of the portfolio too. Nearly 90% of the Fortune 500 now have active agents built with Microsoft’s low-code and no-code tools, first-party agent usage is up 6X year to date, and Copilot queries per user rose nearly 20% quarter over quarter (Microsoft Q3 earnings call, April 29, 2026). Microsoft also said it introduced 625 updates to Microsoft 365 Copilot over the past year, up 50%, and that Agent Mode became the default across Copilot in Word, Excel, and PowerPoint as of last week (Microsoft Q3 earnings call, April 29, 2026).

That breadth matters because it reduces the risk that the AI story lives or dies with one product. Fabric had 35,000 paid customers, up 60% year over year, and Cosmos DB revenue rose 50% year over year, driven by AI app workloads (Microsoft Q3 earnings call, April 29, 2026). LinkedIn’s agentic products in Talent Solutions have already surpassed a $450 million annualized revenue run rate (Microsoft Q3 earnings call, April 29, 2026). That is not just a chatbot selling point. It is a platform business starting to show up in several billing lines at once.

Here is where the bull case gets harder, and more interesting. Microsoft is spending heavily enough that the bill would make most companies sweat. Capital expenditures were $31.9 billion in Q3, and the company said it expects roughly $190 billion in capital expenditures for calendar year 2026, including about $25 billion tied to higher component pricing (Microsoft Q3 earnings call, April 29, 2026). It also said CapEx spend should rise to over $40 billion as more capacity comes online (Microsoft Q3 earnings call, April 29, 2026).

The margin picture is mixed, which is why this section matters. Gross margin was 68%, down year over year, as AI infrastructure costs and heavier product usage weighed on profitability. Yet operating margins still increased slightly to 46%, and cash flow from operations rose 26% to $46.7 billion (Microsoft Q3 earnings call, April 29, 2026). Free cash flow was $15.8 billion, lower because of the spending burden, but the company still returned $10.2 billion to shareholders through dividends and buybacks (Microsoft Q3 earnings call, April 29, 2026).

That is the tension investors are voting on. Skeptics of AI infrastructure spending can point to the gap between capex and free cash flow and say the economics are still being written in pencil. Supporters point to Azure growth at 40%, the $625 billion backlog from Q2, and the fact that Microsoft is increasing capacity while also getting smarter about deployment. The company said it brought its Fairwater datacenter in Wisconsin online earlier this month, six weeks ahead of schedule, and cut dock-to-live times for new GPUs in its biggest regions by nearly 20% since the beginning of the year (Microsoft Q3 earnings call, April 29, 2026).

My read is that investors are comfortable paying up because the spend is starting to look productive rather than wasteful. That is not the same as saying the market has it right forever. It just means the current numbers give the spending story enough credibility to keep the stock bid.

The OpenAI investment is the easiest place to get confused, because Microsoft’s reported earnings have been affected by it in different directions in recent quarters. In Q2 FY2026, net income and diluted earnings per share were lifted by net gains from investments in OpenAI, which increased net income by $7.6 billion and diluted EPS by $1.02 (Microsoft Q2 press release, January 28, 2026). That same quarter, GAAP net income was $38.5 billion, up 60%, while non-GAAP net income was $30.9 billion, up 23% (Microsoft Q2 press release, January 28, 2026).

Q3 was much cleaner. Microsoft said net losses from its OpenAI investment reduced net income by just $14 million and had minimal impact on EPS (Microsoft press release, April 29, 2026). On an adjusted basis, EPS was $4.27, up 21%, while GAAP EPS was also $4.27, up 23% (Microsoft press release, April 29, 2026; Microsoft Q3 earnings call, April 29, 2026). That makes Q3 a better read on the core business than Q2, which had the benefit of a large accounting lift.

This is important because it strips away a common excuse. The quarter was not good just because of one investment line. It was good because Microsoft Cloud, Azure, Copilot, Fabric, Dynamics, and the rest of the machine kept pulling in the same direction.

The cleanest answer to why Microsoft stock is going up is that the company turned AI from a promise into a quarter that looked measurable, broad, and profitable enough to matter. Revenue rose 18%, operating income rose 20%, AI ARR hit $37 billion, and Azure growth stayed at 40% (Microsoft press release, April 29, 2026; Microsoft Q3 earnings call, April 29, 2026). That combination is why the Microsoft stock rally has not looked like a simple post-earnings bounce.

The next test is less flattering. Microsoft itself said Q4 Azure growth should stay between 39% and 40% in constant currency, while CapEx moves above $40 billion as more capacity comes online (Microsoft Q3 earnings call, April 29, 2026). If growth holds and margins stay near current levels, the market will probably keep treating this as a new baseline. If the spending outruns the revenue, the same rally will start to look a lot less like conviction and a lot more like a very expensive act of faith.