Microsoft says the way it finances the tens of billions of dollars it spends on new data centers and AI infrastructure is to avoid customers who want to rent GPUs to train new AI models.
CEO Satya Nadella unveiled the strategy Wednesday during the software giant’s Q1 2025 earnings call, in which executives discussed how the business will generate quarterly revenue of $65.6 billion and net profit. 24.7 billion, an improvement of 16% and 11%, respectively.
Microsoft’s Intelligent Cloud division covers server products and cloud services and is the City of Redmond’s largest single source of funding, with revenue increasing 20% year-over-year to $24.1 billion. Azure and other cloud services grew 33%. Azure Arc, a multi-cloud and hybrid cloud management tool, was highlighted for gaining 39,000 customers.
AI was naturally at the forefront of the earnings call, with Nadella pointing out that Microsoft’s AI business is on track to hit an annual operating rate of $10 billion next quarter. If that happens, it will be the fastest new product in Microsoft history.
Investment analysts invited to ask questions during the call noted that Microsoft is building the massive infrastructure needed to deliver its AI services ($20 billion this quarter alone, most of it in data centers and servers). wanted to know how they were being paid. Nadella and Chief Financial Officer Amy Hood said supply chain delays have sometimes made it difficult to bring new infrastructure online to meet demand, impacting profit margins. he said. But as new AI capabilities come online, so too will revenue.
Another way Microsoft is managing things is by turning off training workloads.
“We’re not really selling raw GPUs for other people to train on,” Nadella said. “In fact, this is the kind of business that we turn down because there is so much demand for inference to power various copilot and other AI services.”
“We literally don’t even really participate in most of it because we’re addressing the real demand in the enterprise space and in our own products like GitHub Copilot and M365 Copilot. So our revenue… “I feel like the quality is low,” Nadella added.
CFO Amy Hood added that Microsoft sees revenue from inference as generating funds to pay for future model training efforts.
But while focusing on reasoning may be a cunning plan, it doesn’t help control costs. Despite Microsoft’s best efforts, costs rose 12% this quarter. Some of the increase was due to the addition of Activision staff, but overall headcount increased by 2%.
Most of Microsoft’s business areas recorded strong growth. LinkedIn is 10%. 14% in dynamics; Xbox was up 61% (thanks to the Activision acquisition). and search and advertising decreased by 17%. Microsoft 365 commercial revenue increased 13% and consumer revenue increased 5 points.
However, Windows OEM and device revenue was down 2% and was expected to decline again.
If that worried Microsoft leaders, they didn’t show it. Additionally, there were plenty of other wins to celebrate, separating the names of blue-chip companies that have brought Microsoft’s AI products to life. CFO Hood also noted that “both Azure and Microsoft 365 are seeing an increase in the number of contracts over $10 million,” and “Azure is seeing an increase in the number of contracts over $100 million.” I did.
All these wins leave Microsoft with $259 billion in future revenue, which its customers have already committed to. Technology Leviathan also expects strong growth across most of its major product lines, with Azure expected to grow 31 to 32 percent and productivity and business process products expected to grow 10 to 11 percent.
However, I did warn you that some small points may become dull in the future. One is that revenue from on-premises server products may decline. However, this is not a sign of decline, but rather reflects the fact that this time last year, Redmond was in the final weeks of selling long-term support for Windows Server 2012.
Similarly, the expected slight decline in Microsoft 365 revenue was also attributed to business cycles.
Microsoft’s stock price soared from about $435 to about $442 in after-hours trading, before settling at about $417. This seems to indicate that investors are hoping for better news on costs, but are taking comfort in the fact that AI is paying off. ®
