Microsoft enterprise AI $37B run-rate

Microsoft just gave investors the cleanest proof yet that enterprise AI is turning into real revenue. Not hype. Not pilot projects. Real money at scale. In its fiscal third-quarter results for the period ended March 31, 2026, Microsoft said its AI business has now passed a $37 billion annual revenue run rate, up 123% from a year earlier, while Azure revenue grew 40%. That matters because the big question hanging over the whole AI trade was simple — are companies actually buying enough AI to justify the infrastructure bill? ### What did Microsoft actually report? Microsoft reported fiscal Q3 revenue of $82.9 billion, up 18%, with diluted EPS of $4.27, up 23%. Microsoft Cloud revenue reached $54.5 billion, up 29%. But the line everyone locked onto was the new AI disclosure: Microsoft’s AI business surpassed a $37 billion annual revenue run rate in the quarter. Microsoft put that figure directly in its earnings materials and repeated it on the earnings call. (microsoft.com) ### What counts as Microsoft’s “AI business”? Microsoft did not break out a full segment P&L for AI. Basically, this is a cross-company number tied to AI infrastructure and AI software sold through Azure, Microsoft 365, Dynamics, GitHub, and related products. The company framed the opportunity as both cloud infrastructure and h(microsoft.com)ding one temporary wave. It reflects AI demand showing up across the stack. (microsoft.com) ### Why is $37 billion such a big deal? Because the last public benchmark was much smaller. Microsoft had previously talked about a $13 billion AI annual run rate in January 2025. Now it is above $37 billion. That is a huge step-up in roughly 15 months, and the 123% year-over-year growth rate says the business is still compounding fas(microsoft.com) way that lets the market see how quickly enterprise AI spending has scaled. (geekwire.com) ### Why does Azure matter so much here? Azure is where a lot of the AI demand lands first. Training, inference, data services, model hosting, and enterprise integrations all pull through cloud usage. Microsoft said Azure revenue grew 40% in the quarter, faster than many analysts expected(geekwire.com) The simplest read is that AI demand is now large enough to move the core cloud business, not just sit on top of it as a flashy add-on. (news.microsoft.com) ### Does this settle the capex debate? Not really — but it changes the tone. The market has been wrestling with whether Microsoft, Amazon, Google, and Meta are overspending on AI infrastructure ahead of demand. Microsoft’s update says demand is very much there, but the catch is that supply still need(news.microsoft.com)mpany builds more capacity for persistent AI demand. (finance.yahoo.com) ### Why do Nvidia and memory names care? Because Microsoft is one of the clearest downstream buyers of the full AI hardware stack. If Microsoft is monetizing AI this quickly, that supports continued orders for GPUs, networking gear, and high-bandwidth memory. Investors tend to read a number like this as evidence that AI i(finance.yahoo.com)stomers are paying for now. That is why the read-through often lands on Nvidia, Broadcom, and memory suppliers almost immediately. This last point is an inference from Microsoft’s disclosed growth and spending trajectory. (microsoft.com) ### So what is the bottom line? Microsoft did not just say AI is popular. It showed that AI is already a business big enough to matter at Microsoft scale. That does not end the debate over margins, capex, or competition. But it does answer the most important question for now — enterprise customers are spending, and they are spending enough to turn AI from a story into a line item.