Microsoft budgets $190B capex 2026
- Microsoft told investors on April 29 that 2026 capital spending will hit $190 billion as it races to add Azure and AI capacity. - The standout detail is cost pressure, not just ambition — CFO Amy Hood said roughly $25 billion of that capex jump comes from pricier parts. - That matters because Microsoft spent about $80 billion in fiscal 2025; the new plan shows the AI buildout is accelerating, not easing.
Cloud infrastructure is the story here. Microsoft did not just post a strong quarter on April 29 — it also told investors to expect a huge new wave of spending in 2026, with capital expenditures reaching about $190 billion. That is an eye-popping number even by hyperscaler standards. And the interesting part is that this is not only about demand. It is also about the cost of building the machines that feed that demand. (cnbc.com) ### What actually changed? The new thing is the size of the budget. Microsoft’s fiscal third-quarter results were strong — $82.9 billion in revenue, up 18%, and Azure growth guidance for the next quarter came in above expectations. But the bigger surprise was the 2026 capex outlook. Microsoft said it now expects $190 billion in capital (cnbc.com)cnbc.com) ### Why is the number so high? Part of it is simple: AI demand is still roaring. Satya Nadella said Microsoft’s AI business has now passed a $37 billion annual revenue run rate, up 123% year over year. If customers keep buying Azure AI capacity, Microsoft has to keep adding servers, networking gear, and data center space. But part of th(cnbc.com) plan reflects higher component prices, especially memory. (microsoft.com) ### Why does memory matter so much? Because AI systems are memory-hungry in a very literal way. Training and serving large models needs advanced chips, but those chips are only useful if they are paired with enough high-bandwidth memory. When memory gets scarce, the whole stack gets more expensive. That is basically what Mic(microsoft.com)te has gone up. (cnbc.com) ### Is this just a Microsoft thing? No — that is what makes the story bigger. Alphabet also raised its 2026 capex outlook on April 29, to as much as $190 billion, while saying cloud demand remains capacity-constrained and 2027 spending should rise again. So this is starting to look less like one company overspending and more like the new baseline for the top AI infrastructure players. (cnbc.com) ### Did Microsoft’s quarter support that spending? Mostly, yes. Microsoft beat on revenue and earnings, Azure growth stayed strong, and the company said Microsoft Cloud revenue reached $54.5 billion, up 29%. Commercial remaining performance obligation climbed to $627 billion, which is a useful signal that enterprise demand is still stacking up(cnbc.com)eciation from all this infrastructure started to bite. (microsoft.com) ### What does this mean for customers? More supply should eventually mean more available AI capacity across Azure. That helps companies that want managed models, inference, and enterprise tooling without building their own clusters. But it also sharpens the tradeoff for firms with very specific latency or cost needs. If hype(microsoft.com) some customers will keep asking whether owning more infrastructure themselves pencils out better. ### Why should investors care? Because this is the new shape of the AI race. The winners are not just writing better models — they are financing industrial-scale compute networks. Microsoft is showing that demand is strong enough to justify the buildout, but also that the economics are getting tougher at the same time. Bigger revenue, bigger backlog, bigger capex, tighter margins — all of those can be true at once. (microsoft.com) ### Bottom line? Microsoft’s $190 billion capex plan says the AI boom has moved past experimentation and into heavy industry. The opportunity is enormous. So is the cost of keeping up.
