Microsoft hikes capex to $190B

Microsoft’s AI spending story just got more concrete — and more expensive. On April 29, during its fiscal Q3 2026 earnings call, Microsoft said it expects to spend roughly $190 billion on capital expenditures in calendar 2026. The headline sounds like simple aggression in the AI race. But the more interesting part is the reason the number jumped so much. A big chunk of the increase is not extra ambition. It is inflation inside the hardware stack. ### What did Microsoft actually say? Microsoft’s investor materials and earnings call put the number plainly: roughly $190 billion in capex for calendar 2026. Amy Hood said that forecast includes about $25 billion from higher component pricing. So this was not just “we want more data centers.” It was also “the same buildout now costs a lot more.” (microsoft.com) ### Why is that detail such a big deal? Because it changes how you read the spending number. If $25 billion of the increase comes from pricier components, then part of the budget growth is cost inflation rather than extra physical volume. In plain English — Microsoft can spend way more dollars without getting 1-for-1 more GPUs, servers, or memory. That m(microsoft.com)n anymore. (microsoft.com) ### What components are getting pricier? The pressure is centered on AI-critical parts — especially memory and advanced chips. HBM, the high-bandwidth memory used alongside AI accelerators, has been in tight supply, and storage and memory pricing has climbed fast as hyperscalers all chase the same gear. Microsoft did not publish a line-by-line bill of ma(microsoft.com)g, and industry coverage points to memory and chip costs as the main culprit. (microsoft.com) ### Is this just a Microsoft problem? Not really. It looks more like an industry condition. Alphabet also raised its 2026 capex outlook to as much as $190 billion, and broader estimates for Microsoft, Amazon, Alphabet, and Meta now put combined 2026 AI infrastructure spending around $725 billion. Basically, the biggest buyers are all bidding into the sam(microsoft.com)e. (cnbc.com) ### So does higher capex still mean stronger AI demand? Yes — but with a catch. Microsoft’s core business still shows real demand for cloud and AI infrastructure. Revenue rose to $82.9 billion in the March quarter, up 18%, and management kept leaning hard into AI capacity. The catch is that demand strength and cost inflation are now intertwined. A giant spending pla(cnbc.com)nal units that money will actually buy. (microsoft.com) ### What does this do to margins? It puts pressure on them. Microsoft’s gross margin has already been feeling the weight of depreciation and AI infrastructure buildout, and pricier components add another squeeze point. The company can offset some of that with software economics and cloud scale, but the near-term pattern is clear — AI is driving growth, and it is also making the business more capital-intensive. (equity-insider.com) ### Why should anyone outside Microsoft care? Because this is what the AI boom looks like once it leaves the demo stage. The bottleneck is no longer just model quality. It is power, data centers, chips, and memory — the industrial inputs. Microsoft’s new number is a reminder that the next phase of AI is being shaped as much by semiconductor pricing as by software breakthroughs. (microsoft.com) ### Bottom line The real news is not just that Microsoft plans to spend $190 billion. It is that $25 billion of that is the cost of the same race getting more expensive. That makes AI capex a noisier signal than it used to be — still bullish on demand, but less straightforward as a measure of how much new capacity Big Tech is actually adding.