Big Tech spends $130B on AI capex
- Microsoft, Alphabet, Meta, and Amazon just turned earnings week into an AI infrastructure arms race, with 2026 capex plans swelling toward $725 billion. - Microsoft put the clearest payoff number on the table — a $37 billion AI revenue run rate — while still planning roughly $190 billion in 2026 spend. - The market is rewarding proof of demand, not spending alone — Alphabet jumped, Meta fell, and capacity itself is becoming the product.
Data centers are the story here — not chatbots, not demos, not vague AI promises. The biggest tech companies just used earnings week to say they are spending at a scale that looks less like software and more like railroads, utilities, or chip fabs. Microsoft, Alphabet, Meta, and Amazon now look set to pour as much as $725 billion into capital expenditures in 2026, mostly for AI infrastructure. That matters because the bottleneck in AI is no longer just model quality. It is power, chips, buildings, networking, and how fast these companies can turn concrete into usable compute. (bloomberg.com) ### Why did this suddenly feel bigger? Because the numbers all landed at once. Microsoft said it expects about $190 billion in capital spending this calendar year. Alphabet lifted its 2026 capex range to $180 billion to $190 billion. Meta raised its range to $125 billion to $145 billion. Amazon kept its earlier $200 billion target in place after a huge March-quarter spending jump. Add those together and you get a spending plan north of $700 billion. (money.usnews.com) ### What are they actually buying? Mostly the physical stack for AI — GPUs and other chips, servers, networking gear, land, buildings, and the power systems that keep giant data centers running. Meta said its higher capex reflects pricier components and more data-center costs. Microsoft said $25 billion of its 2026 spend(money.usnews.com) with supply-chain constraints attached. (money.usnews.com) ### Why is Microsoft the key read-through? Because Microsoft gave investors the cleanest evidence that the spending is producing revenue. It said its AI business has passed a $37 billion annual revenue run rate, up 123% from a year earlier. Azure and other cloud services grew 40% in the March quarter, and Microsoft guide(money.usnews.com)ad of it. (news.microsoft.com) ### Why didn’t Wall Street react the same way to everyone? Because spending only looks good when investors can see the payback path. Alphabet’s shares rose after it paired bigger capex with a 63% jump in Google Cloud revenue to more than $20 billion. Meta, by contrast, dropped hard even after a strong quarter, because it raised c(news.microsoft.com)rent proof points. (cnbc.com) ### What about Amazon? Amazon is the quiet giant in this story. It did not raise its 2026 target this week because it had already set a roughly $200 billion plan earlier. But its March-quarter capex surged, AWS sales rose 28% to $37.6 billion, and the company’s free cash flow got squeezed as it kept building. That is the tradeoff in one line — demand is there, but capacity is expensive before it is lucrative. (ir.aboutamazon.com) ### Why does this change what “software company” means? Because the hard problem is shifting from writing code to operating scarce capacity. If customers want AI now, the winner is the company that can admit workloads, queue them, route them, and keep utilization high without melting margins. That starts to look like running an airline or a power grid — sell too much and users wait, build too much and returns sag. (money.usnews.com) ### So what is the real bottom line? The AI race has moved into its heavy-industry phase. The flashy part was model launches. The decisive part may be who can finance, build, and fill the most compute the fastest. Microsoft has the strongest “show me the money” argument right now, Alphabet is gaining credibility fast, Am(money.usnews.com)cks. (news.microsoft.com)
