Google, Amazon, Microsoft commit $700B+

AI spending is starting to look less like a product cycle and more like an industrial buildout. The headline numbers are huge, but the real story is simpler — Google, Amazon, Microsoft, and Meta are all telling investors they need vastly more physical capacity, fast. Not just chips. Whole campuses. Power systems. Cooling loops. Fiber. Custom silicon. The update this week is that Alphabet and Meta pushed those plans even higher, while Amazon and Microsoft kept signaling that their own spending remains at historic levels. (abc.xyz) ### What actually changed this week? Alphabet raised its 2026 capital spending plan to as much as $190 billion on April 29, after a quarter where Google Cloud revenue jumped 63% and crossed $20 billion for the first time, with backlog nearing $460 billion. Meta, also on April 29, raised its own 2026 capex guide to $125 billion to $145 billion, up from $115 billion to $135 b(abc.xyz)y’d need even a few months ago. (abc.xyz) ### Why are these companies spending so much? Because AI demand is no longer hypothetical. Google said cloud growth was driven by AI products and infrastructure. Amazon said AWS AI services are already running at more than $15 billion in annualized revenue, and Andy Jassy said much of the AWS capex planned for 2026 already has customer commitments behind it. Microsoft, mean(abc.xyz)s is the key shift — the capex is being defended with revenue, not just vision slides. (abc.xyz) ### So is this really a $700 billion story? Not from the clean, sourced numbers in hand right now. The solid figures are Alphabet at up to $190 billion for 2026, Meta at up to $145 billion for 2026, Amazon discussing roughly $200 billion in 2026 capex tied mainly to AI in Jassy’s April 9 shareholder update, and Microsoft still carrying the widely cited $80 billion AI infras(abc.xyz) not a neatly confirmed $700 billion-plus hyperscaler total from company filings alone. The broad direction is right. The exact headline total looks overstated unless more company guidance lands. (money.usnews.com) ### Why does the “not just GPUs” point matter? Because the bottleneck has moved. A GPU cluster is useless without the rest of the stack around it — networking fabric, storage, racks, transformers, substations, chillers, liquid cooling, and enough land with enough power in the right place. Meta’s new Tulsa site is a good example. It (money.usnews.com)round fiber, cooling, and cabling. That is what AI capex looks like on the ground. (about.fb.com) ### Why are investors reacting differently? Because the market is rewarding visible monetization and punishing fuzzier payoff timelines. Alphabet got to pair higher spend with booming cloud growth. Amazon paired heavy spend with a disclosed AI revenue run rate and customer commitments. Meta raised spending too, but investors focused on the size of the increase and the less pr(about.fb.com)sh fastest. (money.usnews.com) ### What does this mean for suppliers? It widens the winner list. Nvidia still matters a lot, obviously, but so do Broadcom, AMD, custom-chip partners, optical networking vendors, data center builders, utilities, and anyone who can get megawatts online quickly. Zuckerberg even highlighted Meta’s own custom silicon rollout with Broadcom plus AMD chips alongside Nvidia systems. The spend is spreading across the whole infrastructure chain. (finance.yahoo.com) ### What’s the bottom line? The important update is not that Big Tech likes AI. Everyone knew that. It’s that the biggest platforms are now budgeting for AI like railroads once budgeted for track — as core industrial infrastructure. And the hard constraint is no longer imagination. It’s how fast they can build.