Alphabet, Amazon, Microsoft plan $700B

Data centers are the story here — not chatbots, not demos, not vague “AI strategy” talk. Alphabet, Amazon, and Microsoft just spent the past week telling investors they are still going all in on the physical buildout behind AI, even with Wall Street getting much less patient about returns. That matters because these companies are no longer talking about tens of billions. They are talking about annual capital spending on a scale that starts to look like national infrastructure. Alphabet raised its 2026 capex guide on April 29, Microsoft gave a roughly $190 billion calendar-2026 figure the same day, and Amazon kept its own 2026 plan near $200 billion. ### Why is capex the real AI signal? Capex is the money these companies spend on the heavy stuff — servers, chips, networking gear, land, buildings, power systems. In AI, that is the bottleneck. Models only get sold if somebody first pays to build the compute factory underneath them. So when the hyperscalers talk capex, they are really telling you how much future demand they think is real enough to underwrite with concrete and silicon. ### What changed this week? Alphabet nudged its full-year 2026 capex guidance up to $180 billion to $190 billion from $175 billion to $185 billion, partly folding in its Intersect acquisition. Microsoft, on its April 29 earnings call, said it expects roughly $190 billion of calendar-2026 capex, including about $25 billion tied to higher component pricing. Amazon did not raise its February target, but it reaffirmed that 2026 capex will reach about $200 billion. ### How big is the total now? Add up Alphabet, Microsoft, Amazon, and Meta, and the industry is now staring at roughly $725 billion in 2026 capex. That is the number that made this week’s earnings feel different. It is not just that AI spending is still rising. It is that the rise has survived a full year of investor pushback, margin pressure, and questions about whether demand is real or just fear of falling behind. ### Why aren’t investors cheering automatically? Because spending is no longer the novelty. In 2024 and 2025, the market mostly rewarded companies for showing they were serious about AI. In 2026, that excuse is wearing thin. Microsoft itself flagged that higher component pricing is inflating the bill, while Amazon’s March-quarter spending cut into free cash flow. ### So who looks best positioned to monetize? Right now, the cloud incumbents still have the clearest path. Microsoft can point to Azure demand and Copilot usage. Amazon has AWS plus its own Trainium and Inferentia chip push. Alphabet has Google Cloud backlog growth and its Gemini stack feeding search, cloud, and enterprise tools. The point is not that monetization is solved. It is that buildings are full. ### What is the catch? The catch is that AI infrastructure is starting to look like an arms race where the cost of standing still is huge. If a company underbuilds, it risks losing customers. If it overbuilds, it risks years of weaker cash flow and lower returns. Microsoft’s note about $25 billion of extra component costs is a reminder that even the leaders do not fully control the economics here. ### What should you watch next? Watch for proof, not promises. Cloud growth tied specifically to AI workloads. Gross margin trends. Free cash flow. Backlog conversion. Pricing power. Basically, the build phase is not over, but the market has moved on. The next winner is not the company that spends the most. It is the one that makes $190 billion look rational.