Meta funnels layoff savings into $125–145B AI infrastructure plan for 2026

Meta’s ad machine is still throwing off huge cash. But instead of letting that cash widen margins, Meta is pouring even more of it into AI hardware. That was the real news on April 29, when the company lifted its 2026 capital-expenditure forecast to $125 billion-$145 billion and made clear the money is headed into infrastructure — data centers, servers, networking, and custom chips. ### What changed this week? Meta didn’t just report a strong quarter. It also raised its 2026 capex outlook from $115 billion-$135 billion to $125 billion-$145 billion. That is a huge number even by hyperscaler standards, and it tells you Meta thinks the AI race is now mostly about physical capacity — who can build and power enough compute, fast enough. ### Why is capex the story? Because capex is where strategy turns into concrete. Hiring a few more researchers is one thing. Committing up to $145 billion means Meta is locking itself into a buildout that will shape the company for years. In the same period, Meta also disclosed additional multi-year infrastructure not a flexible experiment anymore. ### What is Meta buying? Mostly compute. Meta has been expanding AI-optimized data centers, building larger clusters, and pushing harder into its own silicon. In March it said four new generations of MTIA chips are coming within two years. In April it expanded its Broadcom partnership to co-develop multiple generations of a chip after that. ### Why does the data-center piece matter so much? Because AI at Meta’s scale is now an industrial project, not just a software project. In February, Meta broke ground on a 1GW campus in Lebanon, Indiana, calling it an investment of more than $10 billion. That one site alone gives a sense of the physical footprint, land, power, cooling, and time. ### So why do layoffs show up in this story? Because Meta is trying to fund two things at once — a giant AI expansion and a leaner operating structure. The company has been seeking cost savings through planned layoffs while still paying up for AI talent and infrastructure. That is the trade: fewer people, things matter most. ### Can the core business afford this? For now, yes. Q1 2026 revenue rose 33% to $56.3 billion, and reported net income rose 61% to $26.8 billion, though that profit figure got a big lift from an $8.03 billion tax benefit. The more durable point is that Meta’s ad engine is still strong — ad impressions rose 19%, and that cash generation is what gives Zuckerberg room to keep spending. ### Why are investors still uneasy? Because even if revenue is strong, infrastructure spending of this size can crush free cash flow in the near term. Meta’s shares fell about 5% in extended trading after the results. Investors are basically asking one question: does this buildout create a lasting AI advantage, or does it become another giant cost base that takes years to pay back? ### Bottom line? Meta is no longer treating AI as a feature layer on top of its apps. It is rebuilding the company around compute. The layoffs matter politically and culturally, but financially they look more like offsetting moves inside a much bigger decision — spend whatever it takes to own the infrastructure.