Meta raises $13B debt for campus

Meta’s latest AI story is not really about a chatbot. It’s about financing. The company is working on a roughly $13 billion package for its El Paso, Texas, data center campus, with Morgan Stanley and JPMorgan leading the deal, and most of the money expected to come as debt. That matters because Meta is not just buying more servers — it is building utility-scale AI capacity and funding it more like a power or telecom project. (money.usnews.com) ### What actually got announced? The new piece of news is the financing structure. Meta had already told El Paso in March that the campus would expand to more than $10 billion and grow to 1 gigawatt of capacity. This week’s development is that the company is now arranging about $13 billion around that project, with a debt-heavy mix rather than paying for everything straight off the balance sheet. (datacenters.atmeta.com) ### Why El Paso? Because Meta wants one huge site that can handle the ugly physical demands of modern AI — power, land, cooling, and room to keep expanding. One gigawatt is an enormous number for a single campus. It puts the project in the same mental category as industrial infrastructure, not a normal enterprise data (datacenters.atmeta.com) 4,000 construction workers at peak buildout. (datacenters.atmeta.com) ### Why use debt at all? Because the AI buildout is getting too big to treat as routine capex. Meta still has a strong balance sheet, but spreading the cost through project-style financing preserves flexibility while it keeps spending aggressively elsewhere. Bloomberg and Reuters both describe the package as mostly deb(datacenters.atmeta.com)ters as financeable infrastructure with predictable long-term value. (bloomberg.com) ### How big is Meta’s broader spending push? Huge. In its first-quarter 2026 results, Meta raised full-year capital expenditure guidance to $125 billion to $145 billion, up from $115 billion to $135 billion. The company said the increase reflects higher component pricing and additional d(bloomberg.com)e before demand gets worse and supply gets tighter. (investor.atmeta.com) ### Why does 1 gigawatt matter so much? Because power is now the bottleneck. AI training clusters need dense racks, heavy cooling, and reliable electricity at a scale that very few sites can provide quickly. A 1-gigawatt target means Meta is planning for very large training(investor.atmeta.com) place that can actually run them?” That last question now decides who gets capacity first. (datacenters.atmeta.com) ### What changed from a few months ago? The speed and the financing model. In March, Meta raised the El Paso commitment from about $1.5 billion to more than $10 billion. Now, barely weeks later, reports show it assembling a roughly $13 billion financing package around the campus. That sequence suggests the project move(datacenters.atmeta.com)s an inference, but it fits the size, the lenders, and the debt-heavy structure. (cnbc.com) ### What’s the catch? The catch is that these campuses do not just consume money. They consume power, water, grid attention, and construction capacity. Local reporting in El Paso has already highlighted concerns around electricity, gas generation, water use, and tax breaks. So even if the financing gets(cnbc.com)t enough. (elpasomatters.org) ### Bottom line Meta is showing what the next phase of the AI race looks like. Not just bigger models — bigger balance sheets, bigger campuses, and financing that looks a lot more like infrastructure than tech. If that model sticks, the winners will be the companies that can secure land, power, and credit at the same time.