Meta taps $13B Texas financing

Data centers are turning into power plants with servers attached. That is basically what this Meta story is about. The company is putting together roughly $13 billion for a giant data center in El Paso, Texas, and Morgan Stanley plus JPMorgan are leading the financing. Most of the money is expected to come as debt, not straight equity. (bloomberg.com) ### Why is this news now? The immediate news is the financing itself. Meta already told the world it was building a gigawatt-scale AI data center in El Paso, but now the project is big enough that it needs a far more elaborate capital structure. Bloomberg says the package could reach about $13 billion, w(bloomberg.com)ure finance. (bloomberg.com) ### What exactly is Meta building? Meta’s El Paso campus is meant to support the company’s AI push — the training and serving of large models needs huge clusters of advanced chips, and those chips need a lot of electricity and cooling. When Meta first announced the Texas site in October 2025, it framed t(bloomberg.com)ite is expected to come online in 2028. (bloomberg.com) ### Why does the dollar figure keep growing? Because the original headline number was not the final buildout cost. By March 26, 2026, Meta said the El Paso investment had climbed to more than $10 billion. Now the financing under discussion is around $13 billion. That does not automatically mean pure construction (bloomberg.com)s clear: this is much larger than the project Meta first described. (bloomberg.com) ### Why use debt instead of just paying cash? Because AI infrastructure is getting absurdly expensive, even for companies as rich as Meta. Debt lets Meta spread the cost over time, preserve balance-sheet flexibility, and keep spending on chips, models, and other campuses at the same time. The broader bac(bloomberg.com)ry dollar directly onto the parent company’s books. (bloomberg.com) ### Why Texas? Power, land, and scale. El Paso gives Meta room to build a very large campus, and Texas remains one of the main places where hyperscalers think they can still assemble huge loads of electricity. But that does not mean it is simple. A gigawatt-scale data center is not just a warehouse full (bloomberg.com)zation all baked into the economics. That is why the financing matters so much. (bloomberg.com) ### Is Meta doing this only in Texas? No — and that is part of the point. Meta has been scaling multiple giant sites at once. In Louisiana, its Hyperion project has become so power-hungry that Entergy is planning 10 gas plants to support it, with more than 7 gigawatts of targeted output. That shows how the compan(bloomberg.com 1)(bloomberg.com 2) ### What does this say about the AI race? The catch is that the AI race now runs through financing as much as model quality. The winners will not just be the companies with the best chips or smartest researchers. They will be the ones that can secure land, power, construction capacity, and tens of billions in capital before bottlenecks get worse. Meta’s Texas deal is a clean example of that shift. (bloomberg.com) ### Bottom line? Meta is not merely funding another office park. It is assembling utility-scale infrastructure for AI, and the $13 billion Texas financing shows how expensive — and how financialized — that buildout has become. (bloomberg.com)