AI News: Data Center Costs Could Hit $8 Trillion

A recent tweet from StockMarketNews pointed to something pretty eye-opening: companies worldwide might need anywhere from $3 trillion to $8 trillion by 2030 just to keep up with AI data center demand. We’re looking at one of the biggest and fastest infrastructure buildouts in corporate history—and here’s the kicker: most of it’s being funded through private credit deals that regulators are starting to worry about.

If things move fast, we’re talking about $7.9 trillion in total spending. That breaks down to $4.7 trillion on IT gear, $2.6 trillion on data center facilities, and $600 billion on power infrastructure. Even if growth slows down, the bill still comes to $3.7 trillion. Numbers like these are making governments rethink their tax codes, especially since there’s growing concern about companies taking on too much debt, potential bankruptcies down the line, and talented workers leaving traditional industries for AI.

The cost of sitting this out could be steep. Banks aren’t handling most of this anymore—it’s private credit funds, insurance firms, pension managers, and securitized lending vehicles picking up the tab. Financial advisers are warning that if profit tax rules don’t get updated, governments could end up with serious budget holes as companies find clever ways to work the system.

The way money’s moving in this space is changing fast. Take Meta’s deal with Blue Owl Capital: Blue Owl technically owns most of a new data center, but Meta runs it. Since Meta owns less than half, they get to keep nearly $30 billion in liabilities off their books. Other big tech companies are doing similar things, and that’s raising red flags about transparency, shrinking tax revenues, and risks spreading through the shadow banking world. Oracle’s debt is projected to hit $290 billion by 2028, and Amazon, Microsoft, and Google are setting up comparable arrangements.