Investors question how Big Tech will fund the next AI data centers

In short: The next wave of AI data center building in the US may be harder to pay for as financial markets tighten and big tech leans more on outside money.

What's going on

Building AI systems takes a lot of physical infrastructure, especially data centers, which are large buildings packed with computers (like warehouses full of servers). The Financial Times reports that US data center projects already being planned could average about 3.6 percent of US GDP each year through 2032, according to Columbia Business School professor Stijn Van Nieuwerburgh.

So far, much of this spending has been carried by a handful of very large companies. These include Alphabet, Amazon, Meta, Microsoft, and Oracle. The article says that together they recently committed $33 billion more to capital spending than they generated from their operations, meaning their “free cash flow” turned negative (they spent more cash than they brought in).

That shift is pushing companies to rely more on raising money from investors and lenders. The FT points to new stock and debt issuance, including Meta selling about $95 billion of stock. Investors also worry that some core businesses that fund these efforts could weaken. For example, Google executives have said its newer AI search brings in less advertising money than traditional search, and a related comment previously helped trigger a sharp, one day drop in Alphabet’s market value.

What to watch

Several big tech firms have said they expect to raise spending again in 2027, on top of an estimated $800 billion in AI investment this year. At the same time, higher interest rates, equipment shortages, and local and political pushback could slow new data center construction. A slower build could ease funding pressure, but it could also delay the sales growth that many investors are expecting.

Source: Financial Times