Meta’s AI push is no longer an abstract future bet. It is now showing up in the bluntest possible place: cash flow.

The company’s second-quarter numbers exposed the trade-off. On one side, Meta is still a formidable money machine, with operating cash rising and ad revenue growing fast. On the other, that cash is being vacuumed up by an AI infrastructure buildout so aggressive that free cash flow fell 91% to $784 million.

That is the core divide in how to read the quarter. The bearish interpretation is simple: “Meta’s free cash flow sinks.”¹ The more strategic view is that the drop says less about a weakening business than about the sheer scale of Meta’s spending ambition.

Business Insider framed it even more starkly, arguing that “Meta’s AI bets are swallowing almost all its free cash flow.”² That tracks with the underlying math. Operating cash flow rose 25% year over year to $31.86 billion, but capital expenditures on servers, networking, and data centers surged 83% to $31.08 billion, nearly erasing what was left over.²

Still, this is not a classic retrenchment story. Meta’s ad engine is growing, and the company says AI is already improving content recommendations and ad targeting, helping drive a 27% rise in advertising revenue.² In other words, the spending is hurting near-term financial optics while also feeding the product systems that keep Meta’s core business humming.

The broader argument in Meta’s favor is that this is deliberate front-loading: build the data centers now, monetize later. The counterargument is that investors are being asked to tolerate enormous spending today on the promise that future AI products, agents, and enterprise tools will justify it. For now, both things are true: Meta’s business is strong, and its AI bill is getting brutally expensive.¹²