Meta Wants $95 Billion for AI. Nvidia Might Chip In. Literally.

Meta Wants $95 Billion for AI. Nvidia Might Chip In. Literally.

There's a classic move in tabletop games where the player who owns the most property also starts lending money to the bank. Meta just tried a version of that in real life: it needs staggering amounts of cash for AI, and some of the companies selling it the chips are thinking about buying its stock. Monopoly money this is not.

$35 Billion Now, $60 Billion Maybe Later

On Monday, October 26, Meta announced a plan to raise up to $95 billion to fund its AI infrastructure and superintelligence ambitions. It starts with $35 billion in underwritten public equity offerings, followed by up to $60 billion more through an at-the-market program beginning in early 2027. Meta was careful to note that the second tranche isn't guaranteed.

The twist is in the cap table. Nvidia, AMD, and Broadcom have expressed interest in buying up to $5 billion of shares combined, though Meta stressed those expressions of interest are "neither binding agreements nor firm purchase commitments." Meta also nudged its capital expenditure outlook for this year to $135–145 billion, raising the low end from $130 billion. Investors weren't throwing confetti: the stock dipped about 1.6% in morning trading.

When Your Suppliers Become Your Shareholders

Chipmakers investing in their own biggest customer is a neat little loop. Meta buys the hardware, the hardware makers buy Meta, and everyone's quarterly earnings look a bit cozier. It isn't illegal or even unusual in a boom, but it's the kind of circular arrangement that makes analysts squint, and reports suggest they're more skeptical of this raise than of Google's and Intel's earlier stock offerings.

The angle most people miss is what this says about the economics. Meta prints money from ads and still has to go to the equity markets to keep up. If the biggest players need outside capital to fund the AI race, the infrastructure bill is far larger than the friendly chatbot interfaces suggest. And that cost doesn't vanish. It shows up later in API pricing, platform terms, and which features get prioritized.

The AI arms race just switched from "move fast and break things" to "move fast and issue shares."

Whatever the giants spend, your business doesn't need a superintelligence budget to get real value out of AI, and we build practical, right-sized AI integrations that hold up no matter whose stock is doing what.

Source: CNBC