AI is entering its financing era

Apparently, building the future is expensive—so much so that Big Tech wants someone else to help pay for it.

Amazon is reportedly considering putting about $8 billion worth of Nvidia Grace Blackwell chips into a special-purpose vehicle funded by outside investors, then leasing the chips right back. The vehicle would raise debt and could sell investors an equity stake of up to 10%, giving Amazon a more asset-light way to keep feeding AWS’s appetite for computing power.

Passing the bill

Why bother, you ask? Amazon expects to spend a whopping $220 billion on capex this year, much of it to expand AWS and AI infrastructure, up from an earlier forecast of $200 billion. But even Big Tech’s deep pockets have limits, and bringing in outside capital lets companies keep building without footing the entire bill themselves.

Amazon’s proposed deal is part of a broader shift in who finances the AI boom. The five largest AI hyperscalers are expected to spend more than $800 billion this year and $1.2 trillion next year, according to Goldman Sachs, and they’ll take any opportunity to offload those expenses.

That’s why SoftBank raised $11.1 billion last week to fund AI bets that include OpenAI, while Nvidia recently teamed up with Wall Street firms on platforms designed to mobilize more than $500 billion for AI infrastructure.

Follow the money

For long-term investors, the question is whether all this AI infrastructure will eventually earn enough to justify the spending. Credit markets are already starting to wonder about that: AI-related bonds are offering higher yields than the broader investment-grade market, a sign that investors want more compensation for the risk.

And some of the new collateral in these deals is unusually difficult to evaluate. A data center can operate for decades, but GPUs become obsolete much faster. Banks typically depreciate them over three to four years, making billions of dollars of chip-backed financing harder to value as newer models arrive.

The AI trade is no longer just a tech-stock trade. As the boom spreads into bonds, private credit, and infrastructure funds, whether AI generates enough returns to pay for today’s spending could increasingly affect investors across their portfolios.—HC