Wall Street backs Nvidia’s AI financing push

Wall Street is preparing to finance AI data centers as income-producing infrastructure.

Nvidia CEO Jensen Huang used a CNBC interview on Monday to pitch a new way to pay for AI factories, and he brought six heavyweight backers along for the message. Goldman Sachs, BlackRock, Blackstone, KKR, Apollo and Brookfield said they want to raise up to $500 billion, and possibly more, for AI infrastructure.

The pitch matters because the first wave of AI build-outs has already burned through enormous amounts of capital. Tech giants have leaned on debt, equity and cash flow to fund chips, servers and data centers, while Nvidia’s own earlier OpenAI investment plan never fully materialized.

Wall Street is treating AI gear like an asset class

Huang’s big idea is simple: AI infrastructure is no longer just a pile of hardware that gets depreciated on a spreadsheet. He wants investors to see it as a revenue-producing asset with long useful life, predictable cash flow and enough value to be financed like other large infrastructure projects.

That framing is why the comment from Goldman Sachs CEO David Solomon mattered so much. “You’re starting to see, in a sense, you know, asset-based financing against this infrastructure build-out,” Solomon said on the CNBC panel. “That’s not surprising because these are real assets. They have real value.”

The language is familiar to anyone who has followed project finance, shipping, aircraft leasing or commercial real estate. The twist is that the collateral here is a cluster of Nvidia-powered systems built for AI workloads, and the demand story is tied to model training, inference and the rise of AI agents.

• The six firms named on CNBC were Goldman Sachs, BlackRock, Blackstone, KKR, Apollo and Brookfield.
• Nvidia said it may backstop 25% of each loan, which could improve rates for borrowers.
• Huang said borrowers would need system designs that another operator could take over if needed.
• McKinsey expects global AI outlays to reach $7 trillion by the end of the decade.

The financing pitch is bigger than one company

This is where the story moves beyond Nvidia. Alphabet, Amazon, Meta, Microsoft and Oracle have already raised more than $150 billion this year through debt and equity sales tied to data centers and AI development. Intel also announced a $15 billion stock offering, then increased it to $20 billion.

That money is feeding a build-out that has become one of the most capital-hungry technology cycles in recent memory. The basic problem is that demand for compute is rising faster than even the largest balance sheets want to absorb on their own.

Wall Street sees a chance to step in as the middle layer between the builders and the end users. If the systems can generate steady revenue, then banks, private credit firms and asset managers can package the cash flows into loans, structured products or other financing vehicles.

“This is the very beginning, like what it was when I started in the mortgage-backed securities market in the 1970s,” BlackRock CEO Larry Fink said on CNBC. “I look upon this as as a next future for financial engineering.”

That comparison is useful, but it also raises the obvious question: what happens when investors start assuming every expensive machine will keep earning forever? The AI market is still young, and the underwriting standards for these projects have not been stress-tested through a downturn.

The risks are real, and the market knows it

The biggest red flag is depreciation. Michael Burry, the investor known for betting against subprime mortgages, argued late last year that companies such as Meta, Oracle, Microsoft, Google and Amazon may be overstating how long their AI chips stay useful. If that turns out to be true, the economics of these financing structures get weaker fast.

There is also concentration risk. Apollo president Jim Zelter acknowledged that “there will be excesses, there will be pullbacks,” while also arguing that having more participants reduces the chance that one failure knocks the whole structure off balance.

That tension is exactly why this financing model feels important. It could unlock far more capital for AI infrastructure than corporate balance sheets alone can provide, but it also pushes Wall Street deeper into a market whose long-term cash flows are still being guessed at rather than proven.

• OpenAI’s earlier Nvidia-linked plan was tied to 10 gigawatts of power demand.
• Nvidia later contributed $30 billion to OpenAI’s record funding round this year.
• The new Wall Street effort is framed as third-party capital, not direct balance-sheet spending by Nvidia.
• Brookfield CEO Bruce Flatt said Huang created the format for investors because “there’s hundreds of trillions of dollars of money in the world.”

What happens next depends on the first deals

The real test is whether these firms can turn the concept into lending terms that borrowers will actually accept. Rates, maturities, collateral rules and operating handoff provisions will matter more than the marketing language around AI factories.

If the first transactions close cleanly, expect other lenders to copy the model and push even more capital into AI infrastructure. If the deals stall, the whole pitch will look like a very expensive experiment in financial engineering.

For now, the most useful takeaway is that AI build-outs are no longer being funded only by hyperscaler cash and chipmaker optimism. Wall Street wants a piece of the bill, and the next few financing agreements will show whether Huang’s big concept is a durable new funding channel or just a well-timed pitch to investors with too much cash and too little yield.