Nvidia CEO Jensen Huang stood shoulder to shoulder with the bosses of Goldman Sachs, BlackRock, Apollo, Blackstone, Brookfield, and KKR to announce a $500 billion financing deal for AI datacenters, chip factories, and power stations — and if history is any guide, the utility bills for all that electricity will land on your monthly statement, not on Wall Street's.
Huang posted on X that the partnership marks "a major milestone for Nvidia and the AI industry," framing AI compute — the physical infrastructure of datacenters, chips, networks, and software — as a new investable asset class. "In AI, compute is revenue," he wrote. "It produces revenue, serves a broad market, improves in performance over time and can be redeployed." The pitch is straightforward: Wall Street gets a new thing to finance and securitize, Nvidia gets more buyers for its GPUs, and the public gets to fund the power grid upgrades that make it all possible.
The six Wall Street firms signed memorandums of understanding to offer funding for what the industry calls "compute." Nvidia, valued at $5.3 trillion, did not disclose financial terms, individual firm commitments, or a timetable. Huang said the financing would "create dedicated pools of capital at significant scale at attractive rates" for customers who have demand for compute but lack affordable financing to build quickly.
Here is the problem ordinary Americans should watch. Every one of those AI factories requires staggering amounts of electricity. Datacenters are already straining power grids in Texas and Virginia, and the buildout Huang is cheering will only accelerate that pressure. Utility companies don't eat those costs — they pass them to ratepayers. The same working Americans who just got told their electric bill is going up again are now subsidizing the infrastructure that lets BlackRock and Goldman Sachs collect returns on AI compute.
Business Insider noted that the deal-making frenzy has raised questions about "the shaky nature of circular financing, as hardware firms like Nvidia invest in companies that will turn around and buy things like GPUs from them." Huang addressed this by saying demand comes from real customers — frontier AI labs, startups, enterprises, cloud providers — and that "NVIDIA provides the platform; the investors make independent financing decisions." That is a careful construction. It does not dispute the circularity; it reframes who is making the buying decisions.
The Guardian highlighted what Business Insider buried: the Bank of England's July warning that AI debt financing could threaten global financial stability. If companies taking on debt to build AI infrastructure fail to deliver sustainable profits, the central bank said, it could trigger a credit crunch making it "harder and more expensive for businesses and households to secure loans." The Bank also warned that financial firms may not understand their full exposure to AI industry risks because of opaque funding arrangements and "different levels of transparency." That is a polite way of saying the system is building leverage on an asset class that has never been tested through a downturn.
Big tech companies have signaled AI spending could surpass $730 billion this year alone. Huang's answer to monetization concerns is to point to the long view — AI writing software, discovering drugs, designing products. Maybe. But the returns he promises are speculative, and the costs — to the power grid, to ratepayers, to financial stability if the bets go wrong — are immediate and real.
Wall Street is always happy to finance a new asset class when the upside is privatized and the downside is socialized. The question is who eats the losses when the bill comes due.
