- Nvidia signed MOUs with six asset managers to mobilise $500B+ for AI infrastructure.
- The deal reframes GPUs as financeable infrastructure rather than depreciating hardware.
- It builds on a pattern already visible at CoreWeave, Apollo/xAI and Nscale.
Nvidia has signed memorandums of understanding with Apollo Global Management, BlackRock, Blackstone, Brookfield Asset Management, Goldman Sachs and KKR to build dedicated financing platforms for AI infrastructure, aiming to mobilise more than $500 billion in third-party capital over time.
“NVIDIA has reached an important milestone. We began by building chips; today, we are helping create a new class of productive, investable infrastructure: AI factories. In AI, compute is revenue… That is why we are bringing the world’s leading long-term capital providers together to independently underwrite AI infrastructure,” said Jensen Huang, founder and chief executive of Nvidia.
The platforms are designed to give hyperscalers, frontier AI labs and enterprises a way to fund data centres and buy Nvidia hardware without loading the entire cost onto their own balance sheets. It shifts what capital markets underwrite: compute itself, not just the companies that sell or rent it.
The pitch: chips as toll roads
Huang’s framing, delivered in a joint interview with CNBC alongside the six partners, was blunt. “These are revenue-generating assets now. They’re productive, they’re long-lived, they’re fungible, they’re flexible,” he said.
The comparison he’s reaching for is infrastructure in the traditional sense: commercial property, electricity grids, toll roads. Assets that generate predictable cash flow and can be pledged as collateral.
That’s a shift from how hardware has historically been treated: as a rapidly depreciating IT expense. GPUs get replaced every generation or two; a chip bought today is not obviously worth pledging against 10-year debt. Nvidia’s counterargument is that its hardware is broadly adopted and the CUDA software layer sticky enough that compute capacity can be redeployed across customers as demand shifts, which is what makes it financeable in the first place.
Apollo president Jim Zelter called compute “a scarce, mission-critical asset class,” tying it to what Apollo calls the broader Global Industrial Renaissance. BlackRock chairman and chief executive Larry Fink said the partnership “deepens our relationship with NVIDIA” and connects “long-term capital to essential infrastructure.”
Blackstone president and chief operating officer Jon Grey leaned on the firm’s existing exposure: “We continue to be enormous investors globally across the NVIDIA ecosystem.” Brookfield chief executive Bruce Flatt called compute “the essential layer of infrastructure.”
Goldman Sachs chairman and chief executive David Solomon said the bank was “excited for the new opportunity to create a market for credit backed by NVIDIA compute.” KKR co-chief executives Joe Bae and Scott Nuttall talked about turning “growing demand into real capacity at extraordinary scale.”
A pattern already forming
This isn’t Nvidia inventing GPU-backed finance from scratch. Rather, it’s trying to standardise and scale a structure that’s already been tested piecemeal across the AI infrastructure market.
Tech Funding News has tracked several of these deals directly. CoreWeave closed an $8.5 billion financing facility in March 2026, structured as a delayed-draw term loan, providing initial access to $7.5 billion, with the full amount available as its GPU-backed assets stabilise. Anchored by Blackstone Credit and Insurance and co-structured by Morgan Stanley and MUFG, it carries an A3 rating from Moody’s.
Apollo has separately backed a roughly $3.4 billion chip-leasing arrangement for xAI, buying Nvidia chips through an investment vehicle and leasing them to Elon Musk’s AI company, with Nvidia itself participating as an anchor investor. And in Europe, Nscale has assembled a stack of GPU-backed debt including a $1.4 billion term loan in February 2026 and a $900 million revolving credit facility in July 2026, led by a syndicate of banks including J.P. Morgan and Goldman Sachs.
Nvidia itself has been testing debt markets too. In June 2026, it raised $25 billion in bonds despite holding more than $13 billion in cash — its first bond sale since 2021 and one that drew $85 billion in investor demand.
What Nvidia’s new platforms aim to do is take that improvised, deal-by-deal pattern and turn it into standing infrastructure: repeatable pools of capital that any qualifying customer in Nvidia’s ecosystem can tap, rather than one-off vehicles negotiated separately for each company that needs chips.
Does the collateral hold up?
A toll road doesn’t lose a third of its value when a competitor opens a faster road next door. A GPU might, the moment a better chip ships. Lenders underwriting compute as long-duration collateral are betting that broad adoption and continuous software improvement stretch a chip’s useful life well past its technical replacement cycle, and that demand for AI compute stays strong enough to keep utilisation high even as newer hardware arrives.
That vision is already being placed at scale elsewhere. Big Tech’s AI-related debt has doubled to $350 billion, according to Bank of America estimates cited by TFN, as hyperscalers lean on bond markets faster than their cash flow can cover the buildout. Nvidia’s platforms would add another, much larger layer on top of that.
If that holds, Nvidia’s initiative could shift AI infrastructure funding from a capital-expenditure race led by a handful of hyperscalers into a broader institutional market, spreading the exposure, and the upside, much further into pension funds, insurers and other long-term capital pools that have so far stayed on the sidelines of the AI buildout.
