NVIDIA signed a memorandum of understanding earlier this week with six of the world’s largest financial institutions to build a standalone compute financing platform targeting over $500 billion in third-party capital for AI infrastructure. Apollo Global Management, BlackRock, Blackstone, Brookfield Asset Management, Goldman Sachs, and KKR will each establish dedicated capital pools that offer competitive debt financing to NVIDIA customers building data centers and procuring GPUs at scale.
Jensen Huang framed the deal as a foundational shift in how the technology industry finances infrastructure at the largest scale.
“This is truly the first time that technology chips have become an investable asset class,” he stated. “They are now income-generating assets with high productivity, long useful lives, strong fungibility, and high flexibility.”
He pointed to the A100 chip launched in 2020 as direct evidence of that durability. The chip remains actively deployed across AI training, fine-tuning, and inference six years after its release. Market pricing supports the thesis as well: H100 lease rates rose from roughly $1.70 per GPU-hour in October 2025 to approximately $2.35 by March 2026. Next-generation Blackwell cloud pricing already reaches between $5.30 and $7.05 per GPU-hour.
BlackRock CEO Larry Fink compared the financing model to the creation of mortgage-backed securities in the 1970s, calling it potentially the beginning of “the next future of financial engineering.” Goldman Sachs CEO David Solomon revealed that Huang personally pitched the concept to Wall Street’s leadership directly. Solomon added that the industry stands at “a critical juncture in a historic AI investment cycle” with enormous capital deployment ahead.
The scale of deals already surfacing around the platform is staggering even before final agreements have been signed by participants. NVIDIA is reportedly in talks with OpenAI to provide up to $250 billion in standby financing support for a 10-gigawatt data center project that SoftBank’s SB Energy is developing in Ohio. A separate $350 billion financing package for OpenAI’s chip procurement plans is also under active discussion between the parties. Either deal alone would represent the single largest capital support case in NVIDIA’s entire corporate history.
The “circular financing” concern arrived alongside the announcement and has refused to leave the conversation since the details emerged. Critics have long observed that NVIDIA both invests in AI companies and sells them chips, creating a reinforcing loop where capital and hardware purchases validate each other without truly independent demand signals. Huang addressed this directly on social media, insisting that each capital provider will independently evaluate customer qualifications, utilization rates, cash flow, and residual asset value. NVIDIA may provide residual value guarantees of up to 25% on individual projects, but Huang characterized that as “far lower than other compute financing arrangements” in the current market.
Jim Chanos, the legendary short seller, responded on social media with characteristic precision and historical awareness. He suggested that the next time these executives sit together discussing AI financing, “it wouldn’t be surprising if it’s at a Congressional hearing in 2031.” The comparison drew a direct line to the 2008 financial crisis when Wall Street leaders faced congressional interrogation over the subprime mortgage collapse.
Morgan Stanley estimates that hyperscale providers including Meta, Microsoft, Alphabet, and Amazon will collectively invest $3.5 trillion in AI infrastructure between 2026 and 2028. Apollo President Jim Zelter projects total AI infrastructure requirements exceeding $8 trillion over the coming decade. The NVIDIA financing platform positions GPU compute as the collateral layer underneath that spending, creating what amounts to a new credit market backed by silicon rather than real estate or traditional infrastructure.
NVIDIA shares fell 2.86% on the announcement day to close at $217.55. The decline reflected investor nervousness about the model’s risk concentration rather than outright rejection of the underlying concept.
