NVIDIA is bringing some of Wall Street’s biggest firms into the race to finance AI data centers and compute. The chipmaker has teamed up with BlackRock, Goldman Sachs and other financial heavyweights on a plan that could mobilize more than $500 billion for AI compute over time.
The company has signed memorandums of understanding with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to create independent financing platforms for its ecosystem. The goal is to give AI labs, enterprises and cloud providers another way to fund large-scale infrastructure without carrying the entire upfront cost themselves.
NVIDIA wants to make compute a financeable asset
NVIDIA said the planned platforms would create dedicated pools of third-party capital for customers building AI infrastructure. The company is pitching its compute stack as a productive asset capable of generating long-duration, usage-linked revenue while also supporting continued hardware and software adoption.
“In AI, compute is revenue,” NVIDIA CEO Jensen Huang said in the announcement.
Huang argued that NVIDIA systems are broadly adopted, flexible across workloads, and continuously improved through CUDA software, which the company says can help extend their useful life and improve their economics.
The $500 billion figure is not money already committed or deployed. NVIDIA described it as capital the partnerships aim to mobilize over time, and the agreements remain subject to final contracts.
Yahoo Finance reported that the model could help qualified AI labs, enterprises and cloud providers gain access to AI factory infrastructure at scale. In practice, that could give neoclouds and other infrastructure operators more capacity to compete with hyperscalers that have spent heavily on their own AI platforms.
Depreciation could complicate the math
The financing model depends on NVIDIA GPUs retaining enough productive and resale value to support loans over time. CNBC reported that hardware depreciation is one of the plan’s central risks because lenders may eventually need to repossess and resell chips if borrowers default.
“Depreciation is the one key risk here,” Ben Emons, founder of FedWatch Advisors, told CNBC. He estimated that investors could demand yields between 11% and 17%, depending on where they sit in the capital structure, if GPUs are treated as high-depreciation equipment rather than long-lived infrastructure.
CNBC also cited a Bank of America Securities note saying likely borrowers could include non-investment-grade AI startups and neoclouds. The same report pointed to another risk: cheaper compute from China could put additional pressure on hardware values if it enters the market at scale.
NVIDIA’s counterargument is that its CUDA software layer can keep older GPUs productive for longer, helping support the economics behind the financing.
What the $500B plan could mean for the channel
If the MOUs turn into final agreements, the platforms could give more customers a way to finance expensive GPU infrastructure rather than relying entirely on their own capital budgets.
For cloud providers, managed service firms, resellers and infrastructure specialists, that could change how some AI projects get funded and sold.
Customers that might struggle to finance large deployments upfront could potentially tap dedicated capital pools instead, while partners could remain involved in designing, deploying and operating the underlying infrastructure.
The details will determine how useful the model becomes in practice. Partners will need to watch who qualifies for financing, what rates borrowers pay, how lenders value used NVIDIA hardware, and whether the final agreements include any form of risk-sharing from NVIDIA.
For now, the $500 billion target shows how much outside capital NVIDIA and its financial partners want to bring into the AI buildout. The next test is whether those financing structures can turn that capital into projects customers can actually afford.
NVIDIA is expanding its AI Factory presence through Equinix deployments with Cisco, while Presidio’s P.A.T.H. Lab gives enterprises another environment to test AI infrastructure before broader rollout.