Nvidia announced on August 10 that it will create independent compute financing platforms with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR, aiming to mobilize over $500 billion of third-party capital for AI infrastructure over time. The memorandums of understanding cover the first compute financing platforms of their kind at global scale, built to let long-term capital underwrite the AI factory buildout instead of Nvidia customers and their balance sheets carrying the full load.
The structure is a new category of financing rather than a single fund. Nvidia says it will work with the six firms to create dedicated pools of capital at significant scale and at attractive rates for Nvidia customers, spanning frontier AI labs, enterprises and AI clouds. Under the partnerships, Nvidia compute is positioned as an investable asset with a long, predictable revenue life, supported by the CUDA software ecosystem that keeps the hardware productive across models and workloads. The platforms remain subject to execution of final agreements, so the first capital pools have no announced close date.
Compute as an asset class
CEO Jensen Huang framed the announcement as a turning point for the company. “We began by building chips; today, we are helping create a new class of productive, investable infrastructure: AI factories,” he said in the release. “In AI, compute is revenue.” Huang said the platforms will help customers access scarce compute at scale and build the DSX AI factories Nvidia says will power industries and countries, and he argued Nvidia compute is broadly adopted, fungible across customers and operators, and continuously improved through CUDA software, which extends its useful life and improves its economics over time.
The investor side echoed the scarcity argument. Apollo President Jim Zelter called modern compute a scarce, mission-critical asset class positioned to drive long-term growth, and said Apollo’s flexible long-term capital base pairs with Nvidia’s ecosystem as part of what he called the Global Industrial Renaissance. BlackRock Chairman and CEO Larry Fink said the partnership deepens BlackRock’s relationship with Nvidia, including through the AI Infrastructure Partnership, and connects long-term capital to essential infrastructure.
Blackstone President and COO Jon Gray said Blackstone continues to be an enormous investor across the Nvidia ecosystem. Brookfield CEO Bruce Flatt said the collaboration will build and fund the backbone of AI globally. Goldman Sachs Chairman and CEO David Solomon described the opportunity to create a market for credit backed by Nvidia compute, and KKR co-CEOs Joe Bae and Scott Nuttall said KKR will pair Nvidia’s accelerated computing platform with its long-duration capital and infrastructure expertise, building on KKR’s earlier investment in Helix Digital Infrastructure.
The financing push follows a familiar pattern in the AI buildout: demand for compute is outrunning the balance sheets of the companies that need it. Countries, governments, enterprises and startups are all competing for scarce capacity, and Nvidia’s own customers, including hyperscalers and model labs, have committed hundreds of billions of dollars to data center expansion. This site has covered how the economics of that expansion have made the chip race into a power race, and financing platforms like these are the financial engineering layer underneath the physical buildout.
The deal also widens the field of institutions with direct exposure to AI compute. Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR collectively manage trillions of dollars, and each brings a different instrument to the table, from insurance-linked capital at Apollo and KKR to BlackRock’s infrastructure partnerships and Goldman’s capital markets reach. If the platforms close as described, they would create a standing market for compute-backed credit, letting operators finance systems against their expected revenue rather than against corporate balance sheets.
What comes next is execution. The release lists no timetable for the final agreements, no first-fund size and no named tenants, and the forward-looking language carries the usual caveats about regulatory approvals and market conditions. The next confirmed milestone is the signing of definitive agreements, after which the size and structure of the first dedicated capital pools will become public. Until then, the headline number, $500 billion over time, signals intent more than commitment, but the intent itself is the story: the largest GPU maker is now in the business of helping finance the data centers that buy its GPUs.
