NVIDIA teams up with major asset managers and banks to back AI data-center financing platforms

NVIDIA said it is partnering with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to create AI compute infrastructure financing platforms. The goal, according to the announcement carried by Yahoo Finance, is to mobilize more than 500 billion dollars of third-party capital to help finance “AI factories” and broaden access to the compute and infrastructure needed to run large-scale artificial intelligence workloads.

The companies’ stated approach is to treat NVIDIA’s compute and broader “full-stack AI infrastructure” as an investable, financeable proposition. In practical terms, the plan is framed around building vehicles or platforms that can route capital toward AI data-center capacity, instead of requiring every AI customer to finance and deploy hardware directly themselves.

NVIDIA’s message links the financing model to long-duration, usage-linked revenue. Usage-linked revenue typically means income that depends on how much computing capacity is actually consumed over time, rather than only on upfront hardware sales. NVIDIA positioned the financing platforms as a way to align infrastructure financing with real-world utilization of AI compute.

The partner list is weighted toward firms that already manage large pools of institutional capital and that often arrange complex financing and investment structures. Apollo and KKR are private investment firms, while BlackRock and Goldman Sachs operate across asset management and capital markets. Blackstone and Brookfield similarly have experience financing large infrastructure and real-asset projects, which can be relevant when data centers and related build-outs require long timelines and significant upfront investment.

AI compute has become a bottleneck for many enterprises and startups trying to run new model training and inference workloads at scale. The headline investment concept in NVIDIA’s announcement is that compute capacity, when packaged with financing, could become easier to acquire and scale for customers. That, in turn, could lower barriers for organizations that do not want to take on the full balance-sheet burden of hardware procurement, site development, and capacity planning.

Still, key implementation details were not provided in the Yahoo Finance item beyond the high-level intent and the named participants. The announcement text, as reflected in the page title and description, does not spell out governance terms, the specific platform legal structures, underwriting standards, or how NVIDIA’s customers would contract for access to capacity under the usage-linked approach.

For investors and market observers, the most notable question is how this financing concept will translate into day-to-day procurement, contracting, and risk allocation. For example, it is unclear from the available information whether the platforms focus mainly on new build-outs, upgrades to existing data-center capacity, or both, and what portion of the economics would be tied directly to compute consumption versus longer-term infrastructure availability.

What to watch next is whether NVIDIA or its partners publish additional documentation that clarifies the platform mechanics. That would include disclosures about timelines for launching the financing platforms, the types of AI compute configurations in scope, and any measurable targets or case studies showing how third-party capital is expected to flow to data-center capacity. For now, the public message centers on mobilizing large-scale capital and packaging AI compute as an investable infrastructure class, with usage-linked revenue as a central theme.