NVIDIA announced on August 10, 2026 a partnership with six major firms, Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR, to launch platforms financing AI infrastructure. Over time they aim to mobilize more than 500 billion dollars (about 2 trillion zloty) of third-party capital to build data centers.
Key takeaways
- Announced August 10, 2026
- Financial partners: Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, KKR
- Goal: mobilize over 500B USD (about 2 trillion zloty) of third-party capital for AI infrastructure
- The platforms aim to give NVIDIA customers cheaper access to scarce compute
- The deals are memorandums for now, subject to final agreements
Independent capital for data centers
NVIDIA describes the initiative as independent compute financing platforms, creating pools of capital on attractive terms for the company's customers. The aim is to secure access to scarce AI resources at scale, using long-term institutional capital rather than NVIDIA's own funds alone.
The company argues its hardware suits this role because it is broadly adopted, flexible and transferable between customers, and holds value longer thanks to CUDA software.
NVIDIA compute is broadly adopted, flexible across models and workloads, fungible and transferable across customers and operators, and continuously improved through CUDA software, extending its useful life and improving its economics over time.
Jensen Huang, CEO of NVIDIA.
Aging GPUs and the question of risk
Jensen Huang addressed concerns about circular financing?circular financing: An arrangement where a company funds customers who then use that money to buy its products. Cash circulates in a closed loop., saying the plan is meant to bring independent, long-term capital into the market. According to TechCrunch's analysis, part of the plan also covers aging chips, NVIDIA would guarantee part of the value of GPUs used as collateral, covering up to 25% of any shortfall on their sale. The company's official statement does not mention such a guarantee explicitly, stressing instead the hardware's durability and fungibility.
NVIDIA has previously committed billions to customers such as OpenAI, Anthropic and CoreWeave, which raised questions about dependencies in the financing chain.
Why it matters
The 500 billion dollar scale shows the AI bottleneck is no longer just the chip, but how to finance the data centers that house it. By bringing in large funds, NVIDIA tries to spread risk and make infrastructure less dependent on its own balance sheet and on the circular financing that investors criticize. At the same time, hardware value guarantees, if they materialize, shift part of the GPU aging risk back to the maker. It signals that the market is starting to treat compute as a financial asset, with its own secondary market and hedging mechanisms.
What's next
- The memorandums remain subject to final agreements, per NVIDIA, only those will settle the real scale of the financing
- The emergence of a secondary market for older GPUs will test whether compute really holds value as collateral
