NVIDIA announced on August 10, 2026, that it has signed memorandums of understanding with six major financial firms — Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR — to establish independent compute financing platforms. The platforms aim to mobilize more than $500 billion of third-party capital for AI infrastructure buildout over time.

The financing platforms are designed to provide dedicated pools of capital at scale and attractive rates for companies that need compute capacity but lack access to financing at the required scale or cost. NVIDIA CEO Jensen Huang described the arrangement as a "major milestone for NVIDIA and the AI industry," framing AI factories as the infrastructure of the intelligence era akin to railroads or electricity in prior industrial revolutions.

What's new

  • Partners: Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, KKR
  • Target capital: Over $500 billion of third-party capital over time
  • Structure: Independent compute financing platforms, not direct NVIDIA lending
  • NVIDIA's role: Guaranteeing up to 25% of the residual value difference for GPUs used as collateral if they depreciate faster than expected
  • Status: Preliminary agreements (MOUs); no disclosed timetable, individual firm commitments, or financial terms

The platforms are intended to serve a broad ecosystem — AI companies, enterprises, sovereign AI initiatives, and neocloud providers such as CoreWeave, Nebius, Firmus, and Lambda. Huang emphasized that the model brings "independent, long-term institutional capital" into the market, distinguishing it from vendor financing where the equipment maker bears the credit risk.

Why it matters

The deal addresses a growing bottleneck. Hyperscalers and AI labs have signaled over $730 billion in AI spending for 2026, yet traditional financing channels are straining. Oracle has taken on significant debt. Google has issued equity. Meta has burned cash. Microsoft CEO Satya Nadella recently referenced the 1873 railroad financial crisis as a cautionary parallel. By unlocking institutional capital — pension funds, insurance companies, sovereign wealth — NVIDIA aims to diversify the funding base beyond the balance sheets of a handful of mega-cap tech firms.

The residual-value guarantee is the mechanism that makes the collateral financeable. If a data center operator defaults and the lender must liquidate GPUs, NVIDIA covers up to 25% of any shortfall between book value and market price. This creates what financiers call "wrong-way risk": NVIDIA's obligation grows precisely when demand for its chips weakens, which would also pressure its revenue. Huang has argued the risk is bounded because the platforms are independent and the guarantee covers only a portion of the collateral value.

The Bank of England warned in July that AI debt financing could pose a risk to financial stability if companies taking on debt fail to deliver sustainable profits or face big disruptions. Policymakers noted it may be difficult for financial firms to see the full extent of their direct and indirect exposures to the AI ecosystem, which could increase the risk that exposures are greater than anticipated or not fully reflected in risk management.

Our take

The guarantee is a calculated bet that AI compute will behave more like long-lived infrastructure than consumer electronics. If a secondary market for used GPUs develops — where a three-year-old H100 cluster can be repurposed by a different cloud, enterprise, or nation — the 25% backstop may rarely be triggered. But the model ties NVIDIA's balance sheet to the residual value of its own aging products, a feedback loop that deserves scrutiny as the GPU installed base expands into the millions.

Sources