NVIDIA Partners with Six Wall Street Institutions to Raise $500 Billion for AI Computing Power, Highlighting Hardware Depreciation Risks

NVIDIA CEO Jensen Huang announced on August 10, 2026, that the company has signed memorandums of understanding with six institutions—Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR—aiming to progressively raise more than $500 billion in third-party long-term capital for AI factory construction.

Financing Structure and Asset Definition

The plan defines GPU clusters as infrastructure assets akin to electricity or highways, emphasizing their ability to generate stable cash flows and be reused across customers. According to Forbes, each gigawatt of computing power corresponds to an investment scale of approximately $50–60 billion. NVIDIA has expressed its willingness to provide up to 25% residual value support in certain projects.

Many excellent AI companies, enterprises, and AI cloud service providers have demand for computing power, but have not yet obtained financing of sufficient scale or at sufficiently low cost to build rapidly. —Jensen Huang, CEO of NVIDIA

The memorandum is not a final contract; each project requires a separate agreement, and no specific timeline or capital allocation ratios have been announced to date.

The Fundamental Conflict Between Hardware Aging and Loan Tenors

The core difficulty lies in the fact that AI hardware refreshes far faster than traditional infrastructure. Amazon has already shortened its server depreciation cycle, directly reflecting how technological iteration compresses asset lifespans. If the loan tenor exceeds the hardware's actual profitability window, the question of who bears the losses will become critical.

Whether NVIDIA's residual value support mechanism can cover large-scale default scenarios depends on the specific contract terms. Strong lease prices do not automatically translate into high resale value—especially once new architectures such as Rubin Ultra are released, the market value of existing clusters may decline rapidly.

Impact on the Competitive Landscape of Cloud Giants

Alphabet's share price fell approximately 2% following the announcement. Google has long invested in in-house TPUs to reduce its dependence on NVIDIA. This financing lowers the barrier to acquiring GPU computing power, thereby weakening the exclusivity advantage of self-developed chips. Amazon and Microsoft face similar pressure.

Institutions such as BlackRock manage trillions of dollars in assets and have previously expanded their infrastructure investment capabilities. Goldman Sachs contributes asset securitization expertise, while private equity firms such as Apollo and KKR specialize in identifying long-cycle cash flow projects. The funds will ultimately flow to AI labs, large enterprises, and cloud service providers.

External Warnings on Financial Stability Risks

The Bank of England's July Financial Stability Report noted that if AI-related debt financing grows as expected, losses or debt defaults by AI companies in the event of a shock would have a material impact on global financing conditions. The report also reminded banks and private credit institutions to fully assess their risk exposure to the AI industry.

Global AI capital expenditure is projected to exceed $730 billion in 2026. Easier financing could push this figure even higher while simultaneously amplifying the systemic risks posed by technological obsolescence.

Independent Assessment

At its core, this financing represents NVIDIA's attempt to convert its own products into securitizable infrastructure assets in order to accelerate computing power supply. However, the maturity mismatch between hardware refresh cycles and long-term debt has not been fundamentally resolved by this arrangement. The actual outcome will depend on the specific contractual provisions allocating technology depreciation risk in future agreements, rather than on the sheer scale of capital.