NVIDIA is joining forces with Wall Street to bet on a radical proposition: whether AI chips can break the financial rule of rapid technological depreciation.
This week, NVIDIA announced a consortium with Apollo, Blackstone, GIP under BlackRock, Brookfield, Goldman Sachs, and KKR to raise up to $500 billion for AI infrastructure projects. The funds will cover chip purchases, power production, and data center construction, with financial giants providing semiconductor leasing financing to tech companies.
On August 13, according to sources familiar with the deal, the core logic behind this plan is that chip demand, driven by the AI boom, will keep prices elevated for longer than most analysts previously expected. NVIDIA CEO Jensen Huang stated that the transaction will create a new asset class with chips as the underlying asset, opening the door to the $22 trillion private capital industry.
However, this grand vision faces a fundamental risk: whether financial institutions can accurately gauge the durability and long-term value of NVIDIA's chips remains uncertain. Unlike leasing cars or commercial aircraft, chip technology evolves rapidly and demand fluctuates sharply, making their long-term value highly unpredictable. This $500 billion gamble is essentially a historic bet on whether AI computing demand will persist and whether chips can become a reliable financial asset.
Transaction Structure: Private Capital Enters, NVIDIA Offloads Balance Sheet Pressure
The core structure of this deal involves securitizing chip lease receivables and selling them to debt buyers like insurance companies.
According to reports, private capital firms like Apollo, KKR, and Brookfield plan to package GPU lease contracts into securities, leveraging their managed insurance asset pools to mobilize large-scale funds. Some institutions are considering setting up special-purpose financing vehicles, referencing the collateralized loan obligation (CLO) structure common in private equity buyouts, to slice chip assets into tranches for investors with varying risk appetites.
"You can finance GPUs like a CLO, with different risk layers," one executive involved in the deal said, though others remain cautious, noting that "each institution's financing approach will differ."
For NVIDIA, this arrangement also carries significant financial implications. Bank of America research analyst Vivek Arya pointed out this week that Wall Street's involvement means NVIDIA can gradually exit the "vendor financing" model, where it directly provided guarantees for clients to help them raise capital in the market.
"The burden falls on the consortium, not NVIDIA's balance sheet," Arya said, calling this a clear positive for the company.
Core Controversy: Can Chips Break Tech's Depreciation Laws?
The biggest uncertainty in this deal is whether the long-term value of chip assets can support financialization.
Traditional financial logic has always been cautious toward tech assets. Currently, most lenders require debt backed by chip leases to be fully repaid within three to five years, assuming the underlying asset's value will approach zero thereafter. Unlike leasing cars or commercial aircraft, which have mature secondary markets when customers default, chips, once technologically obsolete, have almost no reliable buyers.
Ben Bajarin, a tech analyst at Creative Strategies, highlighted the risks: "The entire premise relies on continuous investment. There's a risk of overbuilding, demand could slow, models could improve, and we might not need as much computing power."
Additionally, chip buyers or lessees must simultaneously build data centers, meaning expected revenue streams will lag significantly behind spending plans, further straining cash flow.
NVIDIA's Rebuttal: Older Chips Still Create Value
Facing skepticism, Jensen Huang backed his logic with real data.
Huang posted on social media platform X on Monday that NVIDIA's recent chip lease pricing has risen, and even its A100 chip, launched six years ago, remains in use, exceeding its original lifespan. He has repeatedly emphasized that even older H100 chips retain value far beyond market expectations due to the explosive growth in computing demand.
NVIDIA's strategy goes further. By continuously updating its software platform, Cuda, the company actively extends chip lifespans. Cuda is central to NVIDIA's dominance in AI processing, enabling customers to repurpose GPUs originally designed for graphics to accelerate AI applications.
Bajarin believes that strong demand for all chip types will support NVIDIA's plan in the medium term, while long-term success depends on the company's ability to consistently lower its products' total cost of ownership.
On risk-sharing, NVIDIA has made a substantial commitment: guaranteeing that chips retain at least 25% of their residual value during the lease period, with the $5.3 trillion chip giant absorbing initial losses.
According to a report, one executive involved in the $500 billion deal said, "GPUs have intrinsic value. NVIDIA has a decade-long track record on GPU lease pricing," describing the company's guarantee as a "first-loss protection" in future financing, meaning NVIDIA will absorb early value losses beyond expectations.
Despite the risks, Wall Street's enthusiasm is already ignited.
Blackstone's Jon Gray and BlackRock's Larry Fink have both publicly stated that their institutions are ready to deploy capital to capture the massive demand for new data centers, which will train and run the latest AI models.
The report noted that another executive working with NVIDIA predicted that GPU leasing will drive pricing standardization and create economies of scale for AI companies. He said that as "people figure out the ropes," public debt market investors will eventually enter this field, enhancing the liquidity and tradability of this asset class.
Analysts believe the key to this process is whether the market can reach a consensus on chips' long-term value. If NVIDIA's 25% residual value guarantee withstands market testing, chip lease securitization could become another mainstream asset class spawned by the AI investment wave, following data center infrastructure debt. But if AI demand cools and computing power growth slows, this $500 billion gamble led by Jensen Huang will face a severe stress test.
Comments