Amazon is reportedly considering a plan to transfer approximately $8 billion worth of Nvidia AI equipment into a special purpose vehicle (SPV), subsequently leasing back the computational power. This information comes from the Financial Times, citing sources familiar with the negotiations.
The arrangement involves thousands of components and systems being installed across more than 12 Amazon data centers in five U.S. states, including Nevada and Virginia. According to the publication, the equipment would be handed over to the SPV, which would secure external capital and debt financing. Amazon would continue to utilize the same computational capabilities by leasing them from the SPV.
The corporation is also contemplating offering investors up to a 10% stake in the SPV. Notably, Amazon is not expected to retain any ownership in the structure. Representatives from the tech giant have declined to comment on the matter.
Shifting Amazon's Infrastructure Financing Approach
This proposed structure resembles a sale-and-leaseback arrangement, wherein a company transfers an asset to a separate entity and then leases it back. In Amazon's case, this does not indicate a reduction in computational capacity or a move away from Nvidia equipment.
This scheme would enable the company to attract external funding for its already procured AI infrastructure while maintaining access for its own operational needs. Such SPVs are increasingly being utilized for financing expensive data centers and computational hardware.
In September, it was reported that major tech companies had provided up to $300 billion in guarantees for AI infrastructure over the past year. A significant portion of this related debt is being placed in special structures rather than as standard corporate debt for big tech firms.
This model allows for financing to be raised against specific assets or projects. Depending on the deal's structure, the tech company might lease the capacity, offer guarantees, or assume part of the risk associated with the equipment's residual value.
The new plan does not suggest a decrease in Amazon's investment in AI infrastructure. According to the Financial Times, the company's capital expenditures are expected to reach around $220 billion in 2026, with a substantial amount allocated to AWS for building data centers and acquiring computational equipment.
Wall Street's View on Nvidia as Collateral
In August, Nvidia entered into agreements with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to establish independent platforms for financing AI infrastructure. The partners aim to mobilize over $500 billion in third-party capital over time.
Nvidia is positioning its computing infrastructure as a distinct class of investment assets. However, financial institutions remain cautious in evaluating its equipment as long-term collateral. According to Reuters, lenders often estimate the depreciation period for GPUs at only three to four years due to rapid technological obsolescence.
In contrast, Nvidia argues that its computing hardware can retain its economic value for a significantly longer period. Due to differing assessments, investors are demanding additional guarantees, long-term contracts with clients, and other risk mitigation strategies.
The residual value of the equipment is becoming a crucial factor in turning AI computing into an asset that can attract long-term financing.
It is worth noting that in August, AWS and Nvidia agreed to deploy more than 2 million GPUs in cloud infrastructure during 2027-2028.
Follow ForkLog on social media
Telegram (main channel) Facebook X