Amazon is exploring a transaction to move roughly $8 billion of advanced Nvidia Grace Blackwell chips off its balance sheet through a special purpose vehicle funded by external investors. The arrangement allows the cloud computing giant to lease the hardware back while shifting high-cost semiconductor assets away from its direct financial reporting.
Tech giants are reshaping how they finance the astronomical capital required to build out artificial intelligence infrastructure. According to reporting detailed by financial industry coverage, Amazon has spent recent weeks holding exploratory conversations with outside investors to gauge interest in the off-balance-sheet vehicle.
The proposed structure targets thousands of advanced graphics processing units already installed or being deployed across more than a dozen data centers in the United States. Under the framework outlined in the reporting from regional financial channels, the special purpose vehicle would issue debt instruments and sell up to a 10% equity stake to external backers, enabling Amazon to transition toward a lighter asset model while maintaining uninterrupted access to the computing power.
Special Purpose Entity Seeks Investment-Grade Credit Rating
The physical footprint of the proposed deal spans multiple American states. As noted in banking sector analyses, the hardware involved is situated within more than 12 data centers spread across five U.S. states, including Virginia and Nevada.
Should the special purpose entity secure an investment-grade credit rating, long-term institutional backers such as pension funds and insurance companies could join the financing syndicate. As market coverage points out, this arrangement allows Amazon to utilize the hardware continuously via operating leases while transferring a significant portion of the capital expenditure burden off its primary balance sheets.
Amazon Web Services Increases Capital Outlays
The financing discussions coincide with a surge in capital outlays by Amazon Web Services.
Hyperscale Cloud Providers Secure Constrained Computing Capacity
The push to find alternative funding models reflects a wider scramble among hyperscale cloud providers, including Microsoft and Alphabet’s Google, to secure constrained computing capacity. Industry analysts note that advanced data center hardware faces rapid iteration cycles requiring replacement roughly every five years.
Robert Schiffman of Bloomberg Intelligence said that “the scope of AI infrastructure financing is expanding beyond the capital expenditure boom, with total spending by hyperscale cloud providers potentially approaching $9 trillion through 2031, while debt issuances could reach $500 billion in 2027.”
