We checked this claim: is it true that
Amazon plans an outside investment vehicle for roughly $8 billion in Nvidia chips to shift obsolescence risk to investors?
Preliminary discussions reported by the Financial Times have not resulted in a finalized deal and remain unconfirmed by Amazon or Nvidia.
Amazon explores $8 billion deal to shed chip risks
The cloud company is weighing an outside vehicle to fund AI processors without carrying balance sheet depreciation.
In a nutshell
Amazon is exploring whether to transfer about $8 billion of Nvidia Grace Blackwell artificial intelligence processors into an outside financial vehicle. Using a sale-and-leaseback structure across facilities in at least five U.S. states, the cloud provider would continue running the processors while transferring the risks of equipment depreciation and rapid obsolescence to external investors. The setup would help protect cash flow as Amazon faces up to $220 billion in projected capital spending in 2026, though neither company has confirmed the unfinalized plan.
Highlights
- Amazon is considering moving roughly $8 billion of advanced Nvidia artificial intelligence processors into an external investment vehicle.
- The proposed transaction uses a sale-and-leaseback structure across facilities in at least five U.S. states and more than a dozen data centers.
- External institutional investors could receive debt returns and an equity stake of up to 10 percent in the vehicle.
- Amazon projects total capital spending of up to $220 billion in 2026.
- Talks remain unconfirmed and unfinalized, with both Amazon and Nvidia declining public comment.
From the Editor’s Diary
When computing hardware loses value faster than conventional buildings or machinery, technology companies will try to shift equipment risk to outside investors rather than carry the loss on their own books.
Who's involved
Amazon
Cloud computing and online retail company
goal → Seeks to expand computing capacity while keeping equipment depreciation off its corporate accounts
Nvidia
Designer of advanced artificial intelligence processors
goal → Aims to maintain large-scale shipments of its high-end computing systems to major cloud operators
Institutional Investors
Investment funds, lenders, and pension managers
goal → Seek regular interest income from debt and up to 10 percent ownership in high-demand computing hardware
In short
Tech giants face the risk of holding billions of dollars in computer chips that can become outdated within years, threatening their cash reserves. If completed, Amazon's proposal would pioneer a way for cloud providers to rent rather than own cutting-edge hardware, passing both the upfront cost and the risk of fast depreciation to outside investors.
A finalized transaction would make this funding setup a model for other artificial intelligence operators to keep building massive computer systems without loading debt onto their own books.
Whether this deal happens is uncertain because discussions remain exploratory, neither Amazon nor Nvidia has verified them, and no formal contracts or pricing terms exist.
How it unfolded
Financial Times reports Amazon talks on chip financing entity
Amazon began exploratory talks with institutional investors to move about $8 billion in Nvidia Grace Blackwell chips into an external financial structure, the Financial Times reported on October 2, 2026. The structure was planned as a sale-and-leaseback across facilities in at least five U.S. states, allowing Amazon Web Services to operate the computing hardware while removing the heavy equipment cost from its balance sheet.
Market examines off-balance-sheet strategy as companies stay silent
The initial report prompted immediate attention across global media and online trading forums, where observers debated whether the move represented financial innovation or an urgent hedge against rapid hardware depreciation. Both Amazon and Nvidia declined to comment publicly, leaving the transaction unconfirmed and preserving its status as an exploratory concept through October 3, 2026.
Where things stand
Talks remain at an exploratory stage as of October 3, 2026, with no regulatory filings, finalized investor agreements, or confirmed debt terms. Market participants are waiting for future corporate earnings disclosures to see if off-balance-sheet structures become a recognized template for funding artificial intelligence infrastructure.