Amazon seeks to offload $8 billion of Nvidia chips

By Splitfeed EditorialLast updated Oct 2, 2026, 16:01 UTC

On October 2, 2026, Amazon moved to transfer roughly $8 billion worth of Nvidia AI chips to outside investors in a deal that at least one outlet, ABC Money, characterised as a leaseback rather than a conventional sale. Under a leaseback structure, Amazon would sell ownership of the chips to third parties while continuing to use them operationally, freeing up capital without slowing its AI infrastructure deployment. Broadcom was named in related activity on the same date, with Bloomberg reporting that both companies were testing the limits of investor demand for AI buildout financing, suggesting this is part of an emerging pattern across the industry rather than an isolated move by Amazon.

The significance of the transaction lies in what it reveals about the economics of AI infrastructure at scale. Spending on Nvidia chips has become one of the largest capital line items for major cloud providers, and shifting that ownership off the balance sheet reduces reported debt and capital expenditure while preserving operational capacity. Whether this represents savvy financial engineering or a sign that the cost burden of AI buildout is becoming difficult to absorb internally is a question the deal raises but does not answer. The leaseback framing also matters because it changes how investors and regulators would assess Amazon's actual exposure to AI hardware depreciation and obsolescence risk, which for chips running at this pace of generational turnover is not trivial.

None of the three camps, Pro-AI, Anti-AI, or Middle Ground, had published reactions to this story at the time of reporting. Pro-AI voices would typically treat a deal like this as evidence that AI infrastructure investment is maturing into a recognisable asset class, attracting outside capital the way data centres and fibre networks once did. Anti-AI commentators would more likely read it as a warning sign, arguing that if hyperscalers are structuring deals to move chip costs off their books, the financial case for continued AI spending may be weaker than publicly stated. Middle Ground observers would probably focus on the structural question, asking whether leaseback arrangements introduce new systemic risks if chip valuations fall or if the investors taking on ownership have limited ability to absorb losses.

The clearest near-term signal to watch is whether similar transactions close and at what terms, since pricing and investor take-up would indicate how much appetite actually exists for AI hardware as a financeable asset. Any disclosure from Amazon in its next quarterly earnings about capital expenditure treatment or off-balance-sheet commitments would also clarify how material this arrangement is to the company's broader financial position.