Amazon Plans to Sell $8 Billion of NVIDIA Chips! AI Infrastructure Enters a New Financing Era
A new twist has emerged in the AI boom.
According to reports, Amazon plans to sell investors approximately $8 billion worth of NVIDIA AI chip assets through a special purpose vehicle (SPV), then continue using these chips via a lease arrangement.
In simple terms:
Amazon transfers some of its already deployed NVIDIA Grace Blackwell-series AI chips to investors to hold, while continuing to lease them back for use.
What’s the point of doing this?
① Reducing the funding pressure for AI infrastructure
Building AI data centers requires massive capital investment.
GPU procurement, electricity, data center construction, and networking equipment all demand substantial funding.
Through asset securitization, tech companies can unlock the value of their assets while still obtaining computing power.
② AI chips are becoming “investable assets”
In the past:
Chips = corporate equipment
Now:
AI chips → generate computing-related revenue → can be financed
This means AI infrastructure is starting to adopt financial models similar to large assets like aircraft and energy equipment.
③ AI competition is entering a “capital efficiency” phase
In the past few years, the big players competed by:
Who buys more GPUs
Who builds more data centers
In the future, the competition may shift to:
Who can expand AI computing power more efficiently with capital.
Amazon has also continued to deepen its partnership with NVIDIA—AWS plans to deploy more NVIDIA GPUs to support AI training, agents, and enterprise applications.
However, this also reflects a reality:
AI infrastructure is expanding rapidly, and costs are rising just as quickly.
What the market is watching is not only:
“How many GPUs are purchased”
but also:
How much revenue that computing power can ultimately generate.
In one sentence:
In the AI era, competition is upgrading from a chip battle to a comprehensive contest of computing power, capital, and business models.
#NVIDIA #亚马逊拟售80亿美元英伟达芯片