Amazon weighs a $8 billion financing structure for Nvidia AI chips, raising questions about how much investor-backed spending can absorb

is reportedly considering a financing structure that would shift part of its planned Nvidia AI chip purchases onto outside investors, a move that could reshape how large cloud operators fund data center expansion for artificial intelligence workloads.

The plan, as described in a report carried by Yahoo Finance, would involve moving about $8 billion worth of Nvidia Grace Blackwell chips into a special-purpose vehicle, or SPV, financed by investors. After the SPV acquires the hardware, Amazon would lease the equipment back, effectively converting a purchase decision into a longer-term financing and operating expense arrangement.

Grace Blackwell is Nvidia’s data center computing platform aimed at accelerating AI training and inference, combining specialized processors with supporting system-level components. In practical terms, the chip platform is part of the hardware stack that cloud providers buy to meet demand for AI services, and it typically requires large, upfront capital commitments because data center capacity, power, and networking must scale alongside the compute.

The reported SPV concept matters because it highlights that the debate around AI infrastructure is no longer only about chip availability and capacity planning. It is also about financing, balance-sheet treatment, and how quickly capital can be deployed without overwhelming budgets or tying up cash.

The Yahoo Finance report frames the discussion as a question of whether AI spending is becoming too heavy, implying that even large companies may seek structures that spread cost and risk. Leasing is often used to manage large equipment purchases, and an SPV structure can be a way to bring in third-party capital when direct procurement would otherwise require more immediate funding.

Amazon did not publicly confirm the details described in the report, and the post did not specify which investors would participate, what lease terms would look like, or how any regulatory, accounting, or tax treatment would be handled. It also did not disclose whether the figure of roughly $8 billion reflects a firm commitment or an early-stage estimate.

For Nvidia, the potential effect is indirect but significant. Nvidia does not control how cloud customers finance hardware, but chip demand is closely tied to customer investment cycles. If Amazon proceeds with a financing approach that makes spending more manageable, it could support continued procurement activity for Nvidia’s data center platforms, including Grace Blackwell.

For markets, the next key question is whether other large AI infrastructure buyers adopt similar financing models as they accelerate buildouts. If multiple companies pursue SPVs, leasing, or other structured financing, it could change the rhythm of capex and the way investors evaluate the durability of AI-driven infrastructure demand.