Amazon explores $8B sale-leaseback of Nvidia chips for AI expansion

by@TFTC21

Summary

Amazon is currently negotiating to transfer approximately $8 billion worth of Nvidia Grace Blackwell chips into a special-purpose vehicle (SPV) and lease them back, as reported by the Financial Times. This sale-leaseback structure would allow Amazon to offload billions in capital expenditures (capex) from its balance sheet while maintaining access to the same hardware and computing capabilities. With an anticipated capex of around $220 billion in 2026 for data centers and AI, this strategy reflects a growing trend among major technology firms to utilize similar financing arrangements, particularly as rapid advancements in AI hardware contribute to uncertainty regarding the depreciation of such assets. While no agreements are finalized, the move illustrates how even the largest companies are seeking innovative financial solutions amid a competitive AI landscape.

Tokens

$AMZN$NVDA

Analysis

Amazon: Amazon operates extensive cloud infrastructure through AWS and ranks among the largest buyers of advanced AI accelerators for data center expansion. The company is actively exploring a sale-leaseback transaction involving Nvidia Grace Blackwell chips to restructure ownership of its AI hardware while retaining full operational control. This approach reflects broader efforts by major cloud providers to sustain aggressive infrastructure builds amid rising capital demands. Nvidia: Nvidia develops and supplies high-performance GPUs and AI systems, including the Grace Blackwell platform designed for large-scale training and inference workloads. It appears in the news as the chip supplier in the proposed Amazon transaction and as a participant in customer financing programs that support similar GPU-backed structures. These initiatives underscore Nvidia's central position in enabling the current wave of AI infrastructure investments across the industry. CoreWeave: CoreWeave provides specialized cloud services optimized for GPU-intensive AI and machine learning applications, relying primarily on Nvidia hardware. The news cites CoreWeave as one of several companies that have already implemented comparable GPU-backed financing arrangements. This places CoreWeave within the expanding set of AI infrastructure operators adopting creative ownership models to support growth. Financing Structures: Major technology firms are turning to sale-leaseback deals and special-purpose vehicles to manage ownership of AI accelerators while preserving compute capacity. Hardware Depreciation: Rapid advancement cycles in AI silicon create ongoing uncertainty around the residual value of GPUs used as collateral in financing arrangements.

Categories

macrotechai
View Original Tweet