Amazon faces $1B Nvidia chip inventory issue as data centers lack power

by@zerohedge

Summary

Amazon is currently facing a significant inventory challenge, with billions of Nvidia chips, specifically the GB300 racks, accumulating in warehouses rather than being deployed in operational data centers due to infrastructure bottlenecks, particularly in power and supporting electrical systems. Reports indicate that Amazon has begun implementing an "Early Rack Receipt" process to acquire these expensive chips earlier in the supply chain, motivated by fierce competition among cloud providers for AI hardware. However, this has resulted in the company paying for equipment it cannot yet utilize, leading to negative cash flow for its AWS division, which is expected to persist for the next couple of years. The scarcity of energized compute continues to drive up rental rates, emphasizing the mismatch between chip procurement and site readiness.

Tokens

$NVDA$AMZN

Analysis

Amazon: Amazon operates AWS, the world's largest cloud computing platform, with extensive investments in AI infrastructure including data centers and server deployments. The company is actively scaling its machine learning capabilities amid rapid demand growth. The news highlights internal AWS processes for receiving and storing Nvidia hardware ahead of powered data center readiness. Nvidia: Nvidia designs and manufactures advanced GPUs and related hardware essential for AI training and inference workloads. Its latest GB300 racks are central to hyperscaler data center expansions. In the current news, Nvidia chips form the core of the reported inventory buildup at Amazon facilities due to downstream deployment constraints. John Felton: John Felton serves as CFO of AWS, overseeing financial strategy for the cloud division's operations and investments. He participates in internal discussions on the impacts of AI buildout spending. In the reported meeting, Felton addressed the free cash flow effects of accelerated infrastructure commitments. Prasad Kalyanaraman: Prasad Kalyanaraman leads key aspects of Amazon's AI infrastructure development and supply chain efforts within AWS. He manages forecasts and deployment timelines for advanced computing hardware. The news references his statements on demand uncertainty and the challenges of synchronizing equipment arrivals with site readiness. Capex Timing: Major cloud providers are adjusting procurement processes to secure AI hardware earlier amid intense competition for components, even when full deployment sites are not yet ready. Compute Scarcity: Energized and operational AI compute remains in high demand, supporting strong rental rates despite broader supply chain expansions. Infrastructure Bottlenecks: Power availability, grid interconnections, and supporting electrical infrastructure often lag behind hardware delivery schedules for large-scale data center projects.

Categories

predictionstechaimacroai_agents
View Original Tweet