AWS and NVIDIA Expand AI Infrastructure Partnership
AWS and NVIDIA will deploy an additional 2 million advanced GPUs from 2027 to 2028 to expand cloud computing infrastructure for agentic and physical AI.
What are the key facts?
- 1Additional deployment of 2 million GPUs
- 2Blackwell Ultra and Rubin architectures
- 3Deployment planned for 2027–2028
What happened?
Amazon Web Services (AWS) and NVIDIA announced an expansion of their partnership in artificial intelligence infrastructure. According to the announcement, AWS plans to deploy a total of 2 million additional NVIDIA GPUs in its cloud infrastructure between 2027 and 2028. The chips will include next-generation architectures such as Blackwell Ultra and Rubin. The announcement said these computing resources will support the operational needs of next-generation applications including agentic AI and physical AI.
The partnership covers NVIDIA’s advanced GPUs, AWS cloud computing infrastructure, and related systems for artificial intelligence workloads. The companies said that as agentic and physical AI applications develop, market demand for large-scale computing power and next-generation infrastructure continues to grow. AWS will expand its cloud GPU resources to provide infrastructure support for related AI training, inference, and application deployment.
What does this mean for cross-border sellers?
Continued expansion of cloud computing capacity will accelerate the iteration of AI tools for cross-border ecommerce in customer service, content, and operations. However, do not equate increased underlying computing power with automatic store growth; overlooking data quality can still create a competitive gap. Highest-priority action: This week, identify the most time-consuming manual task in your store, test an AI tool in one scenario, and record its accuracy and manual review cost.