AM Intelligence (AMI), an AI infrastructure platform established by Greenko’s promoters, has placed two new firm orders for 20,000 NVIDIA Rubin graphics processing units for deployment in India and Malaysia. The latest order adds about 70 megawatts of compute capacity and takes AMI’s total committed frontier compute capacity to about 29,000 Rubin GPUs and close to 100 MW.
GPU Orders and Deployment Plan
The 20,000 GPUs are scheduled for delivery in the second quarter of 2027. AMI had earlier placed an order for 9,000 NVIDIA Rubin GPUs in August 2026 for its first AI factory in Hyderabad, India. The new and earlier orders together form the first committed tranche of AMI’s compute-as-a-service pipeline.
NVIDIA Vera Rubin NVL72 Systems
The GPUs will be deployed as NVIDIA Vera Rubin NVL72 rack-scale systems. These systems use high rack power density and liquid-cooling, which are design features used for large AI clusters and future generations of AI silicon. Rack-scale systems combine multiple computing components into a single integrated deployment architecture.
Compute Capacity and Investment Plan
AMI plans to bring another 300 MW of capacity to market in the next 15 months. The company has estimated capital expenditure of more than 20 billion for this expansion, in addition to the 6 billion already committed for the first 100 MW. AMI has also stated a target of 1 GW of Compute-as-a-Service capacity globally.
Important Facts for Exams
- GPU stands for graphics processing unit, and it is widely used in parallel computing and artificial intelligence workloads.
- Liquid-cooling is a common data-centre method for managing heat in high-density computing systems.
- MW, or megawatt, is a unit of power equal to one million watts.
- Hyderabad is a major Indian technology hub and a location for AI infrastructure projects.
Global AI Infrastructure Context
AMI’s larger programme covers 5 GW of powered AI data centres across India, the United States, and Europe. The company’s planned deployments also include Malaysia as part of its multi-country compute network.
