AM Intelligence has ordered 9,000 Nvidia Rubin GPUs for a 30MW Hyderabad AI facility, forming the first phase of its planned 1GW compute-as-a-service network.
Data centre company AM Intelligence (AMI) Group has placed an order for 9,000 Nvidia GPUs to deploy 30MW of artificial intelligence computing capacity at its upcoming facility in Hyderabad. The project is part of AMI’s larger plan to develop 1GW of AI infrastructure across multiple markets with an investment of $8 billion.
AMI Group President Mahesh Kolli said the company expects its 1GW computing capacity to be offered through a compute-as-a-service model by 2030.
The company has placed a firm order for Nvidia’s Vera Rubin systems, with deliveries expected in the first quarter of 2027. The 9,000 GPUs will be deployed in Vera Rubin NVL72 rack-scale systems at AMI’s first AI factory in Hyderabad.
According to AMI, the Hyderabad facility will form the first phase of its planned 1GW compute-as-a-service network, which will span India, the US, Finland and Malaysia. The company expects the site to become one of the early frontier AI computing clusters in Asia.
AMI targets 5GW AI data centre capacity
AMI is also developing plans for up to 5GW of powered AI data centre capacity across India, the US and Europe. The company has not yet provided a timeline for achieving this larger target.
The planned facilities will support high-density AI workloads and use liquid-cooling systems to manage the power consumption and heat generated by advanced AI processors. AMI said the infrastructure will be designed to accommodate successive generations of AI chips, helping extend the usability of the facilities as computing requirements evolve.
The company also intends to build 1GW of global compute-as-a-service capacity for AI workloads, with an initial 200MW expected to become operational in the near term.
The Hyderabad investment highlights the growing demand for dedicated AI infrastructure as businesses and AI developers seek access to large-scale GPU computing capacity. Nvidia’s upcoming Rubin platform is expected to provide the computing capabilities required for increasingly complex AI workloads.
