AMI Orders NVIDIA Rubin Systems for Hyderabad Factory
On August 25, 2026, AM Intelligence placed a binding order for about 9,000 NVIDIA Vera Rubin systems for its first AI factory in Hyderabad, Business Standard reports. Economic Times reports delivery is scheduled for the first quarter of 2027 and that the hardware will power an initial 30 MW phase. AMI has outlined more than $8 billion in capital expenditure for an initial 200 MW of compute capacity.
AM Intelligence (AMI), an AI infrastructure platform established by Greenko's promoters, has placed a binding order for about 9,000 NVIDIA Vera Rubin systems for a Hyderabad AI factory, according to Business Standard. Economic Times reports that the order comprises 9,000 Vera Rubin NVL72 rack-scale systems, scheduled for delivery in the first quarter of 2027, for the facility's initial 30 MW phase.
Business Standard reports that AMI intends to bring an initial 200 MW of compute-as-a-service capacity online in the near term, requiring more than $8 billion in capital expenditure. The Hyderabad site is the first tranche of a reported 1 GW compute-capacity program spanning India, the United States, Finland, and Malaysia.
Compute and power buildout
The companies' reported configuration combines NVIDIA's Vera Rubin architecture with high-speed data movement and advanced storage. Business Standard reports that the Hyderabad factory is being designed for very large AI workloads, including trillion-parameter models and agentic AI applications.
Anil Chalamalasetty, chairman of AM Intelligence, described the effort as an "electron-to-token" opportunity, linking the group's power infrastructure experience to frontier AI compute. Economic Times reports that AMI characterized the Hyderabad deployment as among the first frontier AI compute clusters in Asia.
The scale distinction matters for infrastructure teams. A 200 MW target describes planned compute-as-a-service capacity, while the initial 30 MW phase described by Economic Times is the nearer hardware deployment stage. In comparable large-scale AI buildouts, GPU or rack availability alone is insufficient: networking, storage throughput, cooling, power delivery, and cluster orchestration determine whether capacity can be converted into usable training and inference service.
What remains to be established
The reports do not provide pricing for the NVIDIA order, a detailed facility completion schedule, or customer commitments for the compute capacity. They also describe the ordered hardware differently: Business Standard refers to about 9,000 Rubin GPUs, while Economic Times identifies 9,000 Vera Rubin NVL72 rack-scale systems. That distinction is material because an NVL72 is a rack-scale configuration rather than a single accelerator.
For AI practitioners, the reported first-quarter 2027 delivery date places this infrastructure beyond immediate capacity planning. If commissioned on schedule, it would add a substantial prospective regional source of high-end compute for workloads that require large multi-GPU clusters.
Key Points
- 1AMI's reported NVIDIA order targets a Hyderabad AI factory, with Economic Times identifying a 30 MW initial deployment phase in 2027.
- 2Business Standard reports more than $8 billion of capex for 200 MW, underscoring the capital intensity of frontier AI infrastructure.
- 3Comparable AI factory deployments depend on networking, storage, cooling, and orchestration, not accelerator procurement alone, to deliver usable compute.
Scoring Rationale
The reported order and planned 200 MW deployment represent a notable prospective expansion of frontier AI compute capacity in India and Asia. The infrastructure is not scheduled for delivery until the first quarter of 2027, and public reporting does not yet establish customer availability or operating details.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
