AM Intelligence Scales Up AI Infrastructure with Massive NVIDIA GPU Order
HYDERABAD – AM Intelligence (AMI), the burgeoning artificial intelligence infrastructure platform founded by the promoters of Greenko, has announced a significant expansion of its computing capabilities. The company has secured two new binding orders for 20,000 NVIDIA Rubin GPUs, bringing its total commitment to approximately 29,000 units. This strategic move positions AMI as one of the largest purchasers of frontier AI hardware globally, trailing only major players in North America.
The latest acquisition, which builds upon an initial order of 9,000 GPUs placed in August 2026, represents a combined capacity of nearly 100 megawatts (MW). This infrastructure is slated for deployment across facilities in India and Malaysia, with a delivery timeline set for the second quarter of 2027.
Advancing the Global AI Factory Blueprint
These new orders are central to AMI’s broader mission to establish a “Compute-as-a-Service” model that supports the rapidly growing demands of hyperscalers, sovereign AI initiatives, and frontier labs. The company is currently executing a pipeline of approximately 400 MW of compute capacity, with plans to introduce an additional 300 MW to the market over the next 15 months.
To support this aggressive growth, AMI is committing substantial capital—exceeding $20 billion—in addition to the $6 billion already allocated for its inaugural 100 MW phase. According to Anil Chalamalasetty, Chairman of AM Intelligence, the company’s facilities are being standardized to follow the design of their flagship Hyderabad AI Factory.
“Each site is engineered for high rack power density and is fully liquid-cooled,” Chalamalasetty stated. “This approach ensures that our infrastructure can accommodate successive generations of AI silicon within the same physical footprint, maximizing efficiency and scalability.”
Strategic Integration of Power and Compute
A defining feature of AMI’s expansion is its reliance on the NVIDIA Vera Rubin NVL72 rack-scale systems. These are complemented by high-throughput RDMA over Converged Ethernet (RoCE) networking to ensure low-latency data movement, which is critical for training large-scale AI models.
By leveraging the renewable energy expertise and storage capabilities of the Greenko promoters, AMI intends to optimize “electron-to-token” economics. This integration allows the company to provide sustainable, energy-efficient computing power to developers and enterprises, bridging the gap between infrastructure and application-layer AI development.
Vishal Dhupar, managing director, Asia South, at NVIDIA, noted the significance of the partnership: “By combining accelerated computing with high-performance networking and liquid-cooled infrastructure, AMI is creating the foundation for developers and enterprises to build more ambitious AI.”
A Vision for Gigawatt-Scale Infrastructure
The company’s long-term roadmap is even more ambitious. AMI aims to deploy 1 gigawatt (GW) of Compute-as-a-Service capacity as part of a wider 5 GW program spanning India, the United States, and Europe. With facilities under development through 2028, the company is positioning itself as a critical node in the global AI supply chain, providing the necessary hardware access for domestic developers and sovereign AI projects to compete on a global stage. As the demand for generative AI continues to surge, AMI’s focus on long-term asset development and energy-integrated infrastructure may prove to be a vital differentiator in the competitive landscape of AI factories.
Disclaimer: This content is auto-generated for informational purposes only.
Source: Read Original News
