Indian Firm AM Intelligence Places Order For 9,000 Nvidia Vera Rubin System Chips For Hyderabad AI Factory

AM Intelligence, an Indian AI infrastructure company, has placed a binding order for roughly 9,000 Nvidia Vera Rubin systems, positioning itself as one of the first companies in Asia to adopt Nvidia's newest computing platform, Bloomberg reported.

The GPUs will be deployed as Vera Rubin NVL72 rack-scale systems at AM Intelligence's first AI data centre in Hyderabad, with delivery scheduled for the first quarter of 2027. According to Bloomberg, the servers are expected to come online next year in southern India.

AM Intelligence was founded by the promoters of Greenko Group, the renewable energy producer. The Hyderabad order forms the opening phase of the company's broader plan to build 1 gigawatt of Compute-as-a-Service capacity spread across India, the United States, Finland and Malaysia, backed by more than $8 billion in capital expenditure. AM Intelligence intends to bring an initial 200 megawatts of AI compute capacity to market in the near term.

Bloomberg reported that the Hyderabad facility is being built to handle some of the most demanding AI workloads in use today, including trillion-parameter models and next-generation agentic AI systems. The company expects the site to serve large cloud-service providers, AI research labs and organisations building homegrown Indian AI models.

Customer already signed on

Per Bloomberg's report, the initial compute capacity at the Hyderabad facility has already been bought by a US customer, though Greenko Group founder and president Mahesh Kolli declined to name the buyer, citing a non-disclosure agreement. Kolli told Bloomberg that customers globally are 'hunting desperately' for AI compute capacity, and said India's undersea cable infrastructure lets US firms access compute there with latency of around 300 milliseconds, removing what he described as a major operational barrier.

Energy cost as a differentiator

AM Intelligence is leaning on Greenko's renewable-energy assets to keep its computing costs competitive, an important factor given how much energy pricing affects the economics of running large AI models at scale. Kolli was quoted by Bloomberg as saying that energy prices are a major factor in global token economics, and the company has positioned itself as among the lowest-cost AI computing infrastructure providers worldwide.

Nvidia supply remains tight

The report noted that global supply of high-end Nvidia compute clusters remains constrained, with demand from data centre operators and companies including Microsoft, Google and Amazon driving multiyear order backlogs. Japan, separately, is also planning to acquire Nvidia's next-generation Rubin chips to build a domestic foundational AI model for robotics, with its first such data centre expected to begin operating in June 2028, Bloomberg reported.

AM Intelligence's ambitions extend well beyond the Hyderabad site. The company is targeting 5 gigawatts of AI data-centre capacity by 2030 across India, Europe and other markets, to be funded through a combination of debt and equity.