AM Intelligence, an Indian AI infrastructure company, is seeking to raise as much as $3 billion in debt to pay for a previously announced order of Nvidia's next-generation Vera Rubin chips. The financing could be made up of bonds and bank loans, among other instruments, and the mix is still being finalized.
On top of the debt, AM Intelligence expects $300 million in advance payments from potential offtakers, the customers who would buy the compute capacity once it is running. Talks are in the initial stages and terms could change. No arranging bank or offtaker has been named.
AM Intelligence was set up by the founders of Greenko Group, the Indian renewable energy producer, and is based in Hyderabad. The company sells AI compute as a service, with GPU capacity it plans to bring online across several countries.
Key facts
- 01AM Intelligence is seeking as much as $3 billion to fund a previously announced purchase of Nvidia Vera Rubin chips.
- 02The package could include bonds and bank loans; the firm also expects $300 million in advance payments from offtakers.
- 03In August the company ordered 9,000 Rubin GPUs for a 30 MW Hyderabad AI factory under an $8 billion project.
- 04On October 6 it added two orders for 20,000 more Rubin GPUs, bringing commitments to about 29,000 GPUs and close to 100 MW.
- 05Talks are in early stages and terms could change; AM Intelligence declined to comment and Nvidia did not respond.
In August, AM Intelligence said it had placed a firm and binding order for 9,000 Nvidia Rubin GPUs, deployed as Vera Rubin NVL72 rack-scale systems, for 30 MW of capacity at its first AI factory in Hyderabad. That order sits under an $8 billion project, and the company has said it wants to be one of the first adopters of the Vera Rubin platform in Asia. The Hyderabad facility is designed to run trillion-parameter models and agentic AI applications, with delivery slated for the first quarter of 2027.
The order book has grown since. On October 6, AM Intelligence announced two more firm orders for 20,000 Rubin GPUs, bringing its total commitments to about 29,000 GPUs and close to 100 MW. That follow-on capacity, about 70 MW in Malaysia and India, is scheduled for delivery in the second quarter of 2027. The company says its sites use liquid cooling throughout.
“Each site is engineered for high rack power density and liquid cooled throughout”— Anil Chalamalasetty, Chairman, AM Intelligence
The longer-range plan is larger still. AM Intelligence has outlined about 400 MW of Compute-as-a-Service capacity across India, the United States, Europe and Malaysia, with the balance due in 2027 and early 2028, and a long-term target of 1 GW inside a broader 5 GW program of powered AI data centers. It has also said it plans more than $20 billion in capital spending for an additional 300 MW within 15 months.
The $3 billion raise fits a wider pattern of AI buildouts financed with borrowed money: the AI-linked borrowing boom has brought more than $650 billion of debt to market in less than two years. In India specifically, a KPMG report released in July estimated about $90 billion in investment opportunity across the country's data center value chain by fiscal year 2035.
A representative for AM Intelligence declined to comment on the financing talks, and Nvidia did not respond to requests for comment.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
