Indian artificial intelligence infrastructure company Nvidia’s Rubin graphics processing unit.
Hyderabad-based company AM Intelligence plans to raise up to $3 billion through a bond issue and bank loans to buy Nvidia Corp’s next-generation chips, Bloomberg reported on Friday.
The bond and loan split has not yet been finalized, but the company expects to receive $300 million upfront from a potential buyer, the report said.
AM Intelligence, which claims to be one of the “largest and most avid buyers” of Rubin GPUs outside of North America, did not respond to CNBC’s request for comment.
The company, owned by the founders of India’s leading renewable energy company Greenko, plans to invest $6 billion to install 100MW of computing capacity spread across centers in Malaysia and India, AM Intelligence said in an Oct. 5 release.
The company said it would purchase a total of 29,000 Nvidia Rubin chips for 100MW of capacity, adding that it plans to invest more than $20 billion to add 300MW of capacity to the market within the next 15 months.
AM Intelligence is part of a growing number of AI infrastructure companies in India offering computing as a service. Earlier this year, Mumbai-based Yotta Data Services announced plans to invest $2 billion to set up an artificial intelligence hub using Nvidia’s Blackwell Ultra GPUs.
Neil Shah, vice president at market intelligence firm Counterpoint Research, said many energy companies in India are becoming neo-cloud companies by setting up data centers with computing power that can be leased to hyperscalers, citing Adani Group, Reliance and Tata Power as key examples.
Nvidia’s Rubin GPUs are tailored for agent AI and will give AM Intelligence a first-mover advantage in India, Shah said.
By next year, companies will focus on “low-latency production agent execution” rather than just training raw underlying models supported by Blackwell Ultra GPUs.
AM Intelligence is developing 5 GW of AI data centers and 1 GW of Compute-as-a-Service in India, the US and Europe, a company spokesperson said in a note. The company plans to establish 400 MW of computing capacity within two years.
