AI News · Hot news ·
AM Intelligence orders 20,000 Nvidia GPUs in $4 billion infrastructure deal
AM Intelligence, backed by Greenko founders Anil Chalamalasetty and Mahesh Kolli, has finalized a $4 billion order for 20,000 Nvidia Vera Rubin GPUs. The project targets 70 megawatts of AI compute capacity in India by mid-2027. It also focuses on integrating renewable energy into AI infrastructure.
Key points
- AM Intelligence finalized a $4 billion order for 20,000 Nvidia Vera Rubin NVL72 GPUs, Whalesbook reported.
- The venture targets 70 megawatts by mid-2027; reports differ on whether that includes Malaysia alongside India.
- AMI plans to integrate renewable power and storage from the AM Green and Greenko infrastructure platform.
- Pricing, customer commitments and financing arrangements were not reported.
- The announced capacity is a future buildout, not confirmed compute already available to rent.
What happened: AM Intelligence, an AI infrastructure venture backed by Greenko founders Anil Chalamalasetty and Mahesh Kolli, has finalized a binding $4 billion order for 20,000 Nvidia Vera Rubin NVL72 graphics processing units, Whalesbook reported. The venture is targeting 70 megawatts of AI compute capacity by mid-2027. For businesses evaluating AI infrastructure, the announcement represents planned capacity rather than computing resources already available to rent. Whalesbook reported that the order follows an August 2026 commitment to procure 9,000 Nvidia systems.
The details: The reports differ on the geographic scope of the initial deployment. TechShots described 70 megawatts of new capacity in India, while Whalesbook said the capacity would span India and Malaysia. The allocation between the two countries was not reported. That distinction matters for buyers looking specifically for Indian infrastructure: the headline capacity should not automatically be treated as capacity that will all be located in India. The mid-2027 target is also a future deployment milestone, not confirmation that the facilities have been commissioned.
Background: AMI’s strategy links AI computing with renewable energy. Whalesbook reported that the venture plans to use renewable power and storage from the AM Green and Greenko infrastructure platform to run its compute clusters. Its approach, described as “electrons-for-tokens,” aims to lower energy costs and support competitively priced Compute-as-a-Service offerings. This is the service business through which AMI intends to offer computing capacity to hyperscalers, local AI labs and developers. Pricing, customer commitments and commercial availability dates were not reported.
Who it affects: The project adds to planned AI infrastructure expansion in a market where Reliance Industries, Adani Group and Tata Group are also investing in digital infrastructure, according to Whalesbook. AMI’s longer-term goals are 1 gigawatt of total compute capacity and 5 gigawatts of powered data centers by 2030. Those are company ambitions rather than completed infrastructure. For prospective customers, the renewable-energy strategy explains how AMI hopes to compete on operating costs, but the reports do not establish what customers will pay.
What to watch: Execution and financing remain central questions. Whalesbook highlighted the substantial upfront capital requirements, the risk of rapid hardware depreciation and the burden of servicing debt associated with such infrastructure. The financing arrangements for the order were not reported. Business buyers should watch for commissioning updates, confirmation of deployment locations and actual service availability before treating the announced buildout as a procurement option. AM Intelligence is privately held, with no direct publicly traded stock in the venture.
Our take
The order signals a substantial expansion of India's planned AI capacity. Buyers should distinguish the announced buildout from compute that is already available to rent.