India's AM Intelligence is exploring a financing package of up to $3 billion to support its purchase of Nvidia's next-generation AI chips, according to Bloomberg, as the cost of building large computing facilities pushes infrastructure developers toward increasingly complex financing arrangements.
The company, established by the founders of renewable energy producer Greenko Group, is considering a combination of bank loans, bonds and other financing instruments.
The discussions remain preliminary. AM Intelligence has not announced a completed transaction, and the eventual amount and structure could change.
The company also expects approximately $300 million in advance payments from prospective customers, according to people familiar with the matter who spoke to Bloomberg.
The financing would support an ambitious expansion of computing capacity built around Nvidia's Vera Rubin architecture, the chipmaker's next-generation platform for training and running large AI models.
In August, AM Intelligence announced an initial order for approximately 9,000 Rubin GPUs for a facility in Hyderabad.
It followed that commitment in October with binding orders for another 20,000 GPUs, destined for facilities in India and Malaysia.
Taken together, those commitments represent approximately 29,000 GPUs and nearly 100 megawatts of planned computing capacity.
The company is pursuing a wider infrastructure strategy that combines renewable energy, data centers and high-performance computing.
Its promoters, Anil Chalamalasetty and Mahesh Kolli, previously built Greenko into a major renewable energy producer, developing projects involving wind, solar and energy storage.
That experience could prove relevant as access to electricity becomes one of the primary constraints on AI infrastructure development.
Large AI computing facilities require substantial power supplies, sophisticated cooling systems and expensive networking equipment. Securing the chips is only one part of the investment.
AM Intelligence has outlined plans for an initial 200 megawatts of computing capacity involving more than $8 billion in capital expenditure, within a broader ambition to develop a gigawatt-scale network.
The proposed $3 billion financing is separate from reports earlier this month that the company was also exploring an equity raise of more than $1 billion, with Morgan Stanley acting as an adviser.
Neither transaction has been announced as completed.
The scale of the plans reflects India's effort to establish more domestic AI computing capacity, reducing dependence on infrastructure located in the United States and other overseas markets.
But the economics remain demanding.
Companies building GPU clusters must commit substantial capital before the equipment begins generating revenue. They must also secure long-term customers willing to pay enough for computing services to cover financing costs and the rapid depreciation of advanced chips.
AM Intelligence's strategy is to combine large-scale power infrastructure with demand for computing capacity from cloud providers and AI developers.
The reported financing discussions suggest that executing that strategy will require significant external capital.
For now, the $3 billion figure represents a potential financing package, not money the company has already raised.
Sources: Bloomberg report via The Edge · Moneycontrol · AFP on the GPU order