• Capex and return targets: Out of ₹720 crore raised from the fresh issue, ESDS plans to deploy ₹576 crore into infrastructure—allocating 70% directly to AI infrastructure and GPUs—while targeting an 85% EBITDA margin, 45% ROCE, 40% PAT margin, and 1 asset turn.
• GPU asset dynamics: Management views GPUs as appreciating assets rather than depreciating hardware, noting that older models like A100 and H100 command double or 2.5 times higher rental rates due to tight semiconductor supply expected over the next 5 to 7 years.
• GPU monetization model: Capacity is sold as dedicated private clusters or AI factories billed at fixed rates per GPU per hour (around $5 to $7) rather than on a pay-per-use or token-consumption model.
• AI revenue outlook: AI-related revenue, which historically remained in the single digits, is projected to expand significantly as the company launches a new AI factory next month and taps into a large project funnel.