The country’s expanding artificial intelligence (AI) ecosystem will require computing power of more than 100,000 GPUs in the coming years as AI adoption accelerates across key sectors such as manufacturing, healthcare, agriculture, and ESG compliance.
The growing demand for computing power shows the nation’s push towards building domestic capacity and reducing dependence on foreign infrastructure, said Abhishek Singh, CEO of IndiaAI Mission and additional secretary in the ministry of electronics and information technology (MeitY).
“Today, we have around 38,000 GPUs available at a cost of about Rs 65 per GPU per hour, which is less than a dollar compared with $2.5-3 elsewhere,” Singh said at the TiEcon Delhi-NCR event. “But as we start using AI at scale, the requirement will go beyond 100,000 GPUs,” he added.
He said that building this capacity will require significant public and private investment, as one block of 1,000 GPUs, including data centre and power infrastructure, costs between Rs 500 crore and Rs 800 crore. “We do have visibility on achieving that scale, and both government and private players are working towards it,” Singh added.
Looking ahead, the country’s vision for the next three to five years is to make India a voice-enabled digital economy, where technology becomes accessible to every citizen regardless of literacy or language. “We have about 900 million people in the digital economy and around 500 million outside it. When these 500 million learn to use AI and voice-based tools in their local languages, it will significantly improve productivity, incomes, and national output,” he said.
However, as code generation tools like GitHub Copilot or Cursor AI become mainstream, experts suggest that India’s IT services industry must evolve beyond low-value coding. “Around six million people work in IT, and many do repetitive coding. Indian firms like TCS and Infosys have massive code repositories and they should develop their own code generators to stay ahead,” Singh said, adding that the government is in talks with NASSCOM and industry partners to explore building a code generator.
The IndiaAI Mission, approved by the Union Cabinet in March 2024 with an allocation of over Rs 10,300 crore, is designed to establish a distributed AI compute network comprising more than 10,000 GPUs. The infrastructure is being made available to startups, research institutions, and government bodies to accelerate innovation and AI-driven development across sectors.