New Delhi: Nvidia chief executive Jensen Huang said on Monday that the company’s latest chips are 10 times more energy efficient for running artificial intelligence (AI) applications. For India’s nascent data centre sector, this offers a choice: either select the latest, pricier graphics processing units to eventually save on power costs, or buy older, cheaper chips to set up and scale the computing capacity.
Analysts say companies will embrace both strategies as the country’s data centre build-out has just begun. The operating capacity is expected to soar from 1.5 gigawatts (GW) today to about 16GW by 2032 as the rising adoption of AI drives demand for computing power.
“There are two ways to look at Nvidia’s India revenue. While data centre expansions will increase the market size for Nvidia’s graphic processing unit (GPU) chips, India is price-sensitive in terms of consumption of GPUs, even at the cost of latency,” said Anushree Verma, senior director analyst at consulting firm, Gartner. “This may mean that India’s adoption of Nvidia chips may see strong supplies of less expensive older GPUs that become more affordable, rather than the cutting-edge ones—in about three years.”
Nemotron coalition
At Nvidia’s Global Technology Conference (GTC) 2026 keynote, Huang said that his company’s new computing chips, called Vera Rubin, would be 10 times more energy-efficient in running AI applications. A new general-purpose AI chip, Nvidia Groq 3, is claimed to deliver 35 times the performance per unit of energy.
The $4.5-trillion semiconductor maker also established a global coalition of companies, Nemotron. It includes the likes of Perplexity and Cursor to build large language models (LLMs) that will act as a reference point for enterprises globally. Indian Institute of Technology (IIT) Madras-incubated Sarvam will be part of the coalition to bring India-native foundational LLMs to the world, Huang said.
Still, India is a tiny market for Nvidia. Data sourced by Mint from Tofler showed the company generated ₹6,150 crore ($669 million) in annual revenue from India in FY25. That’s about 0.3% of the $215.9 billion in global revenue the company announced for the February 2025 to January 2026 fiscal.
Huang also skipped the big-splash AI Impact Summit in New Delhi at the last minute, with the company citing “unavoidable circumstances” for it.
“India is a strategically critical market for Nvidia, but its growth will be infrastructure-led, not price-led,” said Sanchir Vir Gogia, founder and chief analyst of tech consultancy firm Greyhound Research. “Demand will scale in line with data centre maturity, power availability, and enterprise monetization of AI. The opportunity is real, but the ramp will be gradual and structurally constrained.”
India opportunity
Nvidia’s technology has seen widespread adoption among Tata Sons and Larsen & Toubro Ltd. Its chips also account for over 85% of the common compute infrastructure that the ministry of electronics and IT (Meity)’s India AI Mission has built. All of this lends ground for Nvidia’s new chips to increase the company’s market share in the country.
The key would be the cost of Nvidia’s new chips, according to Kashyap Kompella, founder of tech analysis firm RPA2AI Research.
“Nvidia’s new chips always come in at a significantly higher price point over older ones. This means that the new chips alone may not make a huge splash for Nvidia when it comes to increasing its India revenue pie, and only a limited volume of cutting-edge facilities handling high-volume AI workloads will adopt them,” said Kompella. “The real growth for Nvidia will come in India as the local data centre market expands, and low-intensity AI tasks for general purposes adopt older-generation GPUs for mass AI applications.”
According to Kompella, Nvidia’s key expansion phase is likely to come through Sarvam’s inclusion in the Nemotron coalition and any future role that India’s IT services industry will play.
“We haven’t seen IT services firms featured in Nvidia’s keynote or announcements so far,” he said. “Nvidia’s physical AI play may increase the role that India currently plays for the chipmaker, which remains to be seen over the next two days.”
Yes, costs will not be the only factor shaping demand for Nvidia’s chips in India.
India “will become important, but not because affordability suddenly unlocks demand overnight”, said Gogia of Greyhound.
“Enterprises across sectors in India are actively exploring AI infrastructure options, but most are still in controlled experimentation phases. There is strong intent, but limited production-scale deployment outside of a few sectors,” said Gogia. “Enterprise buyers are increasingly shifting toward domestic data centres due to latency, compliance, and power economics, but they remain cautious on large capital commitments without clear returns on investments.”
Nvidia’s own roadmap places the availability of these next-generation systems in a future window, he said. “That naturally aligns with India’s infrastructure buildout timeline, which means that India will not lag, even if it does not leapfrog instantly.”
Gartner’s Verma said the newer chips will likely sell in smaller volumes even when they come in two to three years from now, and only for cutting-edge use cases. “This won't make a large volume impact, although some cutting-edge data centres run especially by the US-based Big Tech firms will likely license Nvidia’s latest chips.”