India may not have an NVIDIA, OpenAI or Anthropic yet, but that does not mean it is out of the global artificial intelligence race. In fact, Peak XV Managing Director Rajan Anandan believes the biggest winners in India's AI market will be homegrown companies.
Speaking to Rahul Kanwal on NDTV's Walk The Talk, Anandan argued that India has a genuine opportunity to emerge as a significant AI player, even if its path looks different from that of the US and China.
"AI in India will be won by Indian AI companies," Anandan said, adding that the country has made considerable progress in addressing some of the structural gaps that previously held back its AI ambitions.
ALSO READ: AI Bubble Was Bound To Burst: Govt Sources See Indian IT Gaining From Nvidia Meltdown
One of the biggest challenges has been computing infrastructure. India had fewer than 10,000 active graphics processing units (GPUs) at the end of 2025, according to Anandan. That limited the country's ability to train and deploy advanced AI models.
However, the landscape is changing. He said more than $300 billion in AI-related capital expenditure was announced in the first six months of 2026, involving major Indian conglomerates, global cloud companies and domestic data-centre operators.
Anandan expects India's active GPU capacity to cross 100,000 in the coming months, potentially marking a tenfold increase. He also sees an opportunity for India to become a global computing infrastructure provider, citing its relatively low energy costs and growing interest from states in attracting data-centre investments.
The second pillar of India's AI ambitions is homegrown foundation models. Anandan pointed to the launch of models by Sarvam AI in February, followed by several other Indian companies and BharatChain, developed at IIT Bombay.
Unlike American AI giants, which are investing heavily in increasingly large models in pursuit of artificial general intelligence (AGI), Indian companies are focusing on smaller, specialised models that can deliver practical results at lower costs.
According to Anandan, most enterprise applications do not require trillion-parameter models. A model with up to 100 billion parameters can often handle business requirements effectively while keeping deployment and operating costs under control.
ALSO READ: India Is Beating Wall Street. One Fintech CEO Says That Only Strengthens The Case For US Stocks
Voice AI is one area where Indian companies have an advantage, he said. Success in this segment depends on supporting multiple Indian languages, understanding conversations in noisy environments and offering affordable services.
The opportunity also extends to consumer and enterprise applications. Anandan said India has more consumer AI startups than the US, with businesses emerging across education, healthcare, travel, entertainment and other sectors.
He identified semiconductors as the biggest long-term challenge, noting that building capabilities comparable to NVIDIA would require a multi-decade effort.
Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.