India’s artificial intelligence (AI) landscape is at a crossroads. While the government is ramping up efforts to develop large language models (LLMs) under the IndiaAI Mission, a debate is unfolding over the best way forward. Aravind Srinivas, CEO and co-founder of Perplexity AI, believes that India must not only develop AI applications but also focus on training foundational AI models. His stance directly opposes that of Infosys co-founder Nandan Nilekani, who argues that Indian startups should focus on building applications rather than investing billions in model training.
Credits: The Economic Times
Srinivas’ argument is clear: “India needs the capability to train models independently, which requires both infrastructure and talent.” He believes that without this, the country risks becoming overly reliant on tech giants in Silicon Valley.
The Indian government is taking significant steps toward AI development. Under the IndiaAI Mission, six companies are developing indigenous AI models tailored to Indian languages and culture. To support this initiative, the government has procured 18,693 high-performance graphics processing units (GPUs) that are crucial for AI model training. Of these, 10,000 GPUs are ready for immediate use, while the rest will be gradually deployed.
With a Rs 10,300 crore budget for the IndiaAI Mission, the government is signaling its intent to compete in the global AI race. The focus is not just on creating AI tools but ensuring they are designed for Indian needs, reducing bias, and increasing accessibility for diverse linguistic groups.
The debate over India’s AI strategy intensified when Srinivas publicly disagreed with Nandan Nilekani on X (formerly Twitter). Nilekani had suggested that India should focus on building practical AI applications rather than developing foundational models, arguing that it’s better to leverage pre-existing LLMs from global tech companies.
Srinivas, however, sees this as a limiting approach. “He’s wrong on pushing Indians to ignore model training skills and just focus on building on top of existing models. Essential to do both,” he posted on January 21.
By advocating for independent AI infrastructure, Srinivas is calling for an approach that mirrors India’s success in low-cost space exploration—something that has inspired generations of engineers. He believes AI can follow a similar trajectory if India invests in model training capabilities today.
Srinivas recently met Prime Minister Narendra Modi, who, he said, displayed a deep understanding of AI’s implications. “He was well aware of different AI models, their strengths, and weaknesses,” Srinivas noted.
PM Modi also emphasized the role of AI in preserving India’s ancient scientific knowledge, including Vedic mathematics and literature. This aligns with his broader vision of integrating technology with India’s cultural and historical strengths. Modi stressed that AI should not merely enhance productivity but also spark curiosity and innovation across all Indian languages.
As part of its India strategy, Perplexity AI has launched an ambitious expansion plan. The startup is offering free access to its pro version for students, faculty, and staff at IIT-Madras. “Our goal is to reach a million sign-ups by mid-March,” Srinivas stated, adding that Perplexity wants to make AI-driven knowledge accessible to all students in India.
Beyond education, Perplexity is also entering India’s fintech space through a partnership with Paytm. The startup’s AI-powered search engine will be integrated into the Paytm app, allowing users to access instant, well-researched answers. Srinivas also hinted at future integrations that could enable AI-driven shopping experiences using Paytm’s payment infrastructure.
Perplexity is actively exploring ways to make its AI services more affordable in India. Currently priced at $20 per month, the company is working on integrating local payment solutions, including UPI. “We work with Stripe API right now, but we hope to figure out a way to incorporate Paytm UPI soon,” Srinivas shared.
This focus on affordability aligns with India’s broader push for AI democratization—ensuring that advanced AI tools are not limited to elite institutions or businesses but accessible to everyday users and students.
Credits: Money Control
As AI becomes a cornerstone of technological progress, India must decide whether to be a consumer or a creator. While Nilekani’s pragmatic approach has merit, Srinivas’ vision of a self-sufficient AI ecosystem presents a compelling argument for long-term independence and innovation.
With government support, increasing GPU capacity, and startups like Perplexity pushing for deeper AI integration, India is well-positioned to make a global impact. The question remains: Will India take the bold step of investing in AI model training, or will it settle for building applications on top of existing frameworks? The coming years will determine whether India emerges as a true AI powerhouse or remains dependent on external tech giants.
