New Delhi: Aravind Srinivas, CEO of Perplexity AI, has committed $1 million and five hours of weekly personal involvement to support India’s artificial intelligence (AI) goals. This announcement, made via a series of posts on X (formerly Twitter), reflects his dedication to advancing India’s capabilities in AI development.
Srinivas emphasized the importance of building a robust team capable of surpassing benchmarks set by China’s DeepSeek AI. He expressed willingness to invest an additional $10 million if the team achieves superior results to DeepSeek R1 on all benchmarks.
In his post, Srinivas criticized Infosys co-founder Nandan Nilekani’s recommendation to focus on AI infrastructure rather than training large language models (LLMs). Srinivas argued that India must balance both aspects to ensure competitiveness. “Ignoring model training skills in favor of building on existing models is a mistake. India needs to develop its own globally competitive AI models, not just those optimized for Indic languages,” he asserted.
Highlighting misconceptions about the costs of AI model training, Srinivas pointed to China’s recent breakthroughs as evidence that significant progress is possible without massive investments. Although he does not intend to lead a DeepSeek-like initiative himself, he pledged his support to anyone “obsessed enough” to take on this challenge and open-source the models under an MIT license.
This move marks a significant step toward enhancing India’s AI capabilities and positions Srinivas as a key advocate for the country’s technological growth.
