- Aravind Srinivas urges India to build its own AI models like DeepSeek.
- Western labs ignore Indian voices, a gap India can fill with smart tech.
- Startups should raise money, use open-source AI, and grow step-by-step.
Aravind Srinivas, CEO of Perplexity AI, told Nikhil Kamath on the WTF podcast that India needs to step up and create its own AI models instead of depending on foreign ones. He thinks India has a big chance to shine in artificial intelligence by making homegrown systems that compete worldwide. Srinivas wants an Indian company like DeepSeek to lead the way, building AI that’s great at thinking and solving problems—not just for Indian languages but for global standards too. He says this could push young engineers to dream big and shape the future.
Srinivas sees a huge opportunity for India in fixing a problem Western labs don’t care about: AI that understands Indian voices. Right now, most AI struggles with Indian accents, dialects, and grammar, especially for speech recognition and voice-making tech. He believes if India builds real-time AI that nails all Indian languages and sounds natural, it could change everything. This focus on India’s unique mix of languages could make Indian startups stand out and lead in the AI world.
For Indian startups wanting to jump in, Srinivas has a simple plan: start by raising cash, get users, then raise more. Begin with tweaking free AI models, move to training new ones, and later build data centers. He knows it’s a long road, but India has to start somewhere to catch up and compete. Collecting voice data might be tricky, but Srinivas says it’s worth it—India’s diversity is its strength, and turning that into smart AI could put the country on the global map.
