DeepSeek's fast rise sparks debate on Indian AI models
Deepseek, a large language model (LLM) trained at a fraction of the cost of its Western counterparts, has sent shockwaves through the global AI community leading to calls for India to build its own AI language models.
Budget with ET
India, get ready for the modern warfare of tomorrow
Women, youth, farmers and poor can continue to be Budget 2025's ‘roti, kapda aur makan’
Modi govt has a key task in Budget 2025: Unlocking the PLI goldmine
Darshan Hiranandani, CEO of Hiranandani Group, which owns data centre giant Yotta, said India would need a major overhaul to get up to speed in the AI race. “Most countries have $6 million and 2000 GPUs of H100. We have it at Yotta. But we haven’t created an ecosystem that invents. It’s a mindset issue.”
However, the events surrounding DeepSeek’s launch will push India to create quick and efficient LLMs for Indian languages as well.
Deepseek demonstrates that high-quality LLMs can be built with significantly less investment than previously thought. This breakthrough has ignited a debate about India's AI strategy and the need for increased government support and a shift in mindset.
“The emergence of models like DeepSeek, and at the price that it has been built, are a reflection of the fact that compute power can never be a moat for too long between the countries that have early gains in just compute power compared to those that don't… hopefully Indian AI startups will cause disruptions down the road,” said Rajeev Chandrasekhar, former minister of state for electronics and IT.
Artif
