IT minister Ashwini Vaishnaw reinforced the urgency of this effort, stating, "Very soon, we will have our own LLMs," while confirming that the IndiaAI Compute Facility has secured 18,000+ GPUs to support development. This comes on the heels of DeepSeek, the powerful Chinese AI model built for less than $6 million. DeepSeek's stunning debut has caused a stir in India's AI ecosystem. "The models are a step change and will enable faster progress towards even more powerful and intelligent systems," said Sarvam AI founder Pratyush Kumar, adding that India must "absolutely build its own sovereign models from scratch." He emphasised that these models must balance deep reasoning with high proficiency in Indian languages to be truly effective. The government's expectations are clear. These models must be trained on diverse Indian datasets, comply with Indian regulations, and serve both public and strategic interests. Proposals will be evaluated monthly, with funding provided through direct grants, equity-based investments, and co-financing from industry stakeholders. Enjoy the full story here. |