India launched a program to build native large language models for its many languages, funding startups and pooling compute under the $1.2 billion IndiaAI Mission. The Ministry of Electronics and Information Technology reserved 19,000 GPUs (including 13,000 Nvidia H100s). So far, it's backed Sarvam AI’s 70 billion-parameter multilingual model with voice and reasoning, plus projects from Soket AI Labs, Gan AI and Gnani AI. In The Batch, we break down the projects, model sizes, and implications for voice-first apps. Read the full article: hubs.la/Q03D21jn0

@DeepLearningAI Impressive investment. Will this boost local tech and reduce reliance on global models? A crucial step for digital inclusion.

@DeepLearningAI Great initiative, better late than never.

@DeepLearningAI It's a monumental catch-up game, but a crucial one. Building foundational models for our vast linguistic diversity is essential for tech sovereignty, even with the US & China's massive lead.

@DeepLearningAI need more ppl sharing about this! thank you.

@DeepLearningAI

@DeepLearningAI great and good luck to the project

@DeepLearningAI one of the greatest initiatives

@DeepLearningAI Excited for #India AI mission

@DeepLearningAI These native language models play a large part in taking AI to the core of India. Mistral released Saba to expand their offering in Arabic and South Indian languages. And they seem particularly targeting the regions with a high paying large population. India needs to up its game

@DeepLearningAI This is actually cool

@DeepLearningAI This is nice

@DeepLearningAI Really concerned about the budget allocated for just building a llm

@DeepLearningAI Thank you for covering this mission. This initiative by the Indian government will change the AI landscape of the country and grants the fundamental knowledge of LLMs to Indians.

@DeepLearningAI 🚀

@DeepLearningAI Lessgooo 🚀

@DeepLearningAI interesting initiative. funding for language models is key. how are they training?

@DeepLearningAI lol. half of the money will go into pockets of corrupt people. even in those purchases of hardware there will be commissions and servers labelled with H100s will have something else inside it.

@DeepLearningAI Great initiative! Excited to see the impact!!

@DeepLearningAI Gonna be hit 5M today my god “q1jGxxS5R87USdRipsd9EkemzE5gLeiowvwPL5Cpump”

@DeepLearningAI This is exciting news! India’s investment in native-language LLMs and supporting startups could really accelerate AI adoption and innovation for diverse languages. Looking forward to seeing how these models impact voice-first applications.

@DeepLearningAI India's $1.2B AI push gets it right: · Native LLMs > translated ones · 13K H100s = compute sovereignty · Voice-first for 300M+ non-typing users Make it work on $50 phones though. That's the real test.

@DeepLearningAI Most of the top academic research labs lack access to these GPU clusters. You can't build a foundation without good students and scholars.

@DeepLearningAI @grok explain

@DeepLearningAI 19,000 GPUs vs 19,500 dialects—tech PR teams must be loving this. I’ll believe BharatGPT when it speaks every street tongue. Numbers are easy, delivery’s the kicker.

@DeepLearningAI Is there any ai training work available

@DeepLearningAI It is good.

@DeepLearningAI ‘AI for grassroot governance’ would be the perfect starting point.. 🚀

@DeepLearningAI 👍👍👍

@DeepLearningAI Huge step forward, definitely. Building native models for India’s languages will open up so many opportunities locally.

@DeepLearningAI finally, some focus on local languages. this is key for real impact and inclusion. looking forward to what these models can do