Bhavish Aggarwal pumps ₹2,000 crore into Krutrim, promises ₹10,000 cr investment by 2026

Bhavish Aggarwal on Tuesday announced an investment of ₹2,000 crore in Krutrim, with a "commitment" to pump an additional ₹10,000 crore into his artificial intelligence (AI) venture by next year.

Aggarwal took to X to announce the launch Krutrim AI lab, stating that his team has been working on AI for a year now.

The Ola founder notes that while Krutrim 1, launched on January 24, was a basic model, an improved version is launched today named Krutrim 2, along with more models Chitrarth 1- a vision language model claiming to understand images and documents, Dhwani 1- a speech language model claiming to understand speech translations and Vyakhyarth 1- an Indic embedding model for use cases like Search and RAG (Retrieval augmented generation).

While announcing the company's partnership with Nvidia, Aggarwal reveals, “Also announcing India’s first GB200 deployment in partnership with NVIDIA! Will be live by March and we will make it the largest supercomputer in India by end of year.”

Accepting the negligible standing in terms of global benchmarks, but made good progress in one year, he mentions, “We’re nowhere close to global benchmarks yet but have made good progress in 1 year. And by open sourcing our models, we hope the entire Indian AI community collaborates to create a world class Indian AI ecosystem. We’re still learning to walk before we can run, hopefully within this year!”

Globally, China is ramping its AI efforts with consecutive groundbreaking developments. Recently, DeepSeek caused disruption in the US  AI market, Alibaba has unveiled its AI model ‘Qwen 2.5 Max,’ claiming superior performance than DeepSeek’s R1 model as well as OpenAI’s GPT-4o and Meta’s Llama.

Last week of January kicked off with a dramatic plunge in Nvidia stocks, dropping more than 17% during early trading hours on Monday, as the DeepSeek wave sent shockwaves across the tech world. DeepSeek, the chinese startup reached a major milestone by developing a competitive large language model using the less powerful Nvidia H800 GPUs—chips specifically designed to bypass US export restriction.

DeepSeeks’s ability to deliver superior results, placing it at the forefront alongside the likes of OpenAI’s ChatGPT and Meta’s Llama, has sparked questions about the real need for Nvidia’s high-end hardware.