AI Development: Ola Partners with Lenovo for Krutrim 3

Ola’s artificial intelligence division, Krutrim, declared a strategic alliance with Lenovo, a leading global technology company, on Wednesday. Together, they aim to create Krutrim 3, an impressive AI model that incorporates 350 billion parameters. This announcement comes after Krutrim’s previous initiatives with AI models Krutrim 1 and 2, which launched in 2023 and included parameters ranging from 7 billion to 12 billion. Both earlier models were made open-source to encourage greater innovation in AI.

Navendu Agarwal, the Chief Information Officer of Ola, expressed excitement about the collaboration at the Lenovo Tech World India 2025 event in Mumbai, stating that the two companies are on a mission to create the best AI model that represents India’s capabilities.

Agarwal also emphasised the importance of this project, noting that India currently lacks a sovereign cloud and large-scale hyperscale infrastructure. The development of Krutrim 3 is deemed essential for enhancing the nation’s technological independence.

In a recent post on X, Bhavish Aggarwal, the Founder and CEO of Ola, shared his vision for India’s AI future, stating the necessity for a supercomputer in Krutrim, in collaboration with Lenovo. He encouraged the public to propose names for this innovative supercomputer.

This announcement is part of Ola’s proactive strategy to advance AI technology infrastructure. This effort was showcased in earlier partnerships with corporations like Nvidia, reflecting their dedication to deploying GB200 superchips. Krutrim had also previously integrated advanced AI models from DeepSeek on local servers, significantly increasing data privacy and cutting down training expenses.

The urgent need for a robust sovereign cloud infrastructure within India has become increasingly apparent, especially amid rising global competition, particularly from Chinese and American technology firms. Recently, Tata Group Chairman Natarajan Chandrasekaran also stressed the necessity for indigenous hyperscale computing to preserve the country’s technological sovereignty.