Ola's AI initiative, Krutrim, will design and develop India's first AI silicon chips by 2026, specially designed to handle complex AI tasks and workloads. This is part of the firm's efforts to strengthen its full-stack AI capabilities, which include foundational models, cloud and silicon.
"It's time for AI, it's time for the future of computing. India needs to build its own AI stack," says Ola founder Bhavish Aggarwal at the company's flagship event, Sankalp 2024, at the Ola FutureFactory in Krishnagiri, Tamil Nadu.
The initial family of chips includes Bodhi for AI, Surv for general computing, and Ojas for edge computing. The chips will enable the development of faster and more efficient AI systems and mark a milestone in India's achievement of technological supremacy, the company said in a statement.
The company plans to launch Bodhi 2 by 2028, which will be trained on 10 trillion parameters and scalable to Xscale supercomputing. Exascale systems are supercomputers that can process information much faster than today's most advanced supercomputers.
To support these advancements, the firm plans to scale its data center capacity to 1 GW by 2028. In addition, Am has partnered with global leaders such as Untether AI for the development of its CPU and AI chips.
Along with various purpose-built AI services aimed at helping the Indian developer community build technology applications more affordably, the company has also launched Krutrim Cloud with AI and general-purpose capabilities. Krutrim AI will be available on Ola Electric and Ola Cabs to improve user experience. Currently, more than 25,000 developers are using Krutrim Cloud and more than 250 billion API calls have been made across products since its launch in early 2024, the release said.
In addition, the company introduced more than 50 new services on Krutrim Cloud. These include Bhashik, a language hub with multimodal translation capabilities across text, voice and video, and Customer Experience AI, which introduces multimodal AI agents for enterprises that work across mobile apps.
