Please Share 🤝

Highlights

AI Chip Launch: Krutrim to introduce its first AI chip, Bodhi 1, by 2026, followed by Bodhi 2 in 2028.
Cloud Expansion: Krutrim Cloud scales up with new services, benefiting 25,000+ developers and SMEs.
Data Centre Growth: Krutrim plans to expand data center capacity to 1 GW by 2028 for its AI initiatives.

Krutrim’s AI Chip Vision: Bodhi 1 and Beyond

Krutrim, the artificial intelligence startup founded by Ola’s Bhavish Aggarwal, announced plans to launch its first AI chip, Bodhi 1, by 2026, followed by an advanced version, Bodhi 2, in 2028. These chips are custom-designed to handle complex AI workloads, providing faster and more efficient AI systems. The Bengaluru-based company also revealed its roadmap for four AI chips: Bodhi 1, Bodhi 2, Sarv 1, and Ojas. The chips will target different AI applications, including cloud-native computing and Edge AI for next-gen Ola Electric vehicles.

Key Partnerships for AI Development

Krutrim has partnered with global semiconductor leaders Arm and Untether AI to develop its CPU and AI chips, platforms, and systems. Bodhi 1, which is set to debut in 2026, will be optimized for large language models (LLMs), AI inference, and fine-tuning, with industry-leading power efficiency. Bodhi 2, slated for 2028, will support AI models with over 10 trillion parameters. Sarv 1, Krutrim’s cloud-native CPU, and Ojas, an Edge AI chip, will also launch in 2026, the latter powering Ola Electric vehicles.

Google Pixel 9 Series Launched in India: Price & Specs Overview

Google Pixel 9 Launched in India: Price & Specs Revealed

August 14, 2024 No Comments

Google Pixel 9 series launched in India with advanced specs, starting at Rs 74,999. Pre-order…

CrowdStrike-Microsoft Outage: Largest IT Crash in History?

July 20, 2024 No Comments

Discover the causes and impact of the Microsoft-CrowdStrike IT outage, affecting multiple sectors globally. It…

1 2 3 4 5 6 Next »

Manufacturing Plans Still in Discussion

Krutrim is still finalizing its manufacturing plans for the AI chips, with discussions ongoing with potential partners, including Taiwanese foundries. This announcement follows the successful IPO of Ola Electric, positioning Krutrim as a player to watch in the AI and semiconductor sectors.

Industry Response and Challenges Ahead

Technology analysts like Neil Shah from Counterpoint Research view Krutrim’s roadmap as appealing and well-timed, especially with rising investor interest in AI and indigenous silicon development. However, Shah also highlighted the challenges Krutrim will face in transitioning from design to execution, particularly in building a competitive AI software stack to rival industry giants like AWS and Azure. He emphasized that Krutrim’s business model will need to account for significant capital and operational expenditure to stay viable in the competitive AI space.

Expansion of Krutrim Cloud Services

Krutrim is expanding its cloud services to meet the growing demands of AI and traditional workloads. Krutrim Cloud now offers no-code/low-code platforms tailored to Indian use cases, with services such as GPU-as-a-service and model-as-a-service. More than 25,000 developers are actively using Krutrim Cloud, with over 250 billion API calls made across products since its launch in early 2024. Additionally, over 50 new services have been introduced, including virtual machines, cloud storage, and security features. AI pods for affordable GPU access, AI studios for advanced applications, and a model catalogue with LLMs are also part of the offerings.

Supporting Startups and Developers

Krutrim is incentivizing developers with free cloud services worth Rs 100 crore until Diwali and free access to cloud services for users of ONDC for one year. The company launched the ‘Krutrim for Startups’ Udaan program to further support startups, offering Rs 50 lakh to the first cohort of startups in collaboration with investors like Z47, DeVC, and South Park Commons.

Scaling Data Centre Capacity

To support its ambitious AI and cloud projects, Krutrim plans to scale its data center capacity to 1 GW by 2028, aiming for full-stack AI capabilities, including foundational models, cloud, and silicon.