India has initiated an ambitious program to promote the development of foundational AI models. The government has approved 18 proposals that will receive extensive support, including access to computing infrastructure, financial assistance, and critical datasets, delivering competitive AI within eight to ten months.
One of India’s AI push’s most significant aspects is its computing power investment. The government has exceeded its initial target of securing 10,000 graphical processing units (GPUs), amassing 18,600 GPUs to support AI training and research. Notably, this includes high-performance Nvidia H100 and H200 chips and MI325 models, reports India Today.
The government has identified ten companies to supply the GPUs, including Yotta (backed by the Hiranandani Group), Jio Platforms, Tata Communications, E2E Datacenter, Orient Technologies, and Vensysco Technologies. Notably, half of the GPUs, that is, around 9,000 units, will be purchased from Yotta alone, notes The Indian Express.
These GPUs are essential for developing a common computing facility to provide startups, researchers, and developers access to high-end AI infrastructure.
“Over 1.5 years, our teams have been closely with startups, researchers, professors, etc. Today, we are calling for proposals to develop our own foundational model. The model will take care of the Indian context, languages, culture, are devoid of biases,” Ashwini Vaishnaw, the Minister of Electronics and Information Technology, told reporters.
To accelerate innovation, the Indian government will fund 40% of the computing costs associated with the approved AI projects. This move aims to lower the financial barriers for developers and researchers, ensuring that AI innovation is not confined to deep-pocketed corporations.
India’s AI model is designed to focus on the country’s diverse linguistic and cultural landscape. The ability to process and understand multiple regional languages will be crucial, enabling the AI to cater to India’s vast population more effectively.
Ola launched Krutrim AI in public beta in February 2024. Bhavish Aggarwal, the founder of Ola, announced that the company will launch its first AI chip in 2026.
In the News: Novel botnet caught exploiting Mitel phone vulnerability for DDoS attacks
