India's burgeoning artificial intelligence (AI) landscape requires a significant boost in infrastructure to support the training of increasingly sophisticated AI models, according to a top executive at Krutrim, an Indian AI venture. This call to action highlights a critical need for investment and development in computational resources, data centers, and skilled personnel to propel India to the forefront of AI innovation.
Krutrim, founded by Bhavish Aggarwal, the co-founder of Ola, aims to create AI models specifically tailored for the Indian ecosystem. The company recognizes that "India-first AI" must understand the country's unique cultural context and be trained on datasets specific to its diverse needs. Krutrim has already launched its base model, trained on 2 trillion tokens, and plans to release a more advanced model, Krutrim Pro, with enhanced problem-solving capabilities.
However, the development and training of such large language models (LLMs) demand substantial computational power and robust infrastructure. The current AI models are predominantly trained in English, which may not fully capture the nuances of Indian languages, cultural values, and context. To address this, India needs to build its own AI infrastructure, including indigenous data centers, server-computing capabilities, edge-computing resources, and supercomputers. Krutrim plans to launch a supercomputer within the next two years and develop AI-focused hardware, including chiplets, for data centers.
The lack of adequate infrastructure can hinder the progress of AI research and development in India. Insufficient computing power can slow down training times, limit the size and complexity of models, and make it difficult for Indian researchers and developers to compete with their global counterparts. A robust infrastructure is also crucial for ensuring data privacy and security, as it allows Indian companies to store and process data locally, rather than relying on foreign providers.
Recognizing this need, Krutrim is not only focused on developing AI models but also on building the necessary infrastructure to support them. This includes creating AI infrastructure to develop indigenous data centers and eventually, server-computing, edge-computing, and super-computers. The startup is also working on silicon chips, integrating the models, infrastructure, and silicon tightly for a well-run AI computing business.
Krutrim has also launched Krutrim AI Lab, India's first AI frontier research lab, to make AI research accessible, attract talent, and establish India as a global leader in open-source AI. The lab will focus on critical AI research domains, including multilingual AI, multimodal AI, embodied AI, data-centric AI, and personalized AI. Krutrim AI Lab champions open-source AI and democratizes AI innovation by fostering collaboration with academia, startups, and developers and deploys state-of-the-art computing resources to accelerate AI capabilities.
The Indian government also recognizes the importance of AI and is taking steps to promote its development. Initiatives such as the National AI Strategy aim to create a conducive ecosystem for AI research, development, and adoption. However, more needs to be done to address the infrastructure gap and create a level playing field for Indian AI companies.
To fully realize its AI potential, India needs a concerted effort from both the public and private sectors to invest in and develop the necessary infrastructure. This includes building more data centers, increasing computing power, fostering a skilled workforce, and promoting open-source AI initiatives. By addressing these challenges, India can unlock the transformative power of AI and become a global leader in this critical field.
