India’s Push for Indigenous AI Development
In recent years, artificial intelligence (AI) has emerged as a critical area of technological advancement. Countries worldwide are investing heavily in AI to gain a competitive edge. India is no exception. Many startups in the country are now focusing on developing indigenous AI models, inspired by the success of Chinese AI startup DeepSeek. This article explores the current landscape of AI development in India, the challenges faced, and the potential for future growth.
The Rise of Indigenous AI Models
Indian startups are increasingly prioritizing the development of their own AI models. Companies like Fractal Analytics are leading the charge, with plans to create advanced models for artificial general intelligence (AGI). Srikanth Velamakanni, the founder and CEO of Fractal Analytics, emphasized the importance of collaboration among major players in the industry. He suggested that if ten companies, including TCS and Infosys, each contribute $20 million, the government could match this funding. This could create a substantial fund of $600-$800 million, which would be pivotal in building a frontier AI model.
Fractal Analytics has already developed four small language models (SLMs) named Vaidya, Kalaido, Marshall, and Ramanujan. The company plans to open-source Ramanujan, a complex reasoning model designed for high-level tasks such as advanced mathematics and chess. This move could attract global talent, particularly since many AI researchers are of Indian origin. The initiative aims to establish a robust foundation for AI development in India, ensuring that the country does not lag behind in this crucial technological race.
The Role of Government and Private Sector
While private companies are stepping up to support AI development, experts agree that sustained government investment is essential. Funding, research, and education are critical components in building large language models (LLMs) from scratch. Industry insiders caution that future iterations of models like DeepSeek’s may not remain open-source, especially if developed by entities outside India.
Building LLMs requires significant investments in hardware, data centers, and skilled talent. DeepSeek’s R1 model, for instance, was developed at a cost of under $6 million, excluding hardware and years of capital investment. Paras Chopra, founder of Lossfunk, highlighted the urgency of India catching up in AI technology. He warned that failing to do so could result in India becoming a mere consumer of technology, rather than a creator. This scenario would allow other nations to dictate the terms of technology access, much like the historical importance of India developing its own nuclear technology.
Bridging the Skills Gap
Despite the enthusiasm surrounding AI development, India faces a significant skills gap. Prayank Swaroop, a partner at Accel, pointed out that the country still lacks strong domestic AI capabilities. He emphasized the need for a proactive approach to build these skills. The focus should not only be on AI but also on preparing for future technological disruptions, such as quantum computing.
Over the past year, the Indian AI landscape has divided into two factions. One group advocates for building indigenous LLMs from the ground up, while the other prefers developing SLMs with fewer parameters tailored for specific applications. For instance, Sarvam AI has trained its platform on 2 billion parameters, focusing on Indian languages. In contrast, DeepSeek’s R1 model boasts 671 billion parameters without a specific use case. This division highlights the varying strategies within the industry and the need for a unified approach to maximize India’s potential in AI.
Observer Voice is the one stop site for National, International news, Editor’s Choice, Art/culture contents, Quotes and much more. We also cover historical contents. Historical contents includes World History, Indian History, and what happened today. The website also covers Entertainment across the India and World.