BENGALURU: Two years after OpenAI’s ChatGPT took the world by storm, China’s AI model DeepSeek has now sparked a debate among tech enthusiasts on building low-cost AI models and India’s potential to tap the opportunity, train foundational AI models relatively cheaper than the cost involved currently.
Currently, Indian tech firms have tied up with leading maker of computer chips Nvidia to accelerate AI adoption. While DeepSeek has claimed its R1 model took only $6 million to build and that too using Nvidia’s less-advanced chips, it had cost $100 million for OpenAI to build GPT-4 and Nvidia is the key supplier of advanced chips to the company.
This shows that countries, especially India, need not depend on advanced hardware and that this R1 model drives home the importance of innovation. Bruce Keith, co-founder & CEO, InvestorAi, said DeepSeek R1 has definitely challenged the dominance of a few players in the models and data ecosystem - OpenAI, Google and Meta will feel it the most. This genuinely democratises AI and gives countries who don’t have existing infrastructure a huge leap forward to experiment and be part of the frontier.
The fact that they created this platform with under $6 million investments has shaken the tech CEOs globally highlighting that game changing innovations don’t necessarily need billion-dollar investments, he added. Narendra Bhandari, General Partner, Seafund, a deeptech focused early-stage venture fund, said that this demonstrates a fantastic opportunity for India and that start-ups can aspire to build relevant models at much lower costs.
In its recent India’s Generative AI Startup Landscape 2024 report, Nasscom revealed that the country’s Generative Artificial Intelligence (Gen AI) start-ups have attracted over $750 million in cumulative funding since 2023. Bhaskar Majumdar, managing partner, Unicorn India Ventures, said India is jumping on to AI mission with massive Nvidia GPU (Graphics Processing Units) purchase plans.