Indian artificial intelligence (AI) startups are making significant strides in developing advanced technologies, moving beyond basic applications to create foundational models in fields such as physics and neuroscience. Notable companies include Oru’el, which provides GPU reliability solutions through physics-informed models, and ZenetiQ, which is developing a scientific large language model (LLM) for engineering applications. HumanTronik is working on hyper-personalized language models tailored for enterprise use. These innovations mark a shift from last year's focus on AI applications to a current emphasis on adding real value through technology. However, challenges remain, particularly in human capital and infrastructure, as the expertise needed for advanced model training is scarce in India. Investors are closely monitoring this evolving landscape, particularly in specialized foundation models for healthcare, materials, and physics. The journey towards catching up with global leaders like China and the US continues, as the Indian AI sector strives to overcome these hurdles.