Startups Soket Labs and ZenteiQ have introduced new artificial intelligence foundation models under the government’s IndiaAI Mission. The launch aims to reduce the country’s reliance on foreign AI systems by building sovereign infrastructure for local scientific and industrial applications.
Soket Labs and ZenteiQ have officially launched their indigenous artificial intelligence foundation models as part of the government-backed IndiaAI Mission. The unveiling, which took place in the presence of Ministry of Electronics and Information Technology (MeitY) Secretary S Krishnan, marks a shift in how India aims to handle large-scale AI development.
Soket Labs introduced EKA-1, a model focused on mathematical reasoning and multilingual capabilities. Simultaneously, ZenteiQ unveiled its Dhir series, specifically the 2B and 5B variants, which are designed as scientific foundation models intended for use in engineering and industrial sectors. These launches form part of a broader government strategy to establish sovereign AI infrastructure, meaning India is looking to develop its own technological base rather than relying entirely on systems imported from global AI leaders.
For the domestic startup ecosystem, this development represents a significant step toward commercializing research. By creating foundation models that are tailored to Indian linguistic and technical requirements, these companies aim to provide alternatives to the large language models currently dominated by global tech giants. The government’s role here is to act as a catalyst, providing the necessary platform and support for these startups to mature their technology and scale it for public and private sector workflows.
However, the path ahead for these companies involves significant business and technical challenges. Soket Labs and ZenteiQ are private, unlisted startups, meaning they do not face the same public financial scrutiny as listed firms. As they grow, their success will depend on how they manage the high cost of computing power required to train and maintain these models. Developing AI at this scale is extremely capital-intensive, requiring massive access to GPUs and data infrastructure.
Furthermore, the sector is characterized by intense competition. To be viable, these indigenous models must eventually achieve performance parity with established international counterparts. Investors and industry observers will likely track whether these models can move beyond the prototype stage and find real-world adoption in industrial applications. A key risk for the business model is the level of dependency on government-provided compute subsidies and policy support. If these companies are to thrive long-term, they will need to demonstrate that their technology can be commercially sustained and integrated into mainstream business operations without relying solely on state backing.