India is making significant strides in artificial intelligence by launching its own secure and affordable AI model. Union Minister for Electronics and Information Technology, Railways, and Information & Broadcasting, Ashwini Vaishnaw, announced this initiative in New Delhi at Electronics Niketan. Highlighting India’s commitment to technological advancement, he emphasized that the AI model will be built with inclusivity in mind, ensuring that modern technology reaches all sections of society. With the backing of a high-end common computing facility, India is set to develop multiple foundational AI models tailored to the country’s needs, including Indian languages.
As part of the IndiaAI Mission, researchers, academia, startups, and MSMEs will have access to 18,693 GPUs at competitive prices. Of these, 15,000 are high-end GPUs sourced from major manufacturers and will be offered at best-in-class pricing, with discounts of up to 47 per cent. This initiative aims to empower AI development in India by ensuring affordability and accessibility. The computing facility, currently operating with 10,000 GPUs, will soon integrate the remaining 8,693 GPUs, benefiting researchers, students, and developers. Compared to global models that cost approximately $2.5 to $3 per hour for computation, India’s AI model will be available at less than ₹100 per hour after a 40 per cent government subsidy. Additionally, attractive half-yearly and annual plans will make AIcomputinge services even more affordable.
India’s AI ecosystem is witnessing remarkable growth, with six firms set to develop foundational AI models using the newly provisioned GPUs under the IndiaAI Mission. These models will focus on preserving Indian culture, eliminating biases, and utilizing datasets relevant to Indian citizens. The government is also prioritizing AI safety by establishing an AI Safety Institute, which will follow a hub-and-spoke model to enable researchers, developers, and stakeholders to create tools, frameworks, and processes for AI security. Eight projects focusing on indigenous AI governance have been initiated to ensure privacy, ethical AI auditing, and bias mitigation.
The government has selected 18 projects under the Application Development Initiative. These projects will address crucial areas such as agriculture, climate change, and learning disabilities, leveraging AI to develop impactful solutions. With a focus on democratizing AI, the government has also ensured that a robust common computing infrastructure is in place. The facility, which currently houses 18,693 GPUs, surpasses global benchmarks and significantly exceeds initial projections. DeepSeek, an open-source AI model, will be hosted on Indian servers following security protocol checks, allowing developers, coders, and researchers to benefit from its capabilities.
The IndiaAI Mission, approved with a budget of ₹10,372 crore, aims to bridge gaps in the existing AI ecosystem and position India as a global hub for AI development. The Ministry of Electronics and Information Technology (MeitY), through the IndiaAI Independent Business Division (IBD), initiated a process to empanel AI services on the cloud. Several bidders, including major cloud service providers and data centre companies, have participated in this initiative, offering AI compute units such as Intel Gaudi 2, AMD MI300X, and NVIDIA H100. The lowest bidders in each category will be empanelled first, and other bidders can match their prices to participate.
The establishment of the IndiaAI Safety Institute marks a crucial step in ensuring AI risks and challenges are addressed effectively. The institute will work in collaboration with academia, startups, industry, and government agencies to develop indigenous AI research that aligns with India’s cultural, economic, and linguistic diversity. It will also focus on responsible AI deployment through initiatives such as watermarking, ethical AI frameworks, risk assessment, and deepfake detection. To support the development of foundational AI models, the government has launched a Call for Proposals inviting startups, researchers, and entrepreneurs to collaborate on AI models designed for Indian needs. These models will be based on Indian datasets, ensuring linguistic and contextual relevance.
India’s AI model is expected to be ready within six to ten months, offering cost-effective and innovative solutions for various sectors, including healthcare, education, agriculture, and governance. The AI applications developed under this initiative will enhance predictive analytics in areas such as weather forecasting, disaster management, and soil health monitoring. The affordability of computing facilities further ensures that AI technology is accessible to startups, researchers, and students, helping India strengthen its position as a global AI leader. The initiative has also received international recognition, reinforcing India’s reputation as a trusted technological powerhouse.
Also Read | AI Innovation in India Gets a Boost with 18000 GPUs
With a structured roadmap and a focus on sustainability, the IndiaAI Mission is aligned with the government’s broader vision for semiconductor development and AI innovation. The country has invested over $30 billion in the semiconductor mission, further supporting its long-term technological goals. AI’s real impact extends beyond chatbots and image generation; it is set to transform industries such as oil drilling, railway optimization, and infrastructure management. India also ranks among the top nations in AI education, with over 240 universities offering AI courses and 100 institutions equipped with 5G labs.
The event announcing these initiatives was attended by Minister of State for Electronics & Information Technology, Jitin Prasada, Additional Secretary of MeitY and CEO of IndiaAI, Abhishek Singh, COO of IndiaAI, Smt. Kavita Bhatia, and other senior officials. The government’s commitment to AI development, affordability, and inclusivity ensures that India is poised to emerge as a global AI powerhouse, building technological advancements that benefit society at large.