On March 6th, 2025, the IndiaAI Mission, orchestrated by the Ministry of Electronics and Information Technology, launched “AI Kosha,” a national dataset platform designed to harmonise various ministries’ and departments’ non-personal data streams to train and develop AI models and tools. This wasn’t just a collection; it was the score for India’s AI future. To ensure every musician could play, they also introduced the Common Compute portal, a shared orchestra pit of high-powered GPUs, empowering startups and academia to compose their own AI masterpieces.
After Gemini, ChatGPT, DeepSeek, and other LLMs developed by the USA, China, and other countries, it’s our time to unearth the digital goldmine with India’s own AI model. AI Kosha is an ambitious and prospective initiative towards attaining that objective.
We will cover the following topics in today’s passage to feed your grey cells, which are curious about India’s AI mission and how businesses can contribute and benefit.
- What are the objectives of AI Kosha?
- How will the Common Compute facility provide graphics processing units to help startups, researchers, and developers?
- How many GPUs have been commissioned?
- What is the roadmap of India’s AI mission?
- Why are translation models significant?
Let’s commence.
What is India’s AI Mission?
Launched in 2023 as a collaborative effort between the Ministry of Electronics and Information Technology and NASSCOM, IndiaAI has transitioned from an industry-government partnership to a fully government-led mission. This transformation, unveiled last March, is backed by a substantial investment of Rs. 10,371.92 crore.
Fueled by the explosive growth of generative AI, exemplified by models like ChatGPT and a surge in global AI development, the Union Cabinet approved the IndiaAI mission. Its core objectives are twofold: to foster AI innovation within India and to ensure AI’s beneficial application across the nation—summarised as ‘Making AI in India and Making AI Work for India.”
The IndiaAI mission, amidst its various interventions and resources for private companies and academia, has recently prioritised the Common Compute facility. This facility actively provides GPUs to startups that cannot afford to purchase them in bulk.
GPUs in India: The Journey So Far
The Power of GPUs in AI
Graphics Processing Units (GPUs), traditionally known for rendering images in PCs and phones, have become indispensable for AI research. Their ability to handle complex calculations with massive datasets makes them crucial for training and running generative AI models.
This demand has propelled companies like Nvidia beyond gaming, establishing them as key players in global AI infrastructure. However, the high cost and limited availability of GPUs, such as Nvidia’s Blackwell, which is reportedly sold out until October, pose significant challenges.
IndiaAI Mission’s Response
To address this, the IndiaAI Mission is actively empanelling data centres within India that possess GPUs. The mission will manage access to these computing resources, determining allocation and duration for startups and researchers. This initiative aims to democratise access to essential AI infrastructure, enabling innovation even for those who cannot afford to purchase GPUs in bulk.
Current Status and Future Plans
IT Minister Ashwini Vaishnaw has announced that 14,000 GPUs have already been acquired and commissioned for use by empanelled data centres. The IndiaAI Mission plans to continue acquiring GPUs quarterly, ensuring a steady supply of these critical resources to support the growing needs of India’s AI ecosystem.
What is AI Kosha?
AI Kosha represents the practical application of the IndiaAI Datasets Platform, a critical pillar of the mission. A significant hurdle in developing AI models for India is the prevailing bias of existing datasets, heavily skewed towards English and Western perspectives. Even online contributions from Indian users often default to English, hindering the creation of locally relevant data. AI Kosha aims to rectify this by providing a centralised repository of India-specific data, addressing the inherent limitations of models trained predominantly on Western sources.
Addressing Linguistic and Cultural Nuances
The initial datasets uploaded to AI Kosha, including translation models across various Indian language pairs, exemplify the mission’s commitment to mitigating linguistic bias. By granting researchers and companies broader access to local content, AI Kosha facilitates the development of AI tools that accurately reflect India’s diverse linguistic and cultural landscape. This initiative ensures that AI solutions are robust, contextually relevant, and culturally sensitive.
The Significance of Local Data for Effective AI Deployment
While open-source AI models offer a foundation for building powerful generative AI tools, policymakers and developers emphasise the necessity of local data for effective deployment. Access to India-specific data is crucial for creating AI applications that surpass the capabilities of foreign offerings from companies like OpenAI and Google. AI Kosha’s initiative to democratise access to local data is essential for enabling Indian firms to develop AI solutions that genuinely meet the unique needs and challenges of the nation.
India AI Mission: Seven Pillars and Scope of Work
India’s AI mission has seven pillars overall.
- IndiaAI Compute: Develop a scalable AI ecosystem with compute capacity
- IndiaAI Innovation Centre: Build AI solutions for critical sectors
- IndiaAI Dataset Platform: Provide access to non-personal datasets
- IndiaAI Application development initiatives: Promote AI applications in key sectors
- IndiaAI FutureSkills: Increase access to AI education
- IndiaAI Startup Financing: Provide funding for AI startups
- Safe & Trusted AI: Ensure responsible and ethical AI development
Prioritising AI Safety and Responsible Development
Beyond the spotlight on the Common Compute facility and AI Kosha, driven by cost-effective AI advancements like DeepSeek, Indian policymakers are strategically addressing AI safety. The upcoming AI Safety Institute of India will play a crucial role in mitigating the unintended consequences of AI deployment.
This includes safeguarding against harmful chatbot outputs and ensuring the reliability of specialised AI tools in critical sectors like biomedical research. Aligned with global efforts, this institute will establish guardrails and rigorously examine the risks associated with AI’s rapid expansion across industries and government.
Fostering Innovation and Practical AI Applications
The IndiaAI mission also builds a robust ecosystem for AI innovation and practical application. The IndiaAI Innovation Centre will focus on developing domain-specific foundation models that cater to unique Indian needs. Concurrently, an application development initiative aims to create commercially viable and productive AI use cases.
Futureskills will establish AI labs in smaller cities to democratise AI access, and startup financing will fuel emerging ventures. These initiatives work in concert to translate research into real-world solutions and stimulate the Indian AI market.
Driving Indigenous Foundation Model Development
The mission’s call for proposals to develop an indigenous foundation model is a pivotal effort. Recognising the resource-intensive nature of this endeavour, the Union government is committed to accelerating this process. With 67 applications already received and under evaluation, the mission aims to expedite the creation of a foundational AI platform within months.
This initiative will significantly reduce reliance on foreign models and empower India to build AI solutions tailored to its specific requirements and cultural nuances, solidifying its position in the global AI landscape.