A futuristic robotic hand elegantly presents various educational and technological icons against a modern office background, emphasizing themes of growth and innovation. Auvana | Photo credit: Shinsei Movements
On the three broad core technical ingredients for the AI ecosystem, namely compute, data and fundamental models, India has taken a public-led approach. In computing, India has a common pool access model of over 34,000 GPUs (Graphics Processing Units), democratizing hardware access among research institutions, startups and enterprises. Through AI Kosh there is an open data repository with hundreds of datasets, suitable for model training and general interest applications. Similarly, a public-private-academic triad exists between MeitY’s IndiaAI mission as the organiser, SarvamAI as the developer and IIT Madras as the talent and research center to build India’s sovereign foundational LLM model.
India’s pioneering experience in building DPI is now shaping the way AI is interwoven into the nation’s digital rails. The Aadhaar identity system increasingly uses AI-enabled facial matching to strengthen authentication on a population scale. In digital payments, AI-powered voice-based interfaces such as Hello! UPI expands financial access to users who may not be comfortable with text-based interfaces. Similarly, Open Healthcare Network – a UN digital public good from India – now includes GenAI-based medical writer tools that convert doctor conversations into structured electronic health records, integrated with the Ayushman Bharat Digital Mission’s national health data framework.
Local accessibility
India has also put local accessibility and linguistic diversity at the heart of its AI mission. Through initiatives like Bhashini, under the National Language Translation Mission, the government is supporting the creation of open-source multilingual datasets in 22 Indian languages and is now offering a tool called Shrutlekh for AI-driven real-time speech recognition, translation and transcription. This is complemented by the work of AI4Bharat and BharatGen as the academic center for NLP and Indian language speech recognition to create tools and multilingual LLMs that reflect India’s cultural and linguistic diversity. Together, such efforts are enabling a generation of language-inclusive AI applications, exemplified by startups like Bharat Intelligence, which are using voice-based AI to organize rural agricultural labor markets.
India is also beginning to integrate indigenous epistemologies and cultural context into the development and governance of AI systems. Nishpaksh is a new framework for AI model certification, which adds a crucial layer of assurance to auditing the fairness of models specific to India’s socio-cultural contexts. At the same time, scholars and practitioners are exploring the integration of indigenous knowledge systems, such as reasoning frameworks from the Nyaya Sutra, to enrich AI beyond Western logical structures.
India’s AI innovation ecosystem is at an inflection point, where the building blocks of a clear, public-led AI journey are in place, but the next phase will depend on how the country deals with chip geopolitics, energy challenges and data protection. Today, the global distribution of computers still remains extremely skewed, with an overwhelming majority concentrated in the US (75 percent) and China (15 percent), known as the Compute North/Compute South divide. For Indian AI startups, this translates into less access to training-level computing equipment and slower iteration cycles, which could dampen the momentum of the innovation ecosystem.
This imbalance is exacerbated by the energy needs that shape the geography of computing, requiring AI data centers to be located in regions with cheap, highly reliable electricity and cooling infrastructure, which is to the detriment of energy-importing countries like India. From its population-level DPI implementation, India already manages sensitive identity and payment data, and these must dynamically keep pace with AI-era datasets, especially in sensitive domains such as healthcare.
Essentially, a public-led AI ecosystem with shared computing power, digital public infrastructure, local language inclusion and indigenous knowledge will make India’s approach to building AI ecosystems look different, but chip geopolitics, energy costs and data safeguards will determine how far it scales.
The writer is a professor at the Adam Smith Business School, University of Glasgow
Published on February 11, 2026
#promise #building #distinctive #ecosystem