India is positioning its AI Stack as the backbone for “AI for Humanity,” aiming to democratise access to intelligent applications across sectors while strengthening sovereign capabilities in models, compute, infrastructure and energy. A new government note describes how layered investments in applications, models, compute, networks, data centres and clean power are intended to support population‑scale AI deployment without concentrating capabilities in a few firms or countries.
Application and Model Layers Anchor Real-World Impact
At the top of the stack, India is pushing AI applications in agriculture, healthcare, education, justice and disaster management, with state‑level pilots reporting productivity gains of 30–50 percent in some farm advisory deployments and AI tools aiding early diagnosis of diseases like tuberculosis and cancer.
Initiatives such as AI‑enabled e‑Courts, AI‑supported weather forecasting and AI modules in school curricula are framed as examples of AI moving from experimentation into everyday decision‑making and service delivery. Beneath this, the AI model layer is being built around sovereign, India‑centric models and shared repositories: under the IndiaAI Mission, 12 indigenous models are in development, BharatGen is working on India‑specific foundation and multimodal models, and the IndiaAI Kosh platform now hosts 5,700+ datasets and over 250 AI models from dozens of entities across 20 sectors.
Language technologies such as Sarvam AI’s large language and speech models and Bhashini’s 350+ multilingual AI models are intended to ensure that AI is usable in Indian languages and public‑service contexts.
Compute, Infrastructure and Energy Underpin Scale and Sovereignty
To power this ecosystem, the compute layer combines subsidised cloud access with domestic hardware and supercomputing initiatives. The IndiaAI Compute Portal offers more than 38,000 GPUs and 1,050 TPUs on a compute‑as‑a‑service model at under ₹100 per hour, while a secure national GPU cluster of 3,000 next‑generation GPUs is being set up for sovereign and strategic workloads.
These efforts sit alongside the ₹76,000‑crore India Semiconductor Mission’s 10 approved projects, indigenous SHAKTI and VEGA processor designs, and over 40 petaflops of supercomputing capacity under the National Supercomputing Mission, including systems like PARAM Siddhi‑AI and AIRAWAT tuned for AI workloads.
On the infrastructure side, India has rolled out 5G in all states and UTs with coverage across 99.9 percent of districts, built a nationwide fibre network, and grown data‑centre capacity to about 960 MW with projections of 9.2 GW by 2030, supported by large AI‑linked investments from global cloud providers.
All of this rides on a power sector that has crossed 500 GW of installed capacity, more than half from non‑fossil sources, with plans for large pumped‑storage and battery systems and new nuclear capabilities under the SHANTI Act to provide round‑the‑clock clean power for AI data centres.
Towards an Inclusive, Sustainable AI Future
The government frames this integrated stack as both an economic and social project: by lowering barriers to compute, building indigenous models, localising applications and anchoring infrastructure and energy domestically, India aims to ensure that AI benefits can be delivered at population scale without dependence on a handful of foreign platforms.
The note emphasises that AI is being treated as a general‑purpose technology for public welfare—supporting farmers, students, clinicians, courts and disaster‑response agencies—as much as a driver of private‑sector innovation. By combining affordability, sovereignty and sustainability across each layer of the stack, India seeks to create an AI ecosystem that is resilient, future‑ready and aligned with national priorities of inclusive growth, social equity and climate commitments.
