India Proposes Digital Public Infrastructure Model to Democratise AI Access
India proposes a Digital Public Infrastructure approach to democratise AI access through modular public-good enablers that address coordination gaps. A government white paper outlines this framework to make AI building blocks accessible beyond large firms, leveraging India's DPI successes like UPI, now available in eight countries. However, mixed results from platforms like ONDC, which processes significantly fewer transactions than private competitors, highlight implementation challenges.
*this image is generated using AI for illustrative purposes only.
India is proposing a Digital Public Infrastructure (DPI) model to democratise artificial intelligence infrastructure, building on its successful track record with platforms like Aadhaar and UPI. The government aims to make AI building blocks including data, compute, models and algorithms widely accessible beyond large firms and urban hubs.
Government White Paper Outlines DPI for AI Framework
A recently released government white paper explains the concept as "establishing shared, standards-based layers that improve access interoperability, accountability and trust". The document clarifies that DPI for AI should not be understood as a single platform or monolithic system, but rather as a set of modular public-good enablers that address specific coordination gaps in the AI ecosystem.
The white paper defines the value proposition of this approach:
| Benefit | Description |
|---|---|
| Access Pathways | Predictable, transparent and interoperable access for smaller participants |
| Cost Reduction | Lower barriers for startups, research institutions and smaller firms |
| Standardisation | Common interfaces and governance norms |
| Expanded Participation | Broader base of AI infrastructure beneficiaries |
Building on India's DPI Success Story
India's existing DPI platforms have demonstrated global impact. Aadhaar and UPI are recognised as standout successes in digital identity and digital payments respectively. UPI's international expansion showcases India's DPI expertise:
| Region/Country | UPI Availability Status |
|---|---|
| Singapore | Available |
| UAE | Available |
| France | Available |
| Sri Lanka | Available |
| Bhutan | Available |
| Qatar | Available |
| Mauritius | Available |
| Nepal | Available |
More countries are expected to accept UPI as a payment mode as India seeks to export its biggest DPI success.
Mixed Track Record of DPI Implementations
While some DPI platforms have achieved remarkable success, others face challenges. The Open Network for Digital Commerce (ONDC) provides a contrasting example. Despite completing nearly three years of operations, ONDC's performance remains modest compared to private sector alternatives.
| Platform Performance | Daily Transactions |
|---|---|
| ONDC | 5.90+ lakh transactions |
| Top 3 Quick Commerce Players | 8X more than ONDC |
According to a 2025 year-end review by the Department for Promotion of Industry and Internal Trade, average daily transactions on ONDC have reached 5.90 lakh-plus. However, just the top three companies in quick commerce—BlinkIt, Zepto and Swiggy Instamart—collectively process approximately eight times more transactions than ONDC daily.
AI-DPI Convergence Opportunities
The convergence of AI and DPI presents significant opportunities in governance and service delivery. Potential applications include:
- Enhanced fraud detection capabilities
- Improved grievance redressal systems
- Predictive analytics for risk assessment
- Crop loss forecasting
- Better identification of at-risk sectors or companies
Critical Success Factors
The success of DPI for AI will depend on two key elements: infrastructure availability at its core and the depth of the model and algorithmic ecosystem built around it. A DPI approach is positioned as critical for building openness, interoperability, inclusion and trust into India's AI foundations, potentially offering an alternative to models where private companies drive both AI development and regulation.
20
