India AI Impact Summit 2026

navigating-ai-dpi-geopolitics-an-autonomous-and-strategic-approach-for-the-global-majority

7 min read

TRACK:Democratizing AI Resources
SPEAKERS
  • Dr. Mehdi Snene, UN Office for Digital and Emerging Technologies
  • Jane Munga, Carnegie Endowment
  • Kunal Raj Barua, Aapti Institute
  • Rowan Wilkinson, Chatham House
  • Sourav Das, Co-Develop
SESSION OVERVIEW

As India and other countries integrate AI into Digital Public Infrastructure (DPI), they confront critical geopolitical challenges related to sovereignty, reducing dependency, and enabling global cooperation—while preserving openness and interoperability. Held alongside the India AI Impact Summit, this session will explore public-interest AI-DPI pathways beyond major power centric technology ecosystems. Drawing on global case studies and expert dialogue, it will examine geopolitical implications, distill lessons for the Global Majority, identify mechanisms for collaboration, and propose recommendations for DPI stakeholders and technology funders for 2026. ... Read More

VIDEO RECORDING

Navigating AI-DPI Geopolitics: An Autonomous and Strategic Approach for the Global Majority

Abstract

The panel examined how AI is increasingly being woven into Digital Public Infrastructure (DPI) across the Global Majority, confronting geopolitical tensions around sovereignty, dependence on dominant tech ecosystems, and the need for interoperable, user‑centred services. Drawing on diverse case studies—from India’s AI‑DPI roadmap to Ethiopia’s open‑source digital ID—the discussion unpacked the divergent life‑cycles of AI and DPI, identified bottlenecks to a user‑centric approach, debated a “third‑way” cooperation model beyond US/China dominance, and explored the roles of multilateral donors, development banks, and philanthropic funders in fostering autonomous, sustainable AI‑DPI ecosystems. The UN’s emerging governance mechanisms and the concept of minimal universal standards were also scrutinised, with the audience probing regulatory sandboxes, risk definition, and the practicalities of cross‑border collaboration.

Detailed Summary

  • The moderator described two intersecting technology communities:

    1. Digital Public Infrastructure (DPI) – long‑standing, open, interoperable systems (digital IDs, payment rails, data‑exchange platforms) built primarily by the Global Majority, with principles of transparency, public ownership, and citizen benefit.
    2. Artificial Intelligence (AI) – a newer, largely proprietary stack dominated by US and Chinese giants, high‑capital, closed‑source, and oriented toward commercial returns.

  • She noted the collision of these worlds in government services (chat‑bots, fraud detection, translation, etc.) and posed the central question: “On whose terms will AI integrate with DPI?” – i.e., governance, safeguards, sovereignty, and alignment with public‑interest values.

2. Vision of an Integrated AI‑DPI Future – Dr. Mehdi Snene

  • DPI as a framework: not an end in itself but a means to deliver concrete public‑service use cases (e.g., health data exchange, vaccine roll‑outs, social‑benefit distribution). Its building blocks are ID, financial transactions, and data exchange.

  • AI’s life‑cycle differs: it begins with raw data, proceeds to model refinement, and culminates in either foundational models (large language models) or business‑oriented models.

  • Convergence point: AI should accelerate DPI‑driven outcomes (health, education, crisis resilience) when it is driven by use cases, not by technological hype.

  • Human‑centred imperative:

    • Human rights, inclusivity, and safeguards must be embedded to keep AI a “sustainable technology.”
    • A bottom‑up approach—starting from citizen needs—should dictate where AI and DPI intersect.

  • Strategic recommendation: AI and DPI should co‑originate (“no AI after DPI or DPI after AI”) because both rely on data and share the ultimate goal of serving citizen needs.

3. Bottlenecks to a User‑Centred Approach – Follow‑up (Dr. Mehdi Snene)

  • Misalignment of incentives: Industry’s economic targets (LLM race, profit) often override citizen‑oriented outcomes.
  • Fragmented value chain: Many countries lack end‑to‑end capacity across the AI‑DPI lifecycle, creating hesitation in policy decisions.
  • Policy gap: Sovereignty is discussed without a re‑definition that accounts for the digital‑value chain; consequently, policies remain vague.
  • Economic pressure: Example – OpenAI’s unsustainable spending on LLMs without clear ground‑level implementation, illustrating the risk of “investment without impact.”

4. The “Third Way” – Jane Munga (Carnegie Endowment)

  • Re‑framing the question: Rather than asking whether a “third way” exists, she asked what applications are needed and who should design them.

  • Agency over technology: Emphasised regional cooperation to build solutions for the Global South, not to re‑invent entire stacks.

  • Three pillars of a “third way”:

    1. Leverage existing assets – India’s AI Mission, Latin America’s GPT‑style initiatives.
    2. National budget ownership – encourage governments to fund DPI/AI directly rather than rely solely on donors.
    3. “Third‑space” linkages – create cross‑regional innovation hubs (e.g., Indian innovators pairing with Brazilian partners) to share use‑case‑driven solutions.

  • Call for concrete collaboration: Urged the summit to generate actionable “third‑space” mechanisms that connect innovators across continents.

5. International Cooperation Beyond the UN – Jane Munga (Follow‑up)

  • Cited Mark Carney’s World Economic Forum warning about a “global rupture.”
  • Suggested regional bodies (African Union, Asian Cooperation Forum) as platforms for bilateral or multilateral AI‑DPI agreements, allowing quicker, context‑specific collaborations (e.g., Kenya‑India skill‑exchange).
  • Proposed a democratized cooperation model where countries align on shared challenges rather than being forced into a binary US‑China alignment.

6. Sovereignty vs. Interoperability – Kunal Raj Barua

  • Inherent tension: Sovereignty (control over models, data, scaling) can clash with interoperability, which often requires shared standards and external components.
  • Granular vs. high‑level definitions: While “digital sovereignty” is a common headline, practical differences emerge at data‑layer, model‑layer, and hardware‑layer.
  • Need for a “knowledge platform”: Proposes a third‑way bridging framework that captures nuanced definitions and promotes context‑specific solutions rather than a one‑size‑fits‑all rulebook.

7. Funding the Autonomous AI‑DPI Stack – Sourav Das

  • Philanthropic perspective: Co‑Develop works to catalyse partnerships around DPI; funding is currently donor‑driven, but there’s a shift toward national budget allocations.
  • Challenges with Multilateral Development Banks (MDBs):
    • Long approval cycles (≈2 years) cause technology to become obsolete before deployment.
    • Project‑based financing favors capital expenditures, neglecting operational expenditure needed for long‑term sustainability and capacity building.
  • Call for market‑driven ecosystem: Emphasised the need for vendors, integrators, and open‑source ecosystems that respond to country‑specific needs, not just to MDB‑driven projects.

8. Concrete Use‑Case Illustration – Ethiopia’s Open‑Source Digital ID

  • Ethiopia adopted MOSIP (open‑source digital ID platform) and built in‑house expertise rather than relying on a private system integrator.
  • The government created its own system‑integration unit, allowing full control over the stack and enabling AI layers (e.g., fraud detection) to be added on top.
  • Key lesson: Capacity building (training developers, retaining talent) is essential; without it, countries risk vendor lock‑in and loss of autonomy.

9. UN Governance Mechanisms – Dr. Mehdi Snene

  • Scientific Panel on AI: 40‑person expert panel issuing bi‑annual reports on AI risks/opportunities, modeled on the IPCC.
  • Global Digital Compact (GDC): Adopted by UN member states; establishes minimum standards for digital governance, emphasizing human‑rights‑anchored policies.
  • Global Dialogue: First session in Geneva (July) and second at the UN General Assembly (Sept 2027) to bring together states, civil society, tech operators, and other stakeholders.
  • Shift toward “minimum viable governance”: Proposes defining an irreducible set of policy rules that can be universally accepted, avoiding overly complex, sovereignty‑threatening frameworks.

10. Defining Minimal Standards & Risk

  • Emphasised that states have yet to articulate acceptable AI risk thresholds, making it difficult to craft cohesive governance.
  • Minimal standards should be human‑rights‑based, technology‑agnostic, and flexible enough to accommodate differing levels of AI maturity across countries.

11. Regulatory Sandboxes – Audience Interaction & Panel Insights

  • Question: Could joint regulatory sandboxes (cross‑country test‑beds) accelerate AI‑DPI co‑creation?
  • Panel consensus:
    • Potential – sandboxes enable co‑identification of problems and shared testing of solutions.
    • Practical hurdles – require intensive inter‑departmental liaison, alignment of jurisdictional rules, and resource‑intensive governance.
    • Current experiments – Karnataka (India) developing a sandbox; still early, with challenges around agenda‑setting and cross‑sector coordination.

12. Risks of Fragmentation & Contingency Planning – Audience Questions (Jane Munga & Others)

  • Geopolitical rupture (sanctions, supply‑chain disruptions) could fragment AI‑DPI ecosystems.
  • African context: With only ~40 % internet penetration, basic connectivity and energy remain priority over advanced AI‑DPI resilience.
  • Contingency thinking: Focus on building local data ecosystems (e.g., African‑led data coalitions like Masakane) and ensuring that deployed technologies are adaptable, not locked to a single vendor or geopolitical bloc.

13. Closing Remarks & Outlook

  • UN’s role: Acts as a convening platform, not a rule‑setter; member states retain sovereign decisions on cooperation.
  • Future pathways: Emphasis on South‑South collaboration, open‑source development, capacity building, and minimal universal standards that respect sovereignty while enabling interoperability.

Key Takeaways

  • AI and DPI have divergent life‑cycles but share a common data‑centric goal; successful integration requires a bottom‑up, citizen‑needs‑driven approach.
  • Economic incentives and fragmented value chains are the primary bottlenecks preventing user‑centred AI‑DPI deployment.
  • A “third way” is less about creating a brand‑new stack and more about leveraging existing regional initiatives, national budget ownership, and cross‑regional “third‑space” partnerships.
  • Sovereignty and interoperability are not mutually exclusive; the tension lies in technical layers (hardware, data, models) and can be mitigated through knowledge platforms and nuanced policy definitions.
  • Funding models must evolve: MDBs’ long cycles and project‑based focus hinder agility; philanthropic and national budget funding together with a market‑driven ecosystem can improve sustainability.
  • Ethiopia’s MOSIP‑based digital ID demonstrates that building in‑house expertise avoids vendor lock‑in and enables AI layering on DPI.
  • The UN’s emerging AI scientific panel and Global Digital Compact aim to establish minimum universal standards rooted in human rights, avoiding over‑prescriptive global governance that could clash with sovereignty.
  • Regulatory sandboxes hold promise for joint AI‑DPI testing but require extensive inter‑departmental coordination and clear jurisdictional agreements.
  • Geopolitical fragmentation underscores the need for basic connectivity, energy security, and locally controlled data ecosystems before sophisticated AI‑DPI resilience can be realized.
  • South‑South collaboration, open‑source solutions, and capacity‑building are the most viable routes for the Global Majority to achieve autonomous, inclusive AI‑DPI ecosystems by 2026.

See Also:

Graph View

Backlinks