Countdown to the India AI Impact Summit 2026: Insights from ORF’s Pre-Summit Engagements

Introduction

The countdown to the India AI Impact Summit 2026 is nearly over. Scheduled to be hosted by India in New Delhi on 19–20 February 2026, the Summit will be the largest-ever gathering of AI stakeholders in the world. The convening is the fourth in a series of annual international AI Summits, the first of which took place in the United Kingdom in 2023; the second in South Korea in 2024; and the third in France in 2025, with India as co-chair.

Across these three years, there has been a marked increase in global interest and representation at the Summits: from 28 states and a few tech companies taking part in 2023;^[1] to more than 100 states and a host of tech organisations, multilateral bodies, and civil society organisations (CSOs)—over 1,000 representatives in all—attending the Paris Summit in 2025.^[2],[3] The numbers are expected to be even higher in the New Delhi gathering.

The emphasis has shifted too. In the initial years, the overriding concern was around AI safety and regulation—the curbs and guardrails that AI development requires.  Conversations then transitioned to a focus on the actions being taken to build and govern AI. Four years into the deliberative process, with AI use having become widespread, India views this moment as opportune for nurturing a better understanding of the impact of AI.^[4] The emphasis on AI safety and governance, though, will continue to retain their importance.

Importantly, this is the first AI Summit to be hosted by a country in the Global South. A primary point of focus will be how AI can support the Global South’s development and growth. With the broad theme, ‘Welfare for All, Happiness of All’, and the guiding mantra of ‘People, Planet, Progress’, the Summit weaves a concern for the Global South into the very fabric of its design. The seven chakras or thematic pillars of the conference are: (1) Human capital; (2) Inclusion for social empowerment; (3) Resilience, innovation and efficiency; (4) Promoting scientific cooperation; (5) Democratising AI resources; (6) AI for economic development and social good; and (7) Safe and trusted AI.^[5] Taken together, their developmental and social-economic-environmental imperative is apparent.

Another centrepiece of the Summit will be India’s effort to position itself as the “AI use case capital of the world.”^[6] Its sheer size and scale are an advantage: with India’s multiple demographic, economic, and consumer groups, it would not be an overstatement that if an AI solution succeeds in India, it might find application in other geographies across the globe. Conversely, AI tools developed in other parts of the world, if adapted, are likely to find practical use in India. Showcasing AI use cases, and promoting related knowledge exchanges, will thus be a core priority.

Between August 2025 and January 2026, Observer Research Foundation (ORF) hosted a series of pre-Summit events on many of the above themes and allied issues. Held in different parts of the world, the convenings brought together AI experts from governments, businesses, CSOs, think tanks, and academia. Their insights, and the points of consensus achieved, will inform deliberations at the New Delhi gathering, and contribute towards the Leaders’ Declaration with which the AI Impact Summit will conclude.

This report compiles the principal takeaways and recommendations from ORF’s 22 pre-Summit events, and acts as a ready reckoner for the AI-related opportunities and challenges that confront the world today.

Pre-AI Summit Events Hosted by ORF

Between August 2025 and January 2026, ORF and its partners organised 22 pre-Summit events across locations in the Global North and Global South. These engagements sought to address themes aligned with the Summit’s priorities, were widely attended by prominent AI stakeholders, and resulted in rich insights and recommendations. The following table presents an overview of the events.^[a]

Thematic Breakdown and Analysis

The events hosted by ORF revealed the emergence of a distinctly Global South-oriented AI agenda that called for technology to be shaped by the contexts it serves and not merely adopted from elsewhere. The following thematic analysis distils the principal threads that ran across these discussions. 

1. AI for Development and the Global South 

A thread across all events was an emphatic agreement on the need to centre AI development and discourse around the needs and aspirations of developing countries. The ministerial and senior UN participants at the UNGA side event in New York, ‘At the Heart of Development’ and ‘India Day @ UNGA’, framed aid, trade, and technology as primary levers for achieving the Sustainable Development Goals (SDGs), with India’s digital public infrastructure (DPI) presented as a transformative scalable platform for the broader Global South.

The Cairo roundtable echoed this framing by proposing a Global AI Infrastructure Fund and a Multilingual AI Fund to support AI development for the Global South, and to prevent the monopolisation of compute capacity and foundation models by wealthy nations and incumbents. In a similar vein, the Johannesburg T20 roundtable, ‘Breaking Barriers’, highlighted how continuing competition between US and Chinese AI ecosystems risks marginalising countries that are neither producers nor sizeable consumers of frontier models while calling for interoperability standards for mitigating such marginalisation. The events saw a resounding agreement on ensuring that AI governance and development do not leave the world’s majority population as passive recipients but instead empower them as co-creators of the future. 

2. Inclusion: Health, Education, and Gender 

A number of events focused on how AI can serve historically marginalised populations. The women’s health roundtable in New Delhi brought together government officials, hospital leaders, and foundation heads to argue for interoperable, annotated health datasets that capture women-specific profiles, the integration of gender indicators into digital health frameworks, and AI-readiness training for frontline health workers. An event in Kolkata on “AI as a Teacher’s Ally” emphasised lightweight AI tools for resource-constrained schools; and called for teachers, administrators, and parents to actively participate in the development of educational AI. The Barcelona health session proposed cross-border AI sandboxes linked to India’s Ayushman Bharat Digital Mission (ABDM) and the EU Health Data Space, as well as representative, multi-population datasets to reduce algorithmic bias. These discussions underscored the principle of intersectoral inclusion, which can be achieved only through deliberate design choices, local data, and community agency.

3. Human Capital and Talent Transformation

The workforce dimension of the AI transition surfaced repeatedly through the programmes. The Bhubaneswar roundtable, “AI@India: Work, Markets, and Cities in Transition”, examined how automation is disrupting and reshaping employment across sectors. The discussion identified the core challenge as not just displacement but the speed and flexibility with which individuals and institutions can adapt to AI-driven change. The Barcelona panels similarly argued for a shift from producing narrow, highly specialised AI engineers towards hybrid talent models that combine domain expertise with AI literacy. Several events converged on the recommendation that skilling and reskilling must be embedded within broader institutional reforms ranging from university curricula and vocational training ecosystems, to cross-border talent mobility frameworks.

4. Governance, Regulation, and Trust

Almost every event grappled with the tension between enabling innovation and managing risk. In New Delhi, the dual-use implications of frontier technologies like AI, quantum computing, and biotechnology were examined through a security lens, with panellists stressing the need for adaptive regulatory frameworks and early-warning mechanisms to detect risks before they propagate. The Barcelona panels and the Johannesburg roundtable both cautioned against importing regulatory templates wholesale. They emphasised that regulations should be use-case–based, proportionate to actual harms, and sensitive to the compliance capacity of smaller firms.

The Abu Dhabi event on ‘Secure Frontiers, Shared Futures’ advanced the argument that cybersecurity and resilience should be framed as enablers of AI adoption, while noting the benefits of interoperable, context-specific validation frameworks over a single global standard. And the Paris convening saw arguments for making inclusion a deliberate design choice, not an afterthought. Speakers stressed that trust—between people, institutions, and countries—will be the defining currency of AI adoption. Across these discussions, a consensus emerged that trust in AI should be built through transparency, user literacy, culturally relevant training data, and infrastructure resilience.

5. Sustainability and Infrastructure Resilience

The environmental cost of scaling AI received sustained attention. The Nagpur ‘Tech Huddle’ on AI and sustainability examined the energy and water demands of data centres against India’s net-zero commitments, calling for responsible AI deployment, energy-efficient solutions, and incentives for renewable-powered compute. The Abu Dhabi panels deepened this by noting the benefits of treating AI infrastructure as critical national infrastructure, and emphasised the need for governments to strengthen the resilience of AI infrastructure to prevent the creation of single points of failure. The UAE’s approach of integrating AI deployment into national energy planning, updating grid codes, and investing in small modular reactors (SMRs) was highlighted as a model. Overall, these events positioned the AI-energy nexus and AI’s fast-growing resource-intensity as a central policy challenge that the Summit should address.

6. The Convergence and Symbiosis of AI and Digital Public Infrastructure 

India’s DPI stack featured prominently in discussions as a high-potential toolset for building AI ecosystems from the bottom, up. The Dubai roundtable on ‘The DPI Exchange’ made the most detailed case, arguing that every DPI layer generates structured, purposive, high-quality data that AI models require. This allows AI solutions to be built atop DPIs—in other words, DPI could power AI. At the same time, AI can strengthen DPI too. Several Indian examples of this were presented, e.g., the use of AI to detect payment frauds, and the integration of voice AI into digital identity services. The benefits of DPI in African contexts were also highlighted, and much interest expressed in exploring the ‘AI+DPI’ approach to empowerment. The Spain–India Forum positioned DPI as the infrastructure through which small and medium enterprises (SMEs), startups, and underserved groups can participate more meaningfully in the AI economy. Collectively, the events highlighted DPI’s significance as a service-delivery mechanism and AI enabler.

7. International Cooperation and Other Partnership Architectures 

Finally, ORF’s events sketched a vision of the partnership architectures needed to realise equitable AI development and adoption. In Geneva, speakers outlined the possibility of a “third way” for bridging the gap between deregulatory approaches and centralised control. South–South cooperation featured prominently in the New York and Cairo events; India–Europe collaboration was the focus in Barcelona; and possible India–UAE–Africa linkages were explored in Dubai, Abu Dhabi, and Johannesburg. The latter roundtable proposed an AI Transparency Forum modelled on the Financial Stability Board, while the Cairo event envisioned a Middle East and North Africa (MENA) regional alliance combining energy, talent, data, and industrial capacity. Across the board, plurilateral, interoperable arrangements that respect local context while enabling cross-border collaboration were identified as pathways for equitable AI futures. Another observable trend was the interest in building ‘startup bridges’, and creating mechanisms for AI-focused startups to innovate jointly and collaborate across borders.

Recommendations

1. AI systems should be governed in ways that prioritise trust and resilience rather than relying solely on static compliance frameworks.

Pre-Summit discussions noted the absence of universal “gold standards” for AI safety, with governance instead evolving through use-case-specific validation. Moreover, conversations on cybersecurity highlighted how failures in AI-enabled critical systems erode public trust, making resilience a prerequisite for public support for adoption. Regulatory sandboxes were presented as practical tools for testing and scaling AI products while achieving a balance between innovation and safety. With these considerations in mind, stakeholders should work towards adaptive, context-sensitive governance frameworks that prioritise real-world robustness and accountability over rigid, one-size-fits-all rules, recognising that trust is essential for adoption while overly prescriptive approaches can introduce unnecessary barriers to innovation.

2. Crafting local data protection regimes and infrastructures should be a priority in order to address data sovereignty and curb data extraction from the Global South. 

Several discussions underscored that current AI development pathways remain heavily data-extractive, with data generated in the Global South frequently being used to train models whose economic and strategic benefits are felt elsewhere. Without regulatory intervention, countries risk becoming mere data suppliers while remaining dependent on externally developed models and platforms. Imbalances in data access and the digitisation of legacy data systems between countries often exacerbate these dynamics. Investing in local data protection laws and trusted privacy-protecting infrastructures is thus crucial, with clear stipulations about the consent needed for personal data to be used to train AI models, and frameworks for addressing new, emerging forms of data such as synthetic data.

3. Stronger investments need to be made in AI-fluent human capital through targeted skilling and upskilling programmes and the development of hybrid talent models.

There is a global AI talent shortage, with the demand for AI-aware and AI-trained workers outstripping supply. Newer kinds of collaborations between governments, the tech industry, and educational institutions need to be forged to build a stronger AI skills pipeline. Besides, AI reskilling and upskilling programmes must be designed across sectors, and AI-focused micro-credentials and introductory modules integrated into technical and vocational education and training (TVET). Across discussions, it was also noted that narrow definitions of “AI talent” risk excluding domain experts in health, mobility, agriculture or public services, whose contextual knowledge is essential for effective AI deployment. As such, discussions on education need to frame AI as an assistive tool, while emphasising agility in training systems. Finally, hybrid talent strategies enabling human–AI collaboration must be evolved. As teachers and trainers will typically lead the implementation of such strategies, substantial investments in building their capacities must be made.

4. India can help define global standards for AI deployment by leveraging its vast user base as a testbed for use-case innovation.

An advantage India enjoys is its ability to deploy AI solutions at an unmatched scale. Its young demographic, linguistic and sectoral diversity, and the presence of tech hubs that demand context-specific solutions, could allow it to become a living AI laboratory. This can address a potential bottleneck in global AI deployment: the power to develop scalable, context-sensitive use cases that can work in complex, resource-constrained environments. Knowledge-exchange mechanisms to connect practitioners with their counterparts across the Global South can accelerate this process by enabling countries facing similar challenges to co-develop and implement solutions rather than building from scratch. As such, India’s path to AI leadership should centre on scaling deployment across its domestic market, promoting cross-border learning corridors, and exporting proven solutions across the South in order to position itself as the world’s use-case capital, rather than competing in a frontier model race it need not win.

5. Taking inspiration from the Indian approach, open and interoperable DPI could become a foundation for building inclusive AI ecosystems.

Proprietary systems drive market concentration, risking the creation of a ‘digital underclass’ across swathes of the Global South. The wider adoption of India’s DPI model was strongly recommended as it could act as a counterweight by generating locally valuable data at population scale, while avoiding platform dependency. This data—buttressed by robust consent and data-sharing mechanisms—could prove valuable for building and training AI models. AI solutions developed in this manner would reflect local realities and nuances, boosting the accuracy of outputs. Importantly, access to a common, shared digital infrastructure would allow smaller AI firms to build AI systems on an equal footing as incumbents, creating a more level playing field, and fostering more inclusive innovation ecosystems in the Global South.

6. AI ambitions must be integrated with energy infrastructure planning, and efforts to promote energy and water security.

Discussions at pre-Summit events repeatedly emphasised the growing energy- and water-intensity of AI infrastructure, and the risks this poses to grid stability and water availability for local communities. Negotiating these energy requirements should be mainstreamed into efforts to advance the SDGs and existing climate commitments; and environmental costs must be factored into the present “compute-first” race, that often prioritises innovation without sufficiently considering carbon footprints and the region-specific climate vulnerabilities. Renewable energy sources and SMRs could be explored as power sources for AI infrastructure. Additionally, AI should strategically be incorporated into the management of the very systems it makes vulnerable. For example, AI-enabled predictive systems for grid management, water security, and natural disaster planning systems can enable self-sustaining and more resource-efficient data centre ecosystems. Planning standards for AI infrastructure should be conceptualised, and these should explicitly link compute expansion with energy and water security, disaster preparedness, and development goals.

7. Tech middle powers and Global South states should form coalitions for AI cooperation.

Across conversations, participants recommended that tech middle powers—such as India, Australia, Japan, Brazil, and South Africa—should forge partnerships with other Global South countries to promote AI development and governance. This would help generate valuable new streams of cooperation to set AI standards and frameworks, co-develop AI capabilities and applications, and make global supply chains more resilient. Today, the US tends to control the upper end of the AI value chain for advanced computing, while China largely controls access to the critical minerals needed for electronics manufacturing. Coalitions for AI cooperation are therefore seen as increasingly necessary to counter the potential limits that the United States and China could set on domestic innovation in other countries.

Siddharth Yadav is Fellow, Technology, ORF Middle East.

Elizabeth Heyes is Junior Fellow, Technology, ORF Middle East.

Anirban Sarma is Director, Centre for Digital Societies, ORF.

Annex

Key Discussion Points at Pre-Summit Events

Endnotes

^[a] The Annex to the report provides details of the main issues discussed at each event.

^[1] “India, 27 Other Nations, and EU Sign First International Declaration to Address AI Risks,” The Economic Times, November 3, 2023, https://timesofindia.indiatimes.com/business/international-business/india-27-other-nations-and-eu-sign-first-international-declaration-to-address-ai-risks/articleshow/104904367.cms

^[2] “Paris Action Summit 2025: A Turning Point in AI Governance amidst Global AI Divides,” JustAI, February 15, 2025, https://justai.in/paris-ai-action-summit-2025-a-turning-point-in-ai-governance-amidst-global-divides/

^[3] Jerry Fisayo-Bambi, “World Leaders and Tech Giants Converge in Paris for AI Summit,” EuroNews, February 10, 2025, https://www.euronews.com/next/2025/02/10/world-leaders-and-tech-giants-converge-in-paris-for-ai-summit

^[4] Press Information Bureau, Government of India, https://www.pib.gov.in/PressReleasePage.aspx?PRID=2209578&reg=3&lang=1

^[5] “Working Groups at the AI Impact Summit: Driving Global Collaboration across Seven Themes of AI Impact,” India AI Impact Summit 2026, https://impact.indiaai.gov.in/working-groups

^[6] “India Can be Global AI Use Case Capital: MeitY Secretary,” The Economic Times, January 13, 2026,  https://economictimes.indiatimes.com/tech/artificial-intelligence/india-can-be-global-ai-use-case-capital-meity-secretary/articleshow/126491043.cms?from=mdr