After the Summit, the Hard Work Begins

While the India AI Impact Summit signalled ambition, building credible AI capacity will be harder.

For decades, India’s economic rise was measured in visible achievements in the form of highways laid, ports expanded and software exported to the world. But last week at Bharat Mandapam in New Delhi, the metric of national power shifted during the India AI Impact Summit 2026 - the first such global AI summit led by a nation of the Global South. Which saw delegations from over 100 countries. The summit was about the invisible infrastructure of the future and the strategic question of who controls the intelligence that will shape the 21st century.

Beneath all the diplomatic language and applause, the clear message that emanated from the summit was that India does not want to remain the world’s back office in the AI era. It wants to design, build and govern systems of intelligence.

The New Geopolitics

In the industrial age, sovereignty rested on steel output, oil access and manufacturing depth. In the digital age, it increasingly rests on compute. Control over large-scale computing power determines who trains frontier models, who sets technical standards and who must adapt to rules written elsewhere.

That is why the government’s announcement at the India AI Impact Summit to add 20,000 GPUs to an existing base of roughly 38,000 was not a routine capacity upgrade. It reflected strategic necessity. Countries without domestic compute are compelled to rent it abroad, usually from American hyperscalers, along with the models, safety norms and commercial constraints that accompany them.

This dependence already shapes outcomes. Indian startups training advanced models routinely host data overseas. Public-sector pilots rely on proprietary APIs governed by foreign compliance regimes. Even defence and health research increasingly runs on cloud infrastructure beyond India’s jurisdiction. Compute scarcity, not talent, has become the binding constraint.

Artificial intelligence is not neutral infrastructure. Systems trained primarily on Western data struggle with Indian jurisprudence, informal labour markets or multilingual governance.

India’s emphasis on multilingual large language models for governance, courts and welfare delivery reflects a recognition that scale alone is not enough. Local relevance matters.

If AI becomes the operating system of modern economies, sovereignty in the AI age requires at least partial control over that operating system.

The Implementation Gap

Yet, ambition does not translate to not capacity. And the summit revealed just how wide the gap remains between declaration and delivery. Cybersecurity incidents, including phishing attempts targeting participants, were treated as side stories. Weak security at the showcase stage raises questions about readiness at scale.

Then, there was the Galgotias university contretemps which involved a Chinese-manufactured robot dog that was presented as domestic innovation. Such embarrassing incidents expose a familiar weakness of India’s tendency to privilege optics over verification. In emerging technologies, such shortcuts are costly. Mislabelled demonstrations undermine credibility with investors, researchers and partners who care less about slogans than supply chains.

India has moved fast in narrative construction. But institutional capacity has lagged and AI does not tolerate that imbalance. It cannot be built through announcements or symbolic launches. It requires patient capital, research depth, disciplined procurement and independent auditing mechanisms that resist political pressure.

Digital Public Infrastructure is often cited as India’s comparative advantage, and with reason. Aadhaar and UPI demonstrated that population-scale systems can be built and governed. But AI systems are categorically different. They are probabilistic, opaque and adaptive. They demand continuous evaluation, bias testing, data stewardship and energy planning. A payments rail does not hallucinate. But a language model can.

Demographic Crossroads

The AI push also complicates India’s economic narrative. For decades, growth relied on labour abundance. IT services, back-office processing and global outsourcing thrived on scale, English proficiency and wage differentials.

AI changes the equation. Automation raises productivity without proportionate employment. Tasks once distributed across thousands of people like legal review and customer support are increasingly handled by models running on clusters of servers. The immediate impact will be uneven, but the long-term risk is a gradual decoupling of growth from job creation.

Education reform cannot stop at expanding seats or issuing degrees. It must prioritise computational literacy, domain expertise and adaptability. AI literacy must extend beyond engineering colleges to law schools, medical institutes, agricultural universities and civil-service academies. Governance in an AI-rich state requires administrators who understand systems they regulate.

Between Washington and Beijing

India’s AI strategy unfolds amid sharpening global rivalry. The United States relies on private capital, platform dominance and export controls to maintain its lead. China combines state coordination, manufacturing control and data centralisation to accelerate deployment.

India fits neither model. It lacks America’s venture depth and China’s supply-chain integration. Its stated approach of state-enabled infrastructure with private innovation and regulatory oversight aims for balance. Whether that balance can hold remains uncertain.

True autonomy will depend on more than GPU counts. It will require reliable power grids, access to advanced semiconductors, sustained funding for basic research and partnerships that survive geopolitical pressure.

The summit marked a declaration of intent. But the harder task of aligning universities with frontier research, scaling indigenous models beyond pilots, enforcing data-protection rules that withstand legal scrutiny and building institutions capable of auditing systems, lies ahead.

If India wants to move from being a global service provider to a system architect, it must choose research over rhetoric. That means investing where returns will not be immediate, strengthening cybersecurity before expanding scale and ensuring that AI’s benefits extend beyond a few metropolitan clusters.

For now, India has announced its entry into the AI race. But in the algorithmic age, credibility is not measured by summits but by systems that work and citizens who benefit.

(The author is a strategy and transformation leader who writes extensively on technology and future of work)