Yotta to Adani: India building sovereign, frontier AI with Global South relevance

India has spent three decades as the world’s software workshop, we all know that by this point. But if key voices at the India AI Impact Summit 2026 are any indication, the next decade could see India become the architect of sovereign, frontier artificial intelligence for itself – and for the Global South.

On stage, industry leaders weren’t merely discussing model sizes or GPU clusters. They were sketching out a distinctly Indian blueprint for AI based on local compute, rooted in local context, and deployed at population scale. The goal is not just technological self-reliance, but relevance – especially for billions of users across emerging economies.

Sunil Gupta, Co-founder and CEO of Yotta Data Services, framed the opportunity quite starkly. “While India is creating and consuming 20% of the world’s data, how much of it is actually hosted in India? Just 3%, and that shows the scale of the problem, but also the opportunity.” The imbalance between data generation and local compute capacity, he argued, is the defining infrastructure challenge of India’s AI moment.

That challenge, however, is rapidly turning into an investment cycle. Over the past seven years, India’s data centre capacity has multiplied several times. But Gupta’s message was nuanced, that infrastructure isn’t just about catching up, that it’s ultimately about sovereignty. “Unless India has got its own digital infrastructure, which is in terms of compute, India cannot do that. India cannot be just sitting on imported technology… it is the right time for India to create its own infrastructure,” Gupta emphasized.

Also read: Microsoft-Yotta to power IndiaAI Mission, where Shakti Cloud meets Azure AI

This isn’t infrastructure for its own sake, but  infrastructure that will define what Indian AI looks like, Gupta explained further. “Once we do that, we will be defining what type of model we create… what type of data sets our model will take… which use cases Indians prefer – in agriculture, in healthcare, in education, in space.” In Gupta’s telling, sovereign compute is all about shaping AI that solves for mass-scale development rather than niche enterprise value.

That development-first framing is central to India’s pitch to the Global South. “India is talking about impact. Instead of AI filling the pockets of deep tech companies, India is talking about AI actually serving the needs of billions of people… and that is how India can take its use cases to the whole world.” The ambition is clear: build AI not for the top 1% of global users, but for the next five billion.

At #IndiaAIImpactSummit2026, discussions are focused on how India can build frontier AI that strengthens national capability while delivering meaningful impact for the Global South.
The panel explored:

– How regulatory sandboxes like GIFT IFSC can enable responsible AI… pic.twitter.com/g5rRGlFUll

— nasscom (@nasscom) February 16, 2026

Rangarajan V, Senior VP and CDO at Adani Defence and Aerospace, extended the sovereignty argument into language, security, and context. For him, India’s need for local AI models is as much ideological as it is practical. “You need local models handled by local people in order to be truly relevant for the full population. You can get to 60–70% relevance with base languages, but transforming the remaining 30% means serving a population larger than the entire US.”

That last 30% – dialects, local nuances, cultural context – is where imported models often fall short. In defence and security applications, those gaps can be critical, argues Rangarajan. “You cannot expect somebody sitting in the US to appreciate the nuances we need to understand as Indians to protect the security of our borders,” Rangarajan said, pointing to the need for AI systems trained on local linguistic and strategic realities.

Also read: Nvidia eyes India as AI export hub, partners with Reliance and other Indian IT giants

Data sovereignty also plays a central role. With sensitive data often hosted abroad, control becomes a strategic issue. “When it is on-premise on our own soil, we have better control over who accesses the data. If we are building for the Global South, if we are building for the world, then it is truly ours.” Rangarajan’s proposal for an “India Context Protocol” – an ICP for AI – reflects a growing belief that contextual intelligence will be as important as model scale.

Yet sovereign AI isn’t just about building models or data centres. It’s about deployment at India-scale. That’s where Qualcomm India’s Sahil Arora sees the next frontier. “AI has to be served, has to be democratized, it has to go in people’s hands. Smaller models on device will bring latency and connectivity requirements down,” said Arora.

In India, deployment rarely follows Western enterprise-first patterns. Instead, it moves through devices – smartphones, vehicles, and embedded systems – reaching users directly. “With GPU and NPU coming into devices… from mobile to cars to even a sound box – that’s where real deployment happens… hybrid architecture is what will drive India-scale deployment.”

The convergence of sovereign infrastructure, localised models, and device-level deployment is what could define India’s AI trajectory. Or, as Gupta put it in policy terms, “Compute infrastructure for AI has to be treated as an essential commodity.” 

This means digital highways – data centres, networks, and GPU clusters – may soon be viewed as critical as physical ones. If that vision holds, India’s AI journey will be less about catching up with Silicon Valley and more about charting a parallel path.

Also read: Global AI commons: India’s most ambitious tech diplomacy pitch yet