Sovereignty cannot be taken to mean solitude

This article is authored by Somjit Amrit, CEO, IIT Mandi iHUB and HCI Foundation.

With the conclusion of the India AI Summit 2026, conversations around sovereign AI and its critical role in today’s geopolitical landscape are more relevant than ever. Artificial Intelligence (AI) is no longer just about innovation; it is about capability, resilience, and long-term national strength. The need to build AI systems that are both globally engaged and strategically secure is becoming increasingly clear. This conversation becomes even more important as AI itself evolves, particularly with the rapid rise of multimodal systems that integrate text, vision, audio, and beyond.

AI Summit 2026 (HT photo)

We at IIT Mandi TIH are stepping into the era of Multimodal AI with the establishment of the Multimodal AI Lab. An era where text understands images, audio interacts with vision, and intelligence flows across multiple streams of data at once. AI is no longer single-dimensional. It is layered. Connected. Interdependent.

For a nation or a research lab, building AI today is not about shutting doors. It is about choosing what to open, what to protect, and how to contribute without losing identity.

Then does sovereign AI mean being in solitude? Being secluded? Living in isolation?

As we build sovereign AI, we need to clarify the openness of technology. The four points which need to be clarified and adopted per their merit:

• Open source:
  Making training code, model architectures and evaluation pipelines transparent.
  This enables reproducibility, third party audits, bias detection and security validation.
• Open weights:
  Sharing trained model parameters while safeguarding sensitive datasets and proprietary training strategies.
  This allows local fine tuning and deployment without exporting critical data.
• Open standards:
  Defining common APIs and interoperability protocols. This prevents vendor lock in and ensures systems can communicate across borders and platforms.
• Open infrastructure:
  Owning compute capacity, GPU clusters and national research platforms. Without infrastructure control, model independence remains fragile.

Recent research reinforces this approach. A November 2025 study by Frank Nagle (MIT) and Daniel Yue (Georgia Tech), titled The Latent Role of Open Models in the AI Economy, finds that open models are rapidly narrowing the performance gap with proprietary systems while remaining significantly less expensive to deploy. The authors estimate that broader adoption of open models could unlock nearly $24.8 billion in annual consumer value. In short, openness is not only about trust and control, it is a source of economic strength.

The lab by the Technology Innovation Hub (TIH), at the IIT Mandi campus, supported by DST under NM-ICPS, would play a catalytic role in strengthening these foundations. Initiatives such as the IndiaAI Mission and the Startup India Seed Fund Scheme illustrate how targeted support can nurture deep-technology innovation. With this support, the lab can develop high-quality datasets, reliable compute infrastructure, and robust development pipelines. Government-supported infrastructure and funding mechanisms ensure that AI capability remains broad-based, inclusive, and strategically aligned with India’s long-term vision.

Sovereign AI is not about closing systems; it is about designing them with intention. It means sharing what builds trust, protecting what secures autonomy, shaping the standards that define the ecosystem, and investing in the infrastructure that sustains it. In a connected world, strength does not come from standing alone. It comes from participating with clarity, control, and confidence. Sovereignty cannot be solitude. It must be strategic and collaborative.

At the TIH, IIT Mandi, we are committed to advancing this frontier. A state-of-the-art multimodal AI research facility (MI-RA (Multimodal Intelligence-Real-World Applications) is being established to lead research in audio-visual intelligence, spatial computing, and next-generation AI systems. The focus is not only on building advanced models, but on creating strong foundational capabilities such as high-quality datasets, reliable compute infrastructure, and robust development pipelines.

The lab will operate on the principles of openness and shared capability described above. The aim is to create an environment where innovation is not restricted to a few teams, but can extend to startups, collaborators, and the broader research community. More importantly, the technologies and capabilities developed here are intended to contribute to India’s AI ecosystem at large by supporting indigenous innovation, strengthening national capacity, and enabling solutions that address real-world Indian needs.

This article is authored by Somjit Amrit, CEO, IIT Mandi iHUB and HCI Foundation.