New Delhi’s India AI Impact Summit 2026 kicked off with a stark warning from former NITI Aayog CEO Amitabh Kant. Speaking at a high-profile panel, Kant urged India and other developing nations to develop their own Large Language Models (LLMs) using local data. He highlighted how Big Tech giants are harvesting data from the Global South to train their AI systems, leaving these regions at a disadvantage.
Kant pointed out a glaring statistic: India supplies 33% more data than the United States for LLM training. Yet, the benefits flow disproportionately to Western corporations. ‘Global South data is fueling LLM advancements, but Big Tech could turn this into proprietary models sold back at premium prices,’ he cautioned.
To counter this, Kant advocated for sovereign AI development. Countries must leverage their unique datasets to create inclusive models that address local needs. He stressed that AI should be affordable, accountable, and multilingual, empowering billions in the Global South.
The rapid pace of AI innovation risks exacerbating inequalities, Kant warned. Without deliberate design, AI could widen the gap between haves and have-nots. He called for AI applications in education, healthcare, and nutrition to uplift those below the poverty line.
‘Can we ensure AI reaches the poorest? Can it transform lives in the Global South?’ Kant posed these critical questions. He envisioned AI making the impossible possible, from personalized learning to better health outcomes. But only if developing nations seize control of their data destiny.
As investments pour into AI, the choice is clear: build local LLMs or risk permanent marginalization. Kant’s message resonated strongly, signaling a new era of digital sovereignty for the developing world.