NextFin News - Speaking at a premier AI Summit in New Delhi this Tuesday, February 17, 2026, Amitabh Kant, India’s G20 Sherpa and former CEO of NITI Aayog, delivered a provocative assessment of the global artificial intelligence landscape. Kant revealed that India has surpassed the United States as the largest contributor of data to OpenAI’s ecosystem, yet remains largely dependent on foreign proprietary models. Addressing a gathering of tech leaders, policymakers, and venture capitalists, Kant argued that India’s current role as a primary data provider without equivalent ownership of foundational technology is unsustainable. He called for a radical shift toward building homegrown Large Language Models (LLMs) to ensure India’s digital sovereignty and economic security in an era of increasing technological protectionism.
According to Moneycontrol, Kant emphasized that the sheer volume of Indian user interactions, linguistic diversity, and behavioral data is currently fueling the refinement of models like GPT-5 and its successors, primarily benefiting Silicon Valley shareholders. The timing of this critique is significant. As U.S. President Trump pursues a renewed 'America First' agenda in 2026, focusing on domesticating high-tech supply chains and restricting the export of advanced AI compute, India finds itself at a precarious juncture. The reliance on American infrastructure for AI services—while providing the very data that trains those services—creates a lopsided value exchange that Kant believes must be corrected through domestic innovation.
The data disparity Kant highlighted is backed by staggering metrics. With over 900 million internet users and the world’s cheapest data rates, India generates a digital footprint that is unparalleled in its complexity. However, the economic value of this data is currently being captured upstream. In the traditional 'Data-Information-Knowledge-Wisdom' hierarchy, India is providing the raw 'Data' and 'Information' layers, while U.S.-based firms are extracting the 'Knowledge' and 'Wisdom' through sophisticated algorithmic processing. This creates a 'digital colonial' framework where the value-added intellectual property resides outside the borders of the data-originating nation.
The push for homegrown models is not merely a matter of national pride but a technical necessity. Current Western-centric models often struggle with the nuances of India’s 22 official languages and thousands of dialects. A model trained predominantly on Western datasets frequently exhibits cultural biases or fails to grasp the socio-economic context of the Global South. By developing indigenous models, India can create AI tools tailored for local governance, agriculture, and healthcare—sectors where generic models often fall short. Kant’s vision aligns with the 'India Stack' philosophy, suggesting that AI should be treated as a public digital good rather than a closed-loop corporate monopoly.
Furthermore, the geopolitical climate under U.S. President Trump has introduced new variables into the AI equation. The 2025-2026 period has seen a tightening of GPU export controls and a more transactional approach to international tech partnerships. If the U.S. administration decides to prioritize domestic AI capacity or limit API access to foreign entities under the guise of national security, India’s digital economy could face sudden disruptions. Kant’s advocacy for sovereign AI is a preemptive strike against such vulnerabilities, urging the Indian private sector to move beyond application-layer innovation and dive into the more capital-intensive foundational layer.
Looking ahead, the transition to homegrown models will require a massive infusion of compute power and specialized talent. While India has the software expertise, the hardware bottleneck remains a challenge. The Indian government’s AI Mission, which includes plans for a 10,000-GPU supercomputing facility, is a step in the right direction, but Kant suggests that private capital must play a larger role. We expect to see a surge in 'Sovereign AI' startups in India throughout 2026, supported by local data centers and government-backed datasets. The goal is no longer just to be the 'back office' of the world, but to become the 'brain' of the global digital economy, leveraging the very data that Kant notes is currently being exported for free.
Explore more exclusive insights at nextfin.ai.
