NextFin News - Speaking at the NDTV AI Summit in New Delhi on February 18, 2026, Jay Puri, Executive Vice President of Worldwide Field Operations at Nvidia, delivered a compelling case for the adoption of open-source AI models as the cornerstone of India’s digital future. Addressing a gathering of policymakers, tech leaders, and industry experts, Puri emphasized that for a nation with the scale and complexity of India, relying on closed, proprietary systems poses a risk to data sovereignty and long-term innovation. According to NDTV, Puri highlighted that open models provide the necessary transparency and flexibility for India to harness its own data—often referred to as the country's 'new oil'—to build customized solutions that reflect its unique linguistic and cultural landscape.
The timing of Puri’s remarks is significant, coming just weeks after the second inauguration of U.S. President Trump, whose administration has signaled a shift toward more protectionist technology policies and a focus on domestic manufacturing. In this evolving geopolitical climate, Puri’s advocacy for open models serves as a strategic roadmap for India to achieve technological self-reliance. By utilizing open-source frameworks like Llama or Nvidia’s own Nemotron, Indian developers can build upon existing global research while ensuring that the sensitive data used for training remains within national borders. This approach, Puri argued, is the only viable way to scale AI across India’s 1.4 billion citizens effectively.
From an analytical perspective, Puri’s emphasis on open models reflects a broader shift in the global AI value chain. For years, the industry was dominated by a 'black box' approach where a few Silicon Valley giants controlled the underlying weights and logic of the most powerful models. However, the rise of 'Sovereign AI'—a concept Nvidia has championed globally—suggests that nations must own their intelligence-producing capabilities. For India, the stakes are particularly high. With over 800 million internet users and a digital economy projected to reach $1 trillion by 2030, the volume of data generated is unparalleled. If this data is processed primarily through foreign proprietary APIs, India risks becoming a mere consumer of AI rather than a producer, losing out on the high-value intellectual property and economic rent associated with model ownership.
The economic implications of Puri’s '5-layer AI cake' framework, which he introduced during the summit, provide a structured look at how India can capture this value. The layers—comprising data, infrastructure, models, applications, and talent—must all be synchronized. By focusing on the 'model' layer through an open-source lens, India can bypass the prohibitive costs of developing foundational models from scratch, which can exceed $100 million in compute costs alone. Instead, Indian firms can fine-tune open models using local datasets, such as those in the 22 official languages of India, creating specialized tools for healthcare, agriculture, and governance that proprietary models often overlook due to lack of localized training data.
Furthermore, the push for open models aligns with the infrastructure investments currently being made by the Indian government and private sector. Under the IndiaAI Mission, the government has allocated approximately $1.2 billion to bolster compute capacity. Puri noted that Nvidia is working closely with Indian partners like Tata Communications and Reliance Industries to deploy tens of thousands of H100 and B200 GPUs. When this massive compute power is paired with open-source models, it democratizes access to AI, allowing startups in Bengaluru or Hyderabad to compete with global incumbents. This democratization is essential for maintaining a competitive market and preventing the monopolization of AI capabilities.
Looking ahead, the trend toward open-source dominance in India appears inevitable but not without challenges. While open models offer sovereignty, they require a highly skilled workforce to implement and maintain. Puri’s call to action underscores the need for a massive upskilling initiative. As U.S. President Trump’s administration potentially tightens H-1B visa regulations or shifts trade dynamics, India’s ability to retain its top-tier engineering talent to work on domestic open-source projects will be a critical factor. If India successfully integrates open models with its localized data and growing compute power, it will not only secure its own digital borders but also emerge as the primary exporter of AI solutions for the Global South, where similar needs for data sovereignty and linguistic localization exist.
Explore more exclusive insights at nextfin.ai.
