Sarvam AI’s CEO Pratyush Kumar cautions that heavy reliance on foreign AI infrastructure could create a digital colony. With support from leaders like Zoho’s Sridhar Vembu, the call for building indigenous, full-stack AI is intensifying. Investors should monitor how this shift affects capital spending and research costs for Indian tech firms.
Pratyush Kumar, co-founder and CEO of Sarvam AI, has issued a strong warning about the risks of India’s over-dependence on foreign technology providers for its artificial intelligence infrastructure. Speaking at the Zoholics conference, he argued that relying on foreign-owned systems puts the country in a position similar to a digital colony, where core value and data control remain outside local hands. Instead of building applications on top of foreign frameworks, Kumar advocates for Indian companies to develop their own full-stack AI infrastructure.
Developing a full-stack approach means the company owns everything from the underlying foundational models to the final application interfaces and the inference layer. This approach offers companies deeper control over their data pipeline and operations. While the initial investment in this strategy is significant, Sarvam AI suggests it provides a long-term competitive advantage by reducing dependence on external service providers and capturing more value within the local ecosystem.
Zoho founder Sridhar Vembu, who also participated in the discussion, expanded the argument beyond AI. He highlighted the need for India to build indigenous capabilities in high-tech fields like semiconductors, robotics, and precision engineering. Vembu noted that innovation is often driven by intellectual capital rather than just large budgets, and drew comparisons to the success of the Unified Payments Interface (UPI). The logic is that if Indian companies can build and commoditize their own AI software, it could lower costs for services that currently face high fees from global tech giants.
For investors, the push toward sovereign AI brings both opportunities and risks. Most large Indian IT services companies currently function by integrating global AI tools into client solutions. Transitioning toward building proprietary models—or what is being termed as sovereign AI—requires heavy spending on graphical processing units, data centers, and specialized R&D talent. This shift to full-stack development typically involves significant capital expenditure, which could put pressure on short-term profit margins and cash flow. Furthermore, the success of such models will depend on their ability to match the performance and speed of established global platforms.
Investors monitoring the sector should track how Indian tech companies balance these R&D investments with client demand. The government’s own India AI Mission, which aims to provide infrastructure support, will be an important factor in how this sector evolves. The central question for shareholders will be whether these companies can monetize indigenous AI effectively enough to justify the higher costs of developing their own technical stacks, or if the market will continue to favor established, ready-to-use foreign technologies.