Sovereign silicon, patient capital key to India’s AI ambition

Bengaluru: India’s bid to become an AI powerhouse will depend less on chatbots and more on control over the compute stack, industry leaders said at the IESA Vision Summit, underscoring the need for sovereign silicon and patient capital.India Electronics and Semiconductor Association (IESA) president Ashok Chandak said India stands at a “consequential juncture” as “the global landscape is getting redefined.” “India is not just a participant, but India is playing a key role by becoming an architect of the future,” he said in Bengaluru on Wednesday. He added that the summit is framed around the theme of India as a “product nation, production nation and a skilled nation.

”

Senior executives from MediaTek, NXP Semiconductors, Cyient Semiconductors and domestic startups argued that access to compute — not just power — will define the next phase of the AI race.Dheemanth Nagaraj, co-founder and CEO of Agrani Labs, said that while India’s data centre capacity is projected to scale from roughly 1.2–1.3 gigawatts to 8–9 gigawatts over five years, “access to compute” remains the bigger constraint.

Building world-class AI infrastructure, he said, requires a full-stack approach spanning silicon, interconnects, thermals, cloud integration and software frameworks. “It’s a 10-year view,” he said, urging long-term capital and strategic partnerships.Akshay Aggarwal, director of technology at MediaTek, said next-generation mobile chipsets are increasingly built on heterogeneous architectures that combine CPUs, GPUs and NPUs (neural processing units) to manage tight power and latency budgets.

Looking ahead to 6G, he said work is underway on “AI for wireless and wireless for AI,” including a “device cloud” model where nearby devices share compute resources.Suman Narayan, CEO of Cyient Semiconductors, pointed to the growing energy burden of AI infrastructure. US data centres consume about 176 terawatt hours of power — roughly 4.4% of total US electricity usage — he said, adding that efficiency gains in power conversion and higher-voltage architectures could significantly cut losses and operating costs.Nitin Rajmohan, co-founder of Akeana, said India must accelerate domestic silicon development. “Sovereignty is not an option, it’s a necessity,” he said, calling for dozens of Indian AI chip companies to tape out silicon rapidly and iterate.Hitesh Garg, vice-president and India country manager at NXP Semiconductors, said safety and security must be embedded as AI systems expand into automotive and industrial applications, warning that trust failures could be “catastrophic.