India should avoid building its artificial intelligence capabilities on systems it neither controls nor regulates, Vishal Sikka said on Friday, urging the country to steadily develop its own AI stack.
Speaking at the 70th Foundation Day of the All India Management Association (AIMA), the founder and CEO of Vianai Systems and former CEO and Managing Director of Infosys said India is large and capable enough to build its own AI technologies over time.
He said there is no harm in using components developed elsewhere but cautioned against depending on AI systems that India does not fully understand or over which it has no regulatory rights.
Sikka noted that AI development is currently concentrated in the hands of a small global community. Around 250,000 people operate AI systems worldwide, about 5,000 build foundation models, and roughly 1,000 work in frontier labs, he said, adding that AI skills can be learned more widely.
On computing infrastructure, he said India must push for access to AI hardware and supported the goal of having 200,000 GPUs by the end of the year. He welcomed Micron Technology’s plan to set up a memory and chip assembly facility in Gujarat, but said India should also look ahead to future technologies rather than only replicate existing models.
Sikka also flagged AI’s high energy consumption, calling it a power-intensive technology. To support AI at scale, he said India would need to invest in energy capacity, including nuclear technology and nuclear fission research.
On robotics, he said it often takes years for technologies to become mainstream. He cited Waymo’s robo taxis, which emerged years after the autonomous vehicle challenge by DARPA. He added that such systems still rely on human monitoring and infrastructure support.
Addressing concerns over jobs, Sikka said AI can significantly improve productivity, including in software development, but the impact on employment will depend on how leaders manage the transition. He also said AI developers must be held accountable for their systems and users should apply judgment rather than rely on AI blindly.
On the use of AI in warfare, he cautioned against handing over critical systems to AI agents operating at high speed and scale.