Mumbai, Dec 3 (NationPress) It is imperative for India to establish its own sovereign AI capabilities to avoid reliance on foreign tech companies that leverage Indian data to enhance their models, stated Amitabh Kant, the former CEO of Niti Aayog.

At the Mint All About AI Tech4Good Awards, he emphasized that the future of India’s technology landscape will hinge on how swiftly it develops its local computing infrastructure, safeguards sensitive data, and nurtures startups focused on creating homegrown AI solutions.

“No other technology in recent history has captivated the globe like artificial intelligence,” Kant remarked. He described this moment as a unique blend of economic opportunity, national security challenges, and demographic potential, asserting that India's status as a leading producer of digital data offers a significant chance to develop AI systems tailored to local needs.

Kant cautioned that although India’s digital public infrastructure has facilitated widespread inclusion, the country’s insufficient computing power threatens to hinder its advancement. He highlighted the recent collaboration between OpenAI and Nvidia to generate 10 gigawatts of GPU capacity, roughly equivalent to five million of the latest high-performance processors, in stark contrast to India's approximately 30,000 GPUs.

“Addressing this disparity necessitates substantial investments from the private sector and foreign direct investment. Currently, OpenAI's ChatGPT in India boasts more monthly active users than any other nation, exceeding the United States by around 33%,” he noted.

Kant outlined three reasons why sovereign capability is crucial: Firstly, self-sufficiency would foster a robust domestic startup ecosystem, attracting investments into AI hardware and software. Secondly, models tailored to India’s languages, cultural nuances, and public sector requirements would ensure AI serves every citizen. Lastly, a secure domestic infrastructure is vital for national security as AI systems become increasingly integrated into public services.

He pointed out that global companies are offering free or low-cost services while utilizing this data to train closed-source models at scale. “Our future might involve AI services powered by our data but owned by others, sold back to us,” he warned.

Kant advocated for a regulatory framework that permits global models to function in India but mandates that they operate on infrastructure located within the country. No user data, he insisted, should exit India for applications involving large language models. This strategy, he contended, would draw investment into computing capacity and enhance privacy safeguards.

India's emerging AI startups, such as Sarvam AI, Sokhat AI, Dhani AI, and Gantt AI, are starting to develop foundational models but require access to ample high-quality data and robust computing power to achieve global standards. National initiatives aimed at attracting top AI researchers, improving access to anonymized datasets, and expanding public computing marketplaces will be essential.

Kant cited global instances where countries that entered late into a technological revolution managed to leapfrog established players. “History illustrates that in technology, those who arrive second can sometimes advance quicker, learn smarter, and sidestep earlier mistakes,” he mentioned, referencing how Google surpassed initial search engines and how later entrants reshaped hardware markets.