Now is the time for India to build its AI
India must immediately train its own AI models, leveraging offshore computing environments such as Singapore while rapidly expanding domestic AI infrastructure
Artificial Intelligence (AI) is reshaping economies and societies at an unprecedented pace. India faces an urgent question: Will we take control of our AI future, or will foreign-built models shape our economy, languages, and perception of history?
The US and China are investing billions of dollars in AI, entrenching their dominance. Delay risks India becoming permanently dependent on foreign AI models.
India must immediately train its own AI models, leveraging offshore computing environments such as Singapore while rapidly expanding domestic AI infrastructure. We cannot wait for AI supercomputers to be built in India; we must act now with available resources and scale as we build our own AI capabilities. This is not just a technological ambition; it is an economic necessity, a cultural imperative, and a national security priority.
India’s intellectual heritage spans Sanskrit mathematical treatises, Tamil Sangam literature, and Buddhist scriptures, among others. Yet, much of this knowledge remains inaccessible to AI models predominantly trained on Western and Chinese datasets. If AI is trained only on English and Chinese sources, India’s vast historical and literary wealth risks digital erasure.
Many of India’s languages share common grammar structures, syntactic rules, and phonetic patterns, allowing AI models trained in one to scale efficiently across others. This linguistic and cultural interconnectedness enables even less widely spoken languages to be trained at the same fidelity as dominant ones. These synergies allow India to build scalable AI models faster than many multilingual nations — but only if we act now.
If India does not invest in its own AI, foreign-built systems will determine what knowledge is preserved. Western AI models misinterpret Indian historical and legal concepts through a colonial or modern Western lens. Chinese AI models falsely claim Arunachal Pradesh as Chinese territory. Without AI grounded in Indian history and values, we risk digital colonisation.
The next five years will determine AI superpowers. Those who move swiftly will shape the AI-driven economy. India stands at a crossroads — seize control of AI now or become an afterthought in the global AI order. AI could add $500 billion to India’s economy by 2030, but only if it serves all Indians. Nowhere is this more urgent than in agriculture, employing over 150 million people. AI can transform farming by making weather forecasts, pest control alerts, and government subsidies accessible in Bhojpuri, Kannada, and Punjabi.
A similar challenge exists in India’s 60 million small and medium enterprises (SMEs). AI tools can revolutionise local businesses but must function in the languages entrepreneurs speak.
This urgency extends to education and governance. If AI remains trapped in a handful of languages, it will fail to uplift underserved communities, locking them out of the AI-driven economy. Students in rural India should have AI tutors in their mother tongue. Citizens should interact with government AI systems in their language, not bureaucratic jargon.
Some suggest India should focus on small AI models for lesser-resourced languages, citing cost and practicality. This approach is shortsighted. While small models are cheaper and easier to fine-tune, over-reliance on them risks a two-tier AI system — well-resourced languages receiving advanced AI while others get sub-par tools.
The belief that large models are unfeasible for Indian languages is a self-fulfilling fallacy. India must reject the narrative that our languages lack data for cutting-edge AI. Instead of accepting scarcity as a limitation, we must actively address it — expanding and curating datasets for robust AI in every Indian language. India has consistently overcome technological constraints, it must be no different in AI.
India’s AI ecosystem does not require immediate billion-dollar investments. The first phase — training foundational models using offshore compute infrastructure — can be achieved with an initial investment of tens of millions of dollars. This provides immediate traction, allowing India to deploy AI models while building domestic supercomputing capacity in parallel. As these models generate economic value, the second phase can be self-funded through returns and government-backed AI adoption. A simple projection: An initial $50 million investment enabling AI-driven efficiencies across agriculture, business, and education could drive $500 million in GDP-impact within two years. Reinvesting even a fraction of this return will sustain and scale the next phase, creating a compounding cycle of AI-driven economic growth.
Acting now secures India’s place in the AI-driven future. If India does not immediately begin training its own AI models, it will forever be an AI consumer, not an AI creator. The world is looking for India to lead the way in responsible and sustainable AI.
Arun Subramaniyan is founder & CEO of Articul8 AI. The views expressed are personal
