Where is India in the AI Race?

Every week a new model drops from San Francisco or Hangzhou, and the question comes up in every Indian group chat: where is India in the AI race? The honest answer is more interesting than "behind" or "ahead". India is the world's largest AI user market, a rising AI builder, and still an AI infrastructure underdog.

How deep does AI go in India?

Deeper than most people assume. NASSCOM estimates India has over 420,000 AI/ML professionals, the second-largest talent pool after the US. Stanford's AI Index has repeatedly ranked India first in "AI skill penetration" — the share of workers listing AI skills relative to other countries.

The IndiaAI Mission (₹10,372 crore, 2024) is funding 18,000+ GPUs for a national compute facility, Indic datasets via AIKosh, and a foundation-model programme. Bhashini, the government's language-AI platform, translates across 22 scheduled languages and powers voice-based services for citizens who never typed a query in English.

How Indians actually use AI

• Payments: UPI's fraud detection screens billions of transactions with ML models.
• Agriculture: Apps like Kisan e-Mitra and satellite-based crop advisories reach millions of farmers.
• Healthcare: AI-read chest X-rays for tuberculosis screening (Qure.ai) are deployed in public health programmes.
• Everyday: India is ChatGPT's second-largest user base; students, creators and small businesses are the heaviest adopters.
• Enterprise: TCS, Infosys, Wipro and HCL together employ hundreds of thousands of engineers deploying generative AI for global clients.

The uncomfortable part: foreign dominance

The frontier layer of AI — the biggest models and the chips they run on — is dominated by US companies (OpenAI, Google, Anthropic, Meta, NVIDIA) with China (DeepSeek, Alibaba, Baidu) as the only serious challenger. India:

• Has no domestic GPU manufacturer; compute is imported and expensive.
• Contributes a small fraction of global AI research papers compared with the US and China.
• Sees most of its AI talent employed by, or migrating to, foreign firms.
• Runs a real risk of "algorithmic dependence": Indian languages, culture and law interpreted through models trained elsewhere.

That last point is why sovereign AI has become a national-security conversation, not just a tech one.

Indian companies making real strides

• Sarvam AI — building Indic-first LLMs (Sarvam-1, Sarvam-M) and voice agents; selected under IndiaAI to build a sovereign foundation model.
• Krutrim (Ola) — India's first AI unicorn, with multilingual models and plans for indigenous AI chips.
• BharatGen — a government-backed consortium of IITs building multimodal Indic models (Param).
• Zoho — quietly trained its own LLMs in Chennai, proving bootstrapped AI is possible.
• Tata Electronics & the semiconductor push — fabs in Dholera and Assam are the first step toward chips made in India.
• Vertical players — Qure.ai (radiology), Niramai (breast-cancer screening), CropIn (agri), Yellow.ai and Gupshup (conversational AI) are winning globally in their niches.

So, where is India in the race?

If the AI race is a marathon with three legs — talent, adoption, and infrastructure — India is leading the first, sprinting in the second, and only just lacing up for the third. The difference between 2047's India as an AI superpower or an AI colony will be decided by compute, chips, and whether the best Indian researchers build here.