From Aadhaar to Sarvam: Where are we Indians at AI?

Yesterday, I wrote about China’s DeepSeek—a $50B AI juggernaut racing toward artificial general intelligence.

Here is the link, if you missed the article: What is DeepSeek?

But as I typed, my mind wandered to a tea stall in Mumbai, where a farmer checked his crop prices via a WhatsApp voice note.

That’s India’s AI moment……Boom!!

We built Aadhaar for 1.4 billion people at ₹100 per head. We vaccinated a billion with CoWIN, a platform now used from Jamaica to Iran. And we taught the world to pay with QR codes.

So why is India still an AI underdog?

Because the world measures AI in GPUs and trillion-parameter models, India measures it in voices understood, crops saved, and lives uplifted.

Here’s why that might redefine the game.

The Frugal Tech Legacy: Proof Scale Meets Simplicity

India’s genius lies in turning constraints into revolutions:

Aadhaar: Scaled to 1.4 billion people for less than the cost of one GPT-4 training run. Morocco and Indonesia now beg, “Build us an Aadhaar.”
UPI: Processes 12 billion transactions monthly—more than Visa + Mastercard combined. Kenya’s PesaLink and Nigeria’s NIP are its clones.
CoWIN: Vaccinated a nation of 1.4 billion in 18 months. Ukraine and Bangladesh begged for the code.

Yet, India’s $($ 150B) and the U.S. ($300B).

The gap? Chips, talent, and ambition.

Chips, talent, and ambition?

Let’s dive into each one of the factors one by one….!!!!

Chips? From Import Addiction to Design Disruption

India imports 90% of its semiconductors, relying on U.S. GPUs and Taiwanese fabs. But here’s the twist: 30% of the engineers designing those chips are Indian.

Startups like Mindgrove now craft RISC-V processors for rice fields, not data centers—at 1/10,000th of NVIDIA’s budget.

The lesson?

Skip the fab race and own the design for “India-fit” chips. These chips are low-cost, rugged, and built for monsoon soils, not metaverse fantasies.

AI Models? Skip the Arms Race, Own the Last Mile

While Silicon Valley obsesses over GPT-5,

Chennai’s Sarvam AI trains voice chatbots for WhatsApp, costing 1/10th per query.

Why?

Because 600 million Indians engage via voice notes, not text. Meanwhile, Jugalbandi, a government chatbot, explains welfare schemes in Mundari—a tribal language ignored by Google. India’s AI mantra: Build for the village, and the world will adapt.

Education? Fixing the Broken Pipeline

India produces 2.5 million STEM grads yearly but trails in AI talent.

Blame it on 85% of schools lacking computers and teachers who’ve never coded. Yet, startups like GUVI teach AI in Tamil and Hindi, while the government’s NEP 2020 nudges coding into classrooms.

The fix?

Treat AI like Aadhaar—a public good, not a privilege.

The UPI Playbook? Scale AI Like Chai for All

UPI’s magic wasn’t tech—it was trust. A tea seller with a QR code became India’s fintech mascot. For AI, the recipe is similar:

Voice-first interfaces for farmers who text with their thumbs.
Public data commons for crop patterns and disease outbreaks.
Regulatory sandboxes like RBI’s UPI experiment, now a $3T ecosystem.

The Global South’s Plea: “Lead Us”

At the G20, France’s Macron urged India to “democratize AI for the Global South.”

Why?

Because India’s chaos—22 languages, 700 million rural users, 100 dialects—is a testing ground for humanity. If an AI model works in Odisha’s tribal hamlets, it’ll work in Senegal’s villages.

The AI Manifesto

Sam Altman, once called India’s AI efforts “hopeless.”

Last week, Sarvam AI raised $41 million to build voice LLMs for 500 million Indians. The world laughs until India delivers.

As Nandan Nilekani, architect of Aadhaar and Sarvam’s backer, says:
“We don’t need to chase GPT-5. We need AI that knows ‘mitti’ (soil), ‘bimaar’ (sickness), and ‘bhasha’ (language). Solve for Bharat, and the planet follows.”

Call to Action: Rewrite the Rules

Policymakers: Launch “AI for Bharat” districts—tax-free zones for agri, health, and education AI.
Investors: Fund startups solving for ₹100/month users, not Silicon Valley clones.
Builders: Embrace constraints. A farmer’s voice note is worth a thousand GPT-4 prompts.

Why This Matters?

From Davos to Delhi’s tech meetups, everyone debates India’s AI potential.

This isn’t just about catching up—it’s about proving that frugal, inclusive innovation can outpace trillion-dollar giants.

The chai stall’s QR code was the start. The voice chatbot is next.

Please follow and like us:

Table of Contents