Why the next wave of AI toys will be made in India
India learned to make the toy but still imports the brain. Language, price discipline and Bengaluru engineering say the AI layer gets built here next.
For most of its history, India's toy industry was an importing industry. Most of what sat on an Indian shelf came from abroad, and most of that came from China. Then policy got serious. Make in India put toys on the priority list. Quality control orders raised the floor on what could legally enter the market. The government's toy-cluster support scheme put ₹3,489 crore behind molds, tooling and training, by public reporting. A decade of that, and IBEF's industry profile now describes a country that exports more toys than it imports. The reversal was earned, not announced.
The market underneath kept growing. India's toy market sits somewhere between $1.5 and $2 billion today, with projections that it roughly doubles by the early 2030s. Growth is not the question anymore. The question is which layers of the toy India will make. The most valuable layer is still on a boat.
The shell is Indian. The brain is not.
Walk a toy cluster today and the progress is physical. An OEM can mold the body in Greater Noida. Stitch the plush in Karur. Turn the wood in Channapatna. Print the packaging next door. Then the spec sheet says 'talks to the child', and the purchase order goes overseas. The voice chip is imported. The model behind the voice is imported. The parent app is rebadged from someone else's cloud. The smart layer of a smart toy, the layer with the margin in it, arrives in a carton.
The gap is widening in money terms, not narrowing. India's smart-toy segment is projected to grow faster than the market around it, and global smart-toy forecasts tell the same story at larger scale. When Spielwarenmesse, the world's largest toy fair, named its trends of the year, one of them was 'AI Loves to Play'. The demand side is settled. The unsettled part is supply: who builds the brain.
Four reasons the brain gets built here
We're a Bengaluru company, so discount our enthusiasm as you see fit. The case doesn't rest on enthusiasm. It rests on four structural facts.
Language
An AI toy for India must speak Hindi from day one, and Hindi is the start, not the finish. A child in Coimbatore dreams in Tamil. Her cousin in Kolkata sings in Bengali. Both flip into English mid-sentence and back out without noticing. A toy trained in California speaks none of this; it does English, brightly, to a four-year-old who hasn't started school in it. Multilingual childhood can't be patched in later. It has to be the foundation. Our stack runs more than ten languages, Hindi and regional Indian languages included, because anything less isn't an Indian toy. It's a tourist.
Price
The bestselling toy in India retails under ₹1,000. Sit with that number. It disqualifies the imported smart module priced for a $79.99 shelf before the conversation starts, and it forces a discipline that toymakers designing for richer aisles never learn: every rupee on the bill of materials has to justify itself. That discipline is why our companion module starts at $10 a unit and baby-care modules at $30. A market this hard on price doesn't merely deserve affordable AI. It manufactures the constraints that make affordable AI possible.
Engineering
Bengaluru ships more software than most countries. The engineering depth that built India's payment rails sits a metro ride from our office, and an AI toy is precisely the problem that depth is shaped for: serious computing on a tiny cost base. We build here, backed by NVIDIA Inception, Karnataka Elevate, NASSCOM and Founders Inc. The point isn't our address. The point is that the brain of a toy is software on a small board, and software on a small board is what this city does at world scale.
Data
Children here switch languages the way other children switch toys. Hinglish at breakfast. Tamil and English braided into one sentence after school. That signal is multi-language, code-switched and real, it exists at this scale only in India, and nobody else is collecting it, because nobody else's toys can hold the conversation. We collect it the only acceptable way: with parental consent and residency promises we can show an auditor. India data stays in India, DPDP-aligned. US data stays in the US. EU data stays in the EU.
The China-plus-one moment reaches the toy aisle
Global brands have spent the past few years spreading manufacturing risk. Apparel moved a generation ago. Electronics followed. Toys are moving now. Tariff rounds between Washington and Beijing turned single-country sourcing into a board-level worry, and procurement teams that once asked Indian factories for quotes out of courtesy now ask about capacity. Plush and plastic are already shifting. The open question is whether the smart layer follows the shell.
India's quality regime helps here, oddly. BIS certification looks like a hurdle from the outside. For a category about to put microphones and cameras into children's bedrooms, it's a floor, and floors are what this category is missing. A market that filters out the cheapest unsafe electronics is a market a global brand can certify into once and defend afterward. We treat BIS the way we treat COPPA: as the spec, not the obstacle. Lori Crib, our AI crib, is in development with a BIS-certified design.
What we're doing about it
Kheelona means toy: खिलौना in Hindi, jouet in French, Spielzeug in German, juguete in Spanish. Lori means lullaby the same way: लोरी, berceuse, Wiegenlied, nana. The names aren't garnish; they're the thesis: a global brand, fluent in the languages children actually speak, Hindi and regional Indian languages among them. We build in Bengaluru and carry that meaning onto every box.
Where that stands today:
- Lumi, our companion plush: live with ten Indian pilot families.
- Lori, the AI baby monitor: production-ready, offered on rental.
- Lori Crib: in development, with a BIS-certified design.
- One school MOU signed, and more than five OEM design-partner conversations running across India and abroad. Yours would start here.
Lori's rental model deserves one more sentence. We invented it for Indian price sensitivity, because a new parent shouldn't have to buy hardware a baby outgrows. Then it surprised us by being the better business model outright. Rental is a relationship. A relationship is exactly what a baby-care product already is.
The economics underneath are published, not whispered. PlayOS is free. The Magic Box module that carries it runs $10–50 a unit for companion toys and $30–100 for baby care. Integration takes four to six weeks, and the partner keeps the retail margin. Pricing born on a ₹1,000 shelf turns out to work on every other shelf on earth.
Two readers, one shelf
If you're an Indian OEM, the differentiation you're hunting is already in your catalog. Your bear, your blocks, your bassinet, with a brain that speaks your customer's language at a module price your retail math can absorb. The first product in each aisle that talks back will make the silent ones beside it look ten years older.
If you're a global brand doing China-plus-one math, the shells were already coming here. The new fact is that the AI layer can be Indian too, with compliance that travels: COPPA for the US, GDPR-K for Europe, DPDP at home, and data pinned to the region it came from. Certify once. Sell everywhere you already sell.
India spent a decade proving it could make the toy. The same forces now line up behind the brain: policy that wants it, engineering that can do it, and a market that demands it at a price nobody else will chase. We think 'made in India' is about to mean the whole toy. We're building as if it already does.