Every Indian consumer company we have spent time with in the last twelve months has the same question on the same slide: 'What does AI actually change about our business?' The slide has a lot of arrows. The answer underneath is rarely as clean as the arrows suggest. This is our working map, drawn from the partners' own portfolios and the deals we have looked at this year.
The frame is simple: AI matters at the consumer layer where it compresses a real unit cost or unlocks a previously impossible interaction. Everywhere else it is a marketing line. We'll go through the stack from acquisition down to operations and try to be honest about what is real and what is decoration.
“India is the 5th largest consumption market globally at $2.1 Tn — 60% of GDP — and the fastest-growing of the major economies at 7.2% CAGR over the last decade.”
— Indus Valley Annual Report 2025
Who actually buys: India1 is 140M people, not a billion
Blume Ventures' framing — India1 / India2 / India3 — is the cleanest read on Indian consumer demand we have come across. India1 is roughly 140M people (the top 10%, ~$15K per person, Mexico-sized), and it carries 2/3 of all discretionary spend. India2 is ~300M (the aspirant class, $3K per person, Indonesia-like), responsible for the remaining third. India3 is the billion below that — beyond discretionary spend, for now.
What this means for an AI-native brand: the addressable market for premium AI-mediated experiences is India1, with India2 reachable on a vernacular-and-microtransaction model (see the STAGE / Kuku FM / NammaYatri playbook). Building a $100Cr-revenue consumer brand in India in 2026 means deciding which India you're building for — and pricing accordingly.
The deepening, not widening, pattern matters: India1's share of national income has gone from 34.1% in 1990 to 57.7% in 2022. The urban top-10% over-indexes 13× on durables, 12× on medical, 11× on out-of-home food vs the per-capita average. AI-native brands that price for this cohort capture outsized ARPU per customer.
Acquisition: real, uneven, mispriced
AI has already changed creative production for performance marketing. Brands that used to ship two creative variants per week now ship twenty, and the best-performing brands ship two hundred. The COGS of a tested ad has dropped roughly an order of magnitude. The CAC, however, has not dropped by an order of magnitude — because every brand has the same tool. The arbitrage window is the eighteen months in which a brand has internalised the workflow and most of its peers haven't.
What is genuinely new in 2026: agentic media buying. Not 'AI-assisted dashboards' — actual agents that allocate spend across Meta, Google, and the marketplaces with hourly feedback loops. The early Indian deployments we have seen lift MER by 15–25% over a quarter. We expect this to be a one-time gain that compresses within three years as it becomes table stakes; we still want our portfolio to capture it.
Discovery and merchandising
Conversational discovery on the brand site is the most-pitched, least-converting AI use case we have evaluated in 2026. The data, across the portfolios we have visibility into, says the same thing: users do not want to chat with a brand to find a product. They want to find the product. Where AI does pay off in discovery is in the silent layer — embeddings-based search, dynamic merchandising, on-site personalisation, and the long-tail SEO that AI content can now produce credibly.
Where this gets interesting in India: vernacular discovery. A meaningful fraction of D2C buyers in Tier 2 and Tier 3 now interact with brands in Hindi, Tamil, Telugu, and Marathi. The brands that build their search, support, and discovery in language — not translated, native — are about to compound very fast.
Conversion and checkout
The honest answer is: less than people claim. UPI and the existing payments rails have already made Indian checkout one of the lowest-friction in the world. The AI uplift here is at the margins — fraud, personalised offers, smarter address parsing, automated COD risk scoring. Useful, not transformational.
Retention and CX
This is where the operator P&L moves most. AI-mediated post-purchase — order tracking, returns triage, complaint resolution, replenishment prompts — is replacing the largest line item in many D2C ops budgets. The portfolios we know are running 60–80% deflection of routine tickets to agents that actually resolve, not just respond. The headcount this displaces is real; the customer satisfaction, where it is done well, is meaningfully higher than human-only support.
“Retention is where AI changes a consumer P&L. Acquisition is where it changes the cost of a test. Discovery is where it changes the language.”
Operations: the under-reported revolution
Behind every consumer brand is a supply chain, a finance function, an HR org, and an inventory model. These are not glamorous and they are where the operator-investor advantage shows up. The largest 2026 efficiency gains we are seeing in Indian consumer companies are not in the consumer-facing layer at all — they are in demand forecasting, SKU rationalisation, vendor-side communications, and book-close automation.
A ₹100 Cr D2C brand running tight ops is now functionally a ₹150 Cr brand of three years ago, with the same headcount. That is the lever LPs underwrite when they back this fund's thesis.
The infra is here, and it's cheaper than its peers
Two specifics worth holding: India's data-centre capacity went from 158 MW in 2014 to 942 MW in H1 2024 — a 6× run, now the 2nd fastest-growing DC market in APAC at 28% growth. And India is the cheapest large market in APAC to build a DC in — $4.59M per MW vs Japan at $12.73M and Singapore at $11.23M. The infrastructure to serve AI-native consumer surfaces is being built right where they will be consumed.
On the talent side, India is the second-largest contributor base to public generative-AI projects on GitHub — only the US has more. Pair that with the central Government's ₹20bn ($240M) AI mission, 18,000 high-end GPUs at 40% below market, and the open call for foundational-model proposals (IndiaAI), and the conditions for a serious AI-native consumer cohort to emerge are real for the first time.
What we are funding inside this stack
- AI-native brands where the operating model is rebuilt around AI from launch, not bolted on.
- Vertical agents that own a real workflow inside a consumer company — returns, support, merchandising, vendor management.
- Vernacular consumer surfaces — voice and chat in Indic languages where the model is the product.
- Infrastructure layers that make these brands and agents cheaper to build and run.
What we are not funding: 'AI-powered' versions of categories that already work fine without AI. If the brand is the differentiator, AI is a feature. If AI is the differentiator, the brand has to be earned on top.