6.6%
India real GDP growth, 2025P
Fastest-growing major economy. World avg 3.3%, China 4.8%, US 1.7%.
Source · IMF
The thesis
·5 min read
What we believe about India, AI, and the next decade of founders.
A thesis is a claim about the future, sharp enough that being wrong has a cost. This is ours. If it disqualifies you, that is a feature.
01 — Macro
The macro bet
India is in the middle of three simultaneous structural shifts — a demographic dividend that will not repeat, a payments-and-identity stack that no other large economy has, and the global pricing collapse of intelligence. Any one of these would shape a decade. All three are stacking inside the same five-year window.
The companies that will define India's next decade are not bigger versions of the last one. They are AI-native from the first commit — built to run lean teams, productionise models as core infrastructure, and distribute through rails that did not exist five years ago. Software is no longer eating the world here. Intelligence is.
We are an operator-led pre-seed and seed fund, writing $50K–$500K cheques into the founders building this shift. Our partners have built the consumer, commerce, gaming, and content businesses that operate at the scale Indian founders aspire to. We invest where we can actually help.
01 — Macro · By the numbers
Three forces, one window.
The shifts shaping the next decade aren't hypotheses — they're already in the print. The four numbers below are the ones we underwrite Fund I against.
$4.5 Tn
India market cap, Jan 2025
4th globally — behind only US ($60.8T), the Mag 7 ($11.7T), China ($9.9T), Japan ($6.5T).
Source · Jefferies
$2.1 Tn
India consumption market
5th largest globally · 60.3% of GDP. 10-yr CAGR 7.2% — highest among major economies.
Source · UBS
$1,493
Per-capita consumption, 2023
Where China was in 2010. Indonesia: $2,656. China: $4,936.
Source · World Bank
02 — Where
The four pillars
We invest across four pillars. Each is a category we have an operator-level view on — meaning we have built, scaled, or shipped in adjacent territory, and can debate the roadmap rather than nod at it.
Pillar 01
Agentic AI
Software that acts, not just answers.
The category we are most excited about is the one that didn't exist two years ago: software that takes actions on behalf of a user, end-to-end, with judgment. Browsers that book, voice agents that resolve, back-office agents that close the books, sales agents that qualify.
We back the infrastructure layer (evals, tool use, memory, orchestration) and the vertical applications where domain depth compounds. We avoid wrappers that a model upgrade would erase.
What this looks like
- Vertical agents in regulated workflows — legal, accounting, healthcare ops, financial services back-office.
- Tooling for agent builders — evaluations, guardrails, simulated environments, agent memory.
- Voice-first agents replacing call centers in India and the global south.
Pillar 02
AI-native consumer & commerce
Brands and commerce surfaces built AI-first from day one.
India's consumer internet is being rebuilt around AI-generated content, AI-personalised merchandising, and AI-mediated purchase journeys. The unit economics of D2C, content, and creator businesses are about to be redrawn.
We partner with operators who have already scaled consumer businesses in India — Shantanu (Bombay Shaving), Anurag (Pilgrim), Rahul (Naturis) — and back the next cohort of founders who treat AI as a core operating advantage, not a marketing line.
What this looks like
- AI-native brands where personalisation, content production, and CX are model-powered from launch.
- Commerce infrastructure — AI search, conversational checkout, returns, post-purchase.
- Creator-economy platforms where AI compresses the cost of production by an order of magnitude.
Pillar 03
Platforms beyond software
When the interface is voice, ambient, or invisible.
For two decades, every new platform was a screen. The next platform shift is happening off-screen — voice in vernacular languages, ambient compute in the home and car, wearables that take action, hardware that ships with a model inside.
India will leapfrog desktop and even mobile-first patterns in entire categories. We back founders building for the interfaces that are arriving, not the ones we already have.
What this looks like
- Indic-language voice platforms — commerce, support, education, government services.
- Ambient and on-device AI — wearables, automotive, home, retail point-of-sale.
- Hardware-software companies where the model is the product.
Pillar 04
AI in healthcare
Where AI compresses cost of care for a billion people.
Indian healthcare has a structural shortage of clinicians, a fragmented payer landscape, and a population that will demand care at scale within this decade. AI is the only technology that meaningfully changes the cost curve.
We back founders in diagnostics, clinical workflow, ops automation, and care delivery — where AI is not a feature but the reason the business exists.
What this looks like
- Diagnostics and triage tools deployed in primary-care settings outside metros.
- Clinical workflow automation — documentation, coding, prior auth, RCM.
- AI-mediated care delivery models — chronic disease, mental health, maternal care.
Pillar 02 · Consumer stack
Who actually consumes in India.
Blume's framework (India1 / India2 / India3) is the cleanest read on Indian consumer demand we have seen. The 'consuming class' is ~140M people — Mexico-sized, with Mexico-like purchasing power.
~140M
India1 — the consuming class
~10% of population · ~$15K per person · 2/3 of all discretionary spend. Mexico-like.
Source · Blume Ventures
~300M
India2 — the aspirant class
~23% of population · ~$3K per person · 1/3 of discretionary spend. Indonesia-like.
Source · Blume Ventures
13×
Urban Top-10% over-indexes on durables
vs avg per-capita spend. 12× on medical, 11× on out-of-home food, 10× on jewellery, 9× on education.
Source · Bernstein
57.7%
Top-10% share of national income, 2022
Up from 34.1% in 1990 — India1 is deepening, not widening.
Source · World Inequality Lab
India · Consumer stack
The top 10% share of India’s national income went from 34% to 58% in a generation.
India isn’t widening as much as it’s deepening. The cohort that already consumes is consuming more — which is who AI-native brands are built for.
22.2%
43.7%
34.1%
20.6%
39.9%
39.5%
15%
27.3%
57.7%
1990
2000
2022
Top 10% (India1)
Middle 40% (India2)
Bottom 50% (India3)
Per-adult pre-tax share of national income, % · Source: World Inequality Lab
Pillars 01 & 03 · AI infra
The infra is cheaper here. The talent is already coding here.
India is the world's second-largest contributor base to public generative-AI projects, builds data centres at a fraction of APAC peer cost, and has explicit Govt. backing for foundation-model attempts. The platform layer is real.
6×
India data-centre capacity growth
158 MW (2014) → 942 MW (H1 2024). 2nd fastest-growing DC market in APAC at 28% growth.
Source · Cushman & Wakefield
$4.59M
Cost per MW to build a DC in India
Cheapest in APAC. Japan: $12.73M. Singapore: $11.23M. India turns infra arbitrage into a moat.
Source · Cushman & Wakefield
#2
Global GitHub contributors to GenAI
India is the second-largest contributor community to public generative-AI projects, after the US.
Source · GitHub
$240M
India Govt AI mission allocation
Plus 18,000 high-end GPUs at 40% below market rates, via the IndiaAI initiative.
Source · IndiaAI / Govt of India
AI infra · Cost arbitrage
India builds AI infrastructure at roughly half the cost of Japan, Singapore, or Korea.
Combined with the world’s second-largest GenAI developer pool, the infrastructure-cost gap is what makes Indian AI businesses globally competitive before they raise their A.
Japan
$12.73M
Japan
$12.73M
Singapore
$11.23M
Singapore
$11.23M
South Korea
$9.23M
South Korea
$9.23M
Hong Kong
$9.19M
Hong Kong
$9.19M
Indonesia
$8.59M
Indonesia
$8.59M
Malaysia
$8.53M
Malaysia
$8.53M
Thailand
$7.45M
Thailand
$7.45M
China
$6.84M
China
$6.84M
India
$6.79M
India
$6.79M
Vietnam
$6.70M
Vietnam
$6.70M
Taiwan
$6.15M
Taiwan
$6.15M
Philippines
$4.59M
Philippines
$4.59M
Median price of constructing a data centre per MW of capacity, USD million · APAC peers · Source: Cushman & Wakefield
03 — Who
What we look for in founders
- 01
Builders, doers, blitzscalers
Founders who ship before they pitch. Who would build the thing whether or not we funded it.
- 02
Ahead of the curve, not chasing it
Founders who saw the AI shift before the deck templates caught up. We are not the right partner for a fast-follow.
- 03
Second-order thinkers
Operators who can converge ideas across domains — consumer + AI, ops + AI, healthcare + AI — and see where the wedge actually compounds.
- 04
Time-as-wealth operators
Founders who treat time as the scarcest resource. Who run lean, decide fast, and protect their own focus the way LPs protect capital.
- 05
Founders with pratyaya
The conviction to compound through the unglamorous middle years — when the press cycle has moved on and the work is mostly hard. Pratyaya is what we look for. Pratyaya is what we back.
04 — Not us
What we don’t fund
We’d rather lose a deal we couldn’t help on than waste a founder’s six weeks. If you’re in any of these buckets, the honest answer is no — and we’d rather you spend the time on a fund that fits.
- —Pure-services businesses with no software wedge
- —Climate-tech (a specialist domain — not our edge)
- —Social-cause-led ventures (admirable, not our mandate)
- —Series A+ deals (outside our cheque size)
- —Companies outside India (Fund I is India-only)
05 — Why now
Why this window
Three forces converged inside an 18-month window. We don’t expect them to hold forever — which is why Fund I is being deployed now, not later.
Force 01
Foundation model commoditisation
Inference costs are dropping ~10× a year. The defensible layer has moved up the stack — to data, distribution, workflow lock-in, and operator judgment. India-based founders can build globally competitive products at a fraction of the cost.
Force 02
India's developer + consumer scale
The largest English-fluent developer base in the world, paired with a digital-payments-rails-complete consumer market of 800M+ internet users. No other geography has this combination at this price.
Force 03
Capital pullback rewards discipline
The 2021 cohort over-capitalised on weak unit economics. The 2026 cohort is being underwritten on operator discipline and capital efficiency. This is the regime we are built for.
06 — Tailwinds · Under-penetration
A consumer market the developed world has already had — that India hasn't, yet.
The four numbers below are why a single percentage point of penetration is a category-defining business. They are not aspirations — they are gaps.
3%
Credit-card penetration
vs US 69%, Brazil 29%, China 21%. Most under-penetrated large fintech market on earth.
Source · CLSA
21%
Mutual-fund penetration, 2024
vs USA 95%, UK 65%, Brazil 78%. SIP contributions hit ₹2.1Tn ($25Bn) in FY25.
Source · AMFI
8%
Room AC penetration
vs Global 42%, China ~100%, US 90%. India accounts for ~7% of global AC units sold; China was ~55%.
Source · Jefferies
0.9%
General insurance as % of GDP
vs USA 8.6%, Germany 3.7%, China 1.9%. India is structurally under-insured.
Source · CLSA
Tailwinds · Under-penetration
A consumer market the developed world has already had — that India hasn’t, yet.
Each tile below is a category where a single point of penetration shift in India builds a category-defining business. These are gaps, not aspirations.
% of population
Credit-card penetration
India
3%
USA
69%
Source · CLSA
% of population, 2024
Mutual-fund penetration
India
21%
USA
95%
Source · AMFI
% of households
Room AC penetration
India
8%
China
100%
Source · Jefferies
% of GDP, 2023
General insurance
India
0.9%
USA
8.6%
Source · CLSA
07 — The market · Capital reality
The funding market we are deploying into.
Indian VC isn't 'in a drought' — it's reset to its pre-2021 cadence with bigger seed cheques, more institutional pre-seed capital, and stretched time-to-graduate. The bar to raise is higher. The bar to get help is the same.
$11.2B
India VC, 2024
Down from $37.4B peak (2021). 1,721 rounds — half the 2021 count. Discipline, not drought.
Source · Tracxn
50%
Seed funding now in '$3M+ mango rounds'
Up from <10% in 2017. Sub-$1M rounds fell from 57.1% to 18.2% of seed funding — fewer cheques, larger sizes.
Source · Tracxn
100+
MicroVCs in India
Filling the pre-seed gap between angels and choosy multi-stage funds — typically $100K–$500K cheques.
Source · Blume Ventures
30 months
Avg Series A→B in 2024
Up from 21 months in 2017. Seed→A also stretched from 23mo to 27mo. The bar to graduate is higher.
Source · Tracxn
India seed · 2017–2024
Mango seeds ate the market. $3M+ rounds went from 9% to 50% of India seed funding.
Sub-$1M rounds collapsed from 57% to 18% of seed in seven years. Fewer cheques, larger sizes — and a real pre-seed gap that MicroVCs are now filling.
'172017
'182018
'192019
'202020
'212021
'222022
'232023
'242024
$3M+ (mango seeds)
$1M–$3M
<$1M
India seed funding split by round size, USD millions · Source: Tracxn
Next step
If the thesis fits, send us the deck.
Response within 2 business days. Decision within 21. We share the reason if we pass.