Gravitywell.Research
Sector Analysis · Industry & Sector Research

AI: The India Stack.

The only Indian sector where funding quadrupled into 2026 ($676M in H1, +317%): a sovereign-model programme (Sarvam at $1.5bn), a $1.2bn Blackstone-led compute bet (Neysa), $67.5bn of hyperscaler pledges, and a first unicorn (Krutrim) already in retrenchment.

$676M
AI startup funding H1 2026: +317% YoY
$1.5bn
Sarvam valuation: the sovereign-model bet
38,000+
GPUs empanelled under IndiaAI (54k announced)
₹400 cr
actually disbursed of ₹10,372 cr Mission outlay
CodeGWR-SEC-AI
PillarIndustry & Sector Research
CadenceRefreshed each cycle
VintageJuly 2026

The scorecard

India's AI stack is three different investments wearing one label. The application layer (~75-80% of startups) is capital-light and occasionally explosive: Emergent hit $50M ARR on $99M raised and quintupled its valuation in three months. The model layer is state-anchored: Sarvam trained a 105B-parameter sovereign model on government-provisioned H100s, and its $234M Series B (HCLTech strategic) is the only large private model-layer round. The compute layer is where institutional money actually went: Neysa's $1.2bn. The honest constraint: OpenAI raised 166x India's entire H1 AI funding in one round, and IndiaAI has disbursed just ₹400 cr of its ₹10,372 cr outlay in two years. India's edge is deployment at population scale, 22-language coverage and inference pricing an order of magnitude below global APIs, not the frontier.

Demand Outlook8Compounding

Market at 25-35% CAGR toward $17bn by 2027 (Nasscom-BCG); ~80% of new GCCs carry AI mandates; Gnani alone processes 30M voice interactions/day.

Competition (global)3Lopsided

OpenAI raised 166x India's H1 2026 AI funding in a single round; Sarvam's benchmark claims lack independent replication; Llama is free and Reliance-distributed.

Capital Intensity0Three-tier

Apps are capital-light (Emergent $50M ARR on $99M); models state-subsidised; compute devours capital (Neysa $1.2bn for 20k GPUs).

Monetisation4Unproven at top

Krutrim, the first GenAI unicorn, cut 550→150 staff and pivoted to cloud; sovereign models have no proven demand beyond procurement.

Policy Anchor2Announced > delivered

IndiaAI GPUs at ₹65-92/hr (~one-third global cloud) is real; ₹400 cr/2yr disbursement vs outlay is the under-reported gap.

Risk-Adjusted Return8Barbell

Application-layer entries at reasonable multiples + services cash flows vs $1.5bn pre-revenue model bets; know which side you're on.

Scores are 0-100 favourability. Competition at 35 reflects the structural frontier gap; the sector's investability lives in the layers that don't fight it. Priors reflect the January 2026 read.

The numbers

AI-native startup funding · $ m / yr
-82128338547757BASE 1006762020202220232025H1'26

H1 2026 alone ($676M) ≈ one-third of ALL historical Indian AI funding; scope excludes the Neysa infra megadeal

IndiaAI GPUs empanelled · count
-12001189995000081000112000BASE 100100000Mar'24Mar'25May'25Feb'26'26 tgt

38,000+ accelerators empanelled (incl. first TPUs); 100k target by Dec 2026: empanelled ≠ deployed

India AI market · $ bn
-3281419BASE 10017.0202420252026e2027e

$17bn by 2027 on the Nasscom-BCG path (25-35% CAGR): definitional spread vs IDC's $6bn is wide

Demand · will supply get filled?

The demand question splits by layer: enterprises buy applications and services today; models sell to the state; compute sells to everyone building the other two.

Market CAGR
25-35% → $17bn (2027e)
GCC AI mandates
~80% of 2026 launches
AI professionals
~920,000 (largest pools)
Skills gap
82.9% in GenAI roles
Funding mix · by layer (E, incl. infra)
55%
24%
14%
7%
Compute / GPU cloud 55%Applications & agents 24%Foundation models 14%Data / evals / services 7%

Dollar-weighted, the 'AI boom' is mostly a compute build-out: Neysa's $1.2bn alone was ~81% of the broadest-scope Q1 2026 tally. By company count it's the opposite: ~75% of startups build applications. GCCs are the silent third force, holding ~27% of national AI talent inside foreign-owned centres.

Output, order book & the global gap

Funding charts can't answer the operating question: is anything actually shipping, being used, and contracted? Usage (API calls, GPU onboarding, dataset downloads) is the deployment proof; the sovereign order book is demand depth; the global table is the honest scale check.

Operational output
Sarvam inference
>10M API calls/day
plus >2M conversational interactions/day; claimed reach: 17M farmers, 45M insurance customers (company-claimed)
IndiaAI GPUs onboarded
34,381 → 38,000+
14 empanelled providers; utilisation % UNPUBLISHED: reports promised 'next quarter'; a data gap in itself
Funded model projects
12 organisations
BharatGen ₹1,058.5 cr (4x next-highest), Sarvam 4,096 H100s (~₹99 cr subsidy), Gnani, Fractal, Soket, Gan.ai…
Models actually shipped
3 flagship
Sarvam-30B + 105B (Feb 2026, 128k ctx MoE) and BharatGen PARAM-2 (17B, 22 languages): sub-frontier by design
AIKosh datasets
5,500+ / 251 models
but ~13,000 downloads and 6,000 registered users by Jul 2025: catalogue deep, usage thin
Enterprise adoption
40-62% using AI
Deloitte: 40% significant usage (vs 28% global); Nasscom: 62% on GenAI; EY-CII: 47% with multiple use cases in production
Order book · contracted backlog
BharatGen deliverables
₹1,058.5 cr allocation: voice + document models, India benchmarks, CPGRAMS + Bhashini deployments due end-2026
Sarvam × Odisha
MoU (Feb 2026): 50 MW sovereign AI capacity hub + Vision AI for mining safety + Odia voice: MoU, not yet revenue
Sarvam × Tamil Nadu
'Digital Sangam' sovereign AI research park with IIT Madras, 20 MW DC
Sarvam 'Chanakya'
High-assurance vertical for defence/regulated finance/government (Mar 2026); UIDAI + ministry deployments referenced
IndiaAI Mission 2.0
+20,000 GPUs to >58,000 public units at ₹65/hr; ~₹10,372 cr enhancement; 100k GPUs targeted by late 2026
GCC AI build-out
1,200+ GCCs with embedded AI/ML, 250+ AI CoEs: the private-sector order book nobody tenders
Global gap · India vs the leaders
Private AI investment (2025)~$1.5bn (Q1'26 run)US $285.9bn · China $12.4bn
India cumulative 2013-24 ~$11.1bn, rank #10: a 100x+ annual gap to the US
Notable models (2025)2-3 sovereignUS 50 · China 30
US-China frontier gap narrowed to 2.7%; India ships mid-scale MoE, not frontier
Top AI talent50,460: #2 globallyUS 220,520
still a net exporter of AI talent, but transitioning; H-1B registrations −38% after the $100k fee
AI publications share7.6%China 17.8%
Europe 11.1%: India's research share lags its talent share badly
GPU install base~50k H100-class permitted to 2027Single US campus >100k
India's national two-year allocation is smaller than one hyperscale project
Workforce AI usage>80% report regular useTop tier globally
adoption-intensity leads despite the capex gap: the deployment-market thesis in one row
The honest read

The honest read: India ships adoption, not frontier: enterprise deployment and population-scale reach lead the world while capital, compute and research share trail by one to two orders of magnitude. The sovereign order book is real but mostly MoUs; the revenue line behind the $1.5bn model-layer valuation is ₹45 crore.

Competitive dashboard

Share of tracked private-AI value · GW estimate
Sarvam
30%
Emergent
12%
Krutrim
10%
Fractal
14%
Deccan / Composio / Gnani
10%
Others
24%

By dollar the compute layer dominates; by count the application layer does. This chart tracks named private companies, ex-hyperscalers.

The capital pledges · $bn
AWS pledge
Through 2030: largest single India commitment
$35b
Microsoft
CY2026-29, largest MSFT pledge in Asia
$17.5b
Google
5-yr, gigawatt AI hub at Visakhapatnam
$15b
Neysa
Blackstone-led; India's largest AI financing
$1.2b
IndiaAI Mission
₹10,372 cr outlay: ₹400 cr disbursed
$1.24b

Pledges, not booked capex: $67.5bn announced by three hyperscalers in ten weeks. Delivery is the 2027 question.

Geographic concentration · share
45%
18%
12%
12%
Bengaluru 45%Delhi-NCR 18%Hyderabad 12%Mumbai / Pune 12%Chennai 8%Other 5%

Capital · unit economics, valuation & deals

Emergent
$50M ARR / $99M raised
The application-layer proof: 30x ARR at the $1.5bn talks
Krutrim FY26
~₹300 cr rev (co-claim)
3x YoY with first profit claimed, after cutting 70%+ of staff
Fractal FY25
₹2,765 cr / 17.4% EBITDA
AI services economics at model-company multiples (109x IPO P/E)
GPU spread
₹65 vs ₹250-500/hr
Subsidised vs market H100: the state IS the price setter
Sarvam API input
₹4/M tokens (~$0.047)
vs global frontier
1-2 orders cheaper
H100 market rate
₹249-500/hr
IndiaAI subsidised
₹65-92/hr (~1/3 global)
Services margin
Fractal 17.4% EBITDA

Inference & compute economics: The cost wedge is real at both layers: subsidised compute at a third of global cloud, sovereign inference at a fraction of frontier APIs, but it prices smaller models against frontier capability. Services economics (Fractal at 17% margins, 109x P/E at IPO) carry model-layer multiples.

Recent transactions
Neysa × Blackstone
$1.2bn ($600M equity + $600M debt) at ~$1.4bn: 20,000+ GPUs; India's largest AI financing (Feb 2026)
Sarvam Series B
$234M first close at $1.5bn post: HCLTech $150M strategic for ~10% (Jun 2026)
Emergent
$70M at $300M (SoftBank/Khosla, Jan 2026); $250M at $1.5bn in talks (unclosed)
Reliance × Meta
₹855 cr Llama JV + Meta's first India AI-DC deal: 168 MW at Jamnagar (Jun 2026)
Deccan AI
$25M Series A (A91): post-training data; DeepMind among customers
BharatGen
₹900 cr state grant: Param-2 17B MoE, 22 languages

Follow the instrument, not the label: equity went to applications, debt+equity to compute, grants to models. The Reliance-Meta axis is the wildcard: free Llama distribution through India's largest conglomerate caps every domestic model's commercial ceiling.

Public-market proxies & IPO pipeline
Netweb
+139% 1-yr: the clean infra winner
AI hardware
E2E Networks
+50% 1-yr but rolling over
GPU cloud
Fractal
~flat vs Feb 2026 IPO; 109x P/E
AI services
Affle 3i
−22-30% 1-yr: derated
Ad-tech AI

The public market has already picked its lane: hardware/infra proxies re-rated hard, services flat, AI-adjacent software derated. The listed read matches our private-layer scorecard almost exactly.

Private players & platforms

Where most of the value is still private: startup-, PE- and strategic-backed. Scale, ownership, and the last marker of value.

Sarvam
10M+ API calls/day
Backers
HCLTech (10.46%), Bessemer, Khosla, Peak XV
Value marker
$1.5bn

FY26 revenue ₹45.1 cr: ~275x trailing; the sovereign-AI bet in one multiple

Neysa
20,000+ GPUs planned
Backers
Blackstone, Teachers' VG, TVS Capital, 360 ONE, Nexus
Value marker
$1.4bn

$600M equity + $600M debt: India's largest AI financing

Yotta
>16,000 H100 + 8,000 B200 on order
Backers
Hiranandani
Value marker
IPO intentions reported

Claims 60-70% of India's GPU capacity; IndiaAI's largest empanelled provider

Krutrim
GPU cloud pivot
Backers
Ola/Bhavish Aggarwal
Value marker
$1bn (stale)

550 → ~150 staff; LLM + chip paused; ~₹300 cr revenue with first profit (company-claimed)

Emergent
$50M ARR
Backers
SoftBank Vision Fund, Khosla, Lightspeed
Value marker
$300M → $1.5bn talks

Coding agents; the application layer's velocity print

Deccan AI
1M+ contributor network
Backers
A91, Susquehanna, Prosus
Value marker
Undisclosed

Post-training data for DeepMind/Snowflake: India as the world's RLHF supplier

Startups & emerging players · the VC layer

Where venture capital enters the theme.

Composio$29M (Lightspeed)
Agent integration infra

100k+ developers: picks-and-shovels for agents

Gnani.ai$10M Series B + talks
Voice AI

IndiaAI voice-model awardee; 30M interactions/day

Soket AI / Gan.ai / AvataarIndiaAI-funded
Foundation-model awardees

Three of the 12 sovereign-model grantees

GenLoop / Zenteiq / Shodh AIIndiaAI-funded
Applied + scientific AI

The long tail of the sovereign programme

IntelliHealthIndiaAI-funded
Health AI

Sector-vertical sovereign model

VC white-space

The 12 funded foundation-model organisations are the sovereign programme's bench; the venture-backed layer (Composio, Deccan, Emergent) mostly sells globally from day one: the two ecosystems barely overlap.

Public-market exposure index · rules-based, purity-weighted

A screened, exposure-weighted basket: each listed name weighted by its sector-exposure purity score (not naively equal-weighted), after liquidity and quality screens. Selection is rule-driven and set ex-ante.

3-yr CAGR (purity-wt)
76%
from +442% total over 3y
1-yr return (wt)
31%
0 screened out
Illustrative SIP XIRR
76%
= CAGR under smooth growth; real needs NAV
Constituents
8
purity-weighted, 25% cap, qtrly rebal.
Rebased growth · 100 = 3 years agoReal 1y/3y anchors · purity-weighted
47184321458595BASE 1005423y agonow

Real point-to-point anchors: each name rebased to 100 at −3y; the −1y (319) and now (542) levels from its actual 1Y & 3Y returns, purity-weighted. Intra-period linear (daily shape/drawdowns need a price feed).

Fractal Analytics FRACTAL8521.8%+-1%+0%
E2E Networks E2E8020.5%+85%+1400%
Netweb NETWEB7519.2%+139%+790%
Affle 3i AFFLE359.0%+-25%+35%
Persistent Systems PERSISTENT307.7%+-24%+110%
KPIT Technologies KPITTECH307.7%+-43%+-22%
Tata Elxsi TATAELXSI307.7%+-30%+-40%
Happiest Minds HAPPSTMNDS256.4%+-47%+-62%
Methodology

Rules-based: include a listed name if its AI purity score ≥ 20/100 AND it clears the eligibility screens. Weight by purity (exposure-weighted), single-name cap 25%, overflow redistributed pro-rata. Quarterly reconstitution. Selection is rule-driven, set ex-ante, not a curation of past winners.

  • Liquidity & size: investable free-float, adequate ADTV
  • Quality: positive profitability (excludes loss-makers)
  • Purity: AI revenue-exposure / relevance score ≥ 20 of 100

Rules-eligible, pending verified data: Oracle FS (purity < 20), Zaggle (purity < 20), L&T Technology Services, Cyient, Infosys (too diversified). Purity scores are a documented judgement tier (clean AI-revenue % is rarely disclosed; Netweb's 43.4% AI-systems revenue share is the exception that anchors the scale). Returns are representative point-to-point figures; 3-yr figures partly estimated (E).

⚠ Hindsight / selection bias

Selection-bias caution: built from names already public over the window, which the AI theme has already re-rated (Netweb +139% in a year). Past returns are upward-biased and NOT a forward estimate; the rules, not the hindsight, are what's intended to repeat.

⚠ Disclaimer

Research / informational only: not investment advice or a recommendation. Baskets are illustrative of the rules, not a managed product.

Externalities & policy footprint

The externalities and strategic stakes a government must price in.

AI job openings
1M+ (2026)
demand +40% YoY; 4M jobs projected by 2030
Unfilled AI roles
51%
WEF; only ~16% of IT professionals AI-skilled (MeitY)
GCC AI professionals
~250,000
~27% of national AI talent inside foreign-owned centres
Returnees (2025)
15,100
vs 9,800 in 2024; US-bound outflow halved after H-1B fee: the brain drain is reversing at the margin

Scenarios to 2030

Bear
$17-20bn core AI market by 2030
Nasscom-BCG floor

Frontier price war erases the inference wedge; sovereign models retreat to procurement; services capture the value

Base
$25-30bn by 2030
25-35% CAGR holds

Deployment-market thesis: applications + GCC + data services compound; compute build-out delivers partially

Bull
$40bn+ by 2030
Infra converts

The $265bn DC pipeline + $67.5bn hyperscaler pledges convert to services revenue; EY's $359-438bn GDP-add overlay materialises

The reality check

Three numbers puncture the narrative: Sarvam's ₹45.1 crore FY26 revenue against a $1.5bn valuation (~275x trailing); AIKosh's ~13,000 downloads against 5,500+ catalogued datasets; and IndiaAI's unpublished GPU-utilisation figures. The capacity story is real and accelerating: the demonstrated-usage story is one to two orders of magnitude behind it. Underwrite deployment, not announcements.

Financing · policy · catalysts

Policy & incentives
Uttar PradeshLucknow AI City (260 acres, 10,000 GPUs planned); 8 DC parks targeting ₹2 lakh cr; ₹350 cr AKTU AI institutes for 100k students
MaharashtraAI Policy 2026: ₹500 cr AI venture fund, capital subsidies, stamp-duty and power-tariff benefits; ₹10,000 cr investment target
TelanganaAI City near Hyderabad; AI services to 1 cr+ citizens by 2027; 500,000 AI-skilled target; DCs as essential service
Tamil NaduTAMDEF governance framework + DEEPMAX scorecard; Digital Sangam sovereign AI park with Sarvam + IIT Madras (20 MW)
OdishaSarvam MoU: 50 MW sovereign AI capacity hub, Vision AI for mining, Odia voice stack
KarnatakaLand subsidies + tax exemptions for AI-DC investment; no flagship 2026 policy yet (E)
What to watch
Dec 2026IndiaAI 100,000-GPU milestone: empanelment vs deployment is the test
H2 2026Emergent $250M close at $1.5bn; Sarvam Series B final close (~$66M remaining)
2026-27Sarvam's next frontier model (agentic/coding): the sovereign programme's commercial test
Feb 2027FY28 budget: IndiaAI allocation after the FY27 trim to ₹1,000 cr

Sensitivities · what moves returns

Risks quantified, not just listed: the levers that swing the underwriting. Directional, illustrative.

IndiaAI disbursement accelerates₹2,000+ cr/yr actually flowsModel + compute layers re-rate; Yotta/E2E/Neysa order books inflect
Frontier model price warGlobal API prices fall 5-10xIndia's inference-cost wedge evaporates; sovereign models retreat to pure procurement
GPU supply shockExport controls tightenCompute layer repriced; subsidised access becomes strategic allocation
Application-layer exitsEmergent-class exit at 10x+Validates the barbell's cheap end; VC funding rotates further to apps

Technology roadmap · what changes the game

ModelsMid-scale MoE shipped (Sarvam-105B, PARAM-2 17B)Sarvam agent stack + Chanakya (defence/finance); BharatGen voice + document models by end-2026: explicitly NOT chasing trillion-parameter frontier
Compute38k GPUs empanelled at ₹65-150/hr; first 1,050 TPUs58k → 100k public GPUs (late 2026); ₹1.64 lakh cr semiconductor approvals; Krutrim's chip programme paused: sovereign silicon setback
Applications24% of leaders have deployed agentic AIIndia + Brazil lead pilot→production conversion; 50% IT-productivity ROI realisation: the agentic scale-up is India's to win
DataAIKosh 5,500+ datasets; Bharat Data Sagar corpusIndia as global post-training data supplier (Deccan-model); Indic synthetic corpora for the 22-language stack

Demand drivers

  • Population-scale deployment surface: UPI-grade digital rails + 22 scheduled languages = the world's largest inference market by users.
  • Cost wedge at every layer: subsidised GPUs at ~1/3 global, inference at 1-2 orders below frontier APIs, engineering talent at GCC rates.
  • Sovereign procurement as anchor demand: the government is the model layer's first paying customer.
  • $67.5bn of hyperscaler pledges since Oct 2025: even partial delivery transforms the compute base.
  • GCC densification: 2,100+ centres, ~80% of new launches AI-mandated.

Risks

  • ! Frontier dominance: a 166x ecosystem funding gap cannot be closed by subsidy; India's models compete on price and language, not capability.
  • ! The disbursement gap, ₹400 cr released of ₹10,372 cr, means the policy anchor is mostly still a press release.
  • ! Monetisation unproven where valuations are highest: Krutrim's retrenchment is the base case, not the tail.
  • ! Talent drain in place: GCCs absorb the best AI engineers at premium comp for foreign balance sheets.
  • ! Froth markers everywhere: 5x valuation step-ups in a quarter, 30x ARR, pre-revenue unicorns.

What it means · by capital type

For VC

The barbell is the strategy: application-layer companies with real ARR (Emergent-class) at the cheap end, data/eval infrastructure (Deccan, Composio) as picks-and-shovels. Leave sovereign-model equity to strategics and the state.

For PE / growth

Compute is the institutional lane: Blackstone's Neysa structure (half debt) is the template: real assets, contracted demand, subsidy tailwind. AI services (Fractal-class) offer cash flows but the 109x entry multiple is the warning.

For hedge funds

Trade the layer rotation: Netweb's +139% says infra is priced; Fractal-flat says services aren't believed; the gap closes one way or the other. Watch IndiaAI disbursement runs as the sector's true leading indicator.

For government

The mission's binding constraint is disbursement machinery, not ambition: ₹400 cr of ₹10,372 cr in two years undermines the sovereign-AI thesis more than any benchmark dispute. Fix the pipe before adding GPUs to the target.

Data vintage July 2026. Anchored to 2025-2026 industry and official prints; figures across sources differ and are reconciled to the cited ranges. Sources: Inc42: H1 2026 AI funding (+317%)S · Blackstone: Neysa $1.2bn (Feb 2026)P · Sarvam: Series B announcementP · Medianama: IndiaAI ₹400 cr disbursementS · Digital India: 34,000+ GPU milestoneP · TechCrunch: Krutrim cloud pivotS · Layer mix, market path, geography: GW estimatesE

Data confidence. High on funding rounds, GPU tenders and listed prints (P/S); medium on market-size projections (definitional spread is wide); company-reported ARR/profit claims flagged individually (E).

Data & sourcing policy

Sourcing. Every figure is sourced and dated. We tier provenance: Primary (official, regulatory, exchange or company filings), Secondary (tier-1 industry research and reputable media), and GW estimate (our own reconstruction or opinion, labelled, never presented as external fact). We prefer primary where it exists, reconcile divergent prints to cited ranges, and hold every number point-in-time: dated, and never silently restated; revisions publish as dated changes.

Fact vs opinion. Facts vs opinion: market sizes, official prints, prices, named deals and agency ratings are sourced facts (Primary/Secondary). Scores, grades, purity weights, scenario paths and indicative sparkline points are Gravitywell's analytical opinion (GW estimate): labelled, not presented as external data.

PPrimary: Official / regulatory / exchange / company filingSSecondary: Tier-1 industry research or reputable mediaEGW estimate: Gravitywell reconstruction or opinion: our analysis, not an external fact

The sector, each cycle.

AI: The India Stack refreshed every cycle, with the scorecard, dashboard, and capital read. More sectors rolling out.

All sectors

New reports and coverage updates. No spam. Unsubscribe anytime.