India's AI Mission: The Seven Pillars and What Each One Actually Funds

HOW TO READ THIS NOTE

This is written as a practical access guide, not just an explainer. Most coverage of the IndiaAI Mission tells you what it is; this note tells you how to use it. Each of the seven pillars includes a "How to Access It" box with the exact portal, eligibility, and steps — and Part VII consolidates everything into a single quick-start table. The Cabinet's own approval document and official IndiaAI portals — not shorthand summaries — are treated as the primary sources.

---

India's AI Mission — Key Facts (as of mid-2026)

Table 1: The IndiaAI Mission at a Glance (Sources: PIB, MeitY, Storyboard18, Medianama, Business Standard — full citations in References)

---

Part I — Why the Mission Exists

On 7 March 2024, the Union Cabinet, chaired by Shri Narendra Modi, approved the "comprehensive national-level IndiaAI Mission" with a budget outlay of ₹10,371.92 crore. Source: PIB Press Release ID 2012355, March 2024.

The framing the government used at the time is worth quoting directly, because it sets up everything that follows: the Mission exists "in furtherance to the vision of Making AI in India and Making AI Work for India." Source: PIB Press Release ID 2012355. That double formulation — building AI capability in India, and making sure AI's benefits actually reach India's population and institutions — is the thread that runs through all seven components. It is also the standard against which the Mission's actual performance should be judged, two-plus years in.

The Mission is implemented by the "IndiaAI" Independent Business Division (IBD), a unit created under the Digital India Corporation (DIC) rather than as a standalone new ministry or department. Source: PIB Press Release ID 2012355. This matters organisationally: it means the Mission functions more like a coordinated fund-and-programme office sitting across seven distinct initiatives than a single monolithic agency — a structure that shows up later in this note, in how unevenly the seven pillars have moved.

India's AI ambitions did not begin with this Mission. Six years earlier, in June 2018, NITI Aayog published a discussion paper titled the National Strategy for Artificial Intelligence — known by its hashtag, #AIforAll — following a mandate from the Finance Minister's 2018-19 budget speech to establish a National Program on AI. Source: NITI Aayog, "National Strategy for Artificial Intelligence," June 2018. That 2018 strategy identified five priority sectors (healthcare, agriculture, education, smart cities and infrastructure, and mobility), proposed Centres of Research Excellence and International Centres for Transformational AI, and coined "AI for All" as the governing philosophy for how India would approach AI design and deployment. Source: Regulations.AI; Ikigai Law, 2018.

What #AIforAll lacked was money and a delivery mechanism — it was a strategy document, not a funded programme. The IndiaAI Mission, six years later, is best understood as the moment that 2018 philosophy finally received a Cabinet-approved budget line, an implementing body, and — as this note documents — a genuinely uneven track record of turning that budget line into disbursed funds.

The March 2024 Cabinet approval, in other words, marks the point where those long-standing ambitions acquired a single, named budget line and a formal accountability structure — which is also what makes it possible, two years on, to ask a very direct question: where did the money actually go?

---

Part II — The Seven Pillars, As the Government Defined Them

The Cabinet approval document names seven components, not six. Getting this number right matters, because most casual summaries — including, at times, the government's own social media messaging — round it down or blur the boundaries between components. Source: PIB Press Release ID 2012355.

Table 2: The Seven Official Components of the IndiaAI Mission (Source: PIB Press Release ID 2012355, March 2024)

Two of these seven are frequently conflated in casual coverage: the Innovation Centre (which funds foundation-model development) and Compute Capacity (which funds the infrastructure those models train on). They are related — a team building a foundation model needs both a compute subsidy and, often, direct development support — but they are separately budgeted, separately governed pillars. Sarvam AI, discussed in detail below, is a useful case precisely because it shows both pillars operating on the same company at once.

---

Part III — Pillar by Pillar: What's Actually Been Funded

3.1 — IndiaAI Compute Capacity

This is the pillar most people mean when they talk about "the AI Mission" — and it is also the pillar with the clearest, most measurable progress.

As of mid-2026, more than 38,000 GPUs have been deployed through the IndiaAI compute ecosystem, offered to startups, researchers, and academic institutions at subsidised rates of roughly ₹65–92 per GPU-hour. Source: AICerts, DQIndia, Business Standard, 2026. For eligible projects in areas deemed of national importance, an additional government subsidy — administered by a MeitY committee — brings costs down further, in some cases below ₹100 per GPU-hour after subsidy. Source: TechDodo, Abhishek Gautam, 2026.

Context makes that number meaningful: commercial H100 GPU access from AWS, Google Cloud, or Azure runs roughly $2.50–4.00 per GPU-hour — approximately ₹210–335 at mid-2026 exchange rates. Source: Business Standard, 2026. The IndiaAI compute marketplace is, on paper, offering access at a fraction of open-market cloud pricing — precisely the "democratise the most capital-intensive layer of AI" logic the Mission was designed around.

There is also an expansion story here worth flagging honestly, because it did not fully materialise: mid-2026 discussions floated adding another 20,000 GPUs under a loosely-termed "AI Mission 2.0," alongside a broader push tied to India's growing domestic chip fabrication ambitions (including Tata's semiconductor fab). Source: EE Times, DQIndia, 2026. As Part IV of this note details, that expansion talk did not survive intact into the FY 2026–27 budget.

HOW TO ACCESS IT — Subsidised GPU compute for your own project

Portal: compute.indiaai.gov.in (the official IndiaAI Compute Portal)

Who is eligible: Researchers and academic faculty, IndiaAI fellowship awardees, students at recognised institutions, DPIIT-registered startups, MSMEs, and government/public-sector entities. Source: IndiaAI Compute Portal; TechDodo, 2026.

The steps:

1. Register on the portal using Meri Pehchaan — sign in through your DigiLocker, Parichay, or ePramaan account. Fill the registration form and upload documents establishing your eligibility category (student ID, institutional affiliation, DPIIT certificate, etc.). A designated verifying official reviews it.
2. Submit a project proposal once verified — this includes your technical approach, the envisioned impact, and an estimated bill of materials (how many GPU-hours you need).
3. Approval: If you request under 5,000 GPU-hours, it is auto-approved. Above 5,000 hours, a committee (PMEC) reviews it. Submit requests between the 1st and 25th of a month; the approved list is published on the 10th of the following month.
4. Subsidy: Approved projects of national importance receive a government subsidy (bringing effective rates below the ₹65–92/hour range); you bear any cost beyond the subsidised portion.

Full end-user terms are published as an official PDF on the IndiaAI S3 documentation store, linked from the portal.

3.2 — IndiaAI Innovation Centre: Twelve Funded Teams (Not Just Sarvam)

Sarvam AI gets almost all the headlines here — but it is one of twelve funded teams, and it is not the largest recipient. That distinction belongs to BharatGen, led by IIT Bombay, whose ₹1,058.52 crore allocation is roughly four times the next-highest team's. Source: Medianama, "Centre Funds 12 AI Projects for Sovereign Models," April 2026. If you only know Sarvam, you are missing most of this pillar.

The full list of twelve funded teams and what each is building (allocation figures are total approved compute/project value; the cash subsidy is a portion of each):

Table 3: The Twelve IndiaAI Innovation Centre Teams and Their Allocations (Source: Medianama, DD News, Inc42, 2026)

Where to follow their work: Sarvam (sarvam.ai), Gnani AI (gnani.ai), Gan AI (gan.ai), and Soket AI (soket.ai) publish models and updates on their own sites; BharatGen is a consortium effort anchored at IIT Bombay and releases its open models publicly. The official selection listing sits under the IndiaAI Innovation Centre hub on indiaai.gov.in — the authoritative source if any company link above changes.

IMPORTANT — this funding round is currently CLOSED. Unlike the compute and datasets pillars, the Innovation Centre is not an open door you can walk through today. Its Call for Proposals ran across three phases and closed on 30 April 2025 (506 proposals received, 12 teams selected). As of mid-2026, no new application round is open — official notices state only that "further dates for submission of proposals will be announced subsequently." If you want to build a foundation model with Mission support, the realistic action is to watch the IndiaAI Innovation Centre hub for the next call, and in the meantime use the two pillars that are open — Compute and AIKosh — to build the track record that a future proposal would rest on.

Sarvam as the instructive case study. It is worth walking through Sarvam in detail not because it is the biggest, but because its numbers are the most fully public — showing exactly how the subsidy mechanism works. Sarvam received access to 4,096 Nvidia H100 GPUs for six months, with a total compute bill of ₹246.71 crore, of which the government subsidised ₹98.68 crore — the largest single cash subsidy disclosed in the Mission's early round. Source: Founderpin, Medianama, 2025–2026. In February 2026 it open-sourced two foundation models trained entirely on that compute: Sarvam 30B (a 32B-parameter Mixture-of-Experts model, ~2.4B active parameters per token, 65K context) and Sarvam 105B (106B-parameter MoE, ~10B active per token, 128K context). Source: explainx.ai, Outlook Business, February 2026.

The commercial trajectory since has been striking: Sarvam closed the first tranche of a Series B round, raising $234 million toward a targeted $300 million round at a valuation of approximately $1.5 billion — India's first AI unicorn built substantially on public compute support, with HCLTech leading via a $150 million investment. Source: Business Standard, Forbes India, June 2026. There is also reporting that the government could end up holding a 1–2% minority stake in Sarvam through how the subsidy is structured. Source: Medianama, June 2026. The lesson for a researcher or founder reading this is not "become Sarvam" — it is that this pillar is a real, contestable funding channel with eleven other awardees. But it is a periodic channel, not a standing one: the door opens only when a new Call for Proposals is announced, and as of mid-2026 that door is shut. Plan around the next window, not around walking in today.

3.3 — IndiaAI Datasets Platform (AIKosha)

The Datasets Platform, branded AIKosha, is designed to solve a structural problem for Indian AI builders: high-quality, India-specific, non-personal training data is scattered, siloed, or simply doesn't exist in usable form for many domains and Indic languages. AIKosha is built as a unified, anonymised dataset repository meant to give startups, researchers, and government bodies a common place to access and contribute datasets. Source: IndiaAI Mission documentation, 2024–2026.

This pillar is, by design, harder to measure with a single headline number than Compute or FutureSkills — its output is infrastructure and access, not a count of trained candidates or GPUs deployed — but it underpins nearly every other pillar's long-term success. A foundation model trained under the Innovation Centre pillar, or an application funded under the Application Development pillar, is only as good as the data available to train and fine-tune it.

HOW TO ACCESS IT — Free datasets, models, and notebooks on AIKosh

Portal: aikosh.indiaai.gov.in — the platform is branded AIKosh (also written AIKosha).

What's on it: A searchable repository of non-personal datasets, pre-trained models, and use cases, plus a built-in Notebook environment (aikosh.indiaai.gov.in/home/notebook) for working with data directly in-browser. Datasets span text, audio (WAV, MP3, FLAC), video (MP4, MKV), images (JPEG, PNG, TIFF), and structured formats (CSV, TSV, Parquet, Excel, JSON, XML). Source: AIKosh User Manual & FAQs, 2026.

How to register and download:

1. Click Register on the homepage. As an individual, choose individual sign-up, verify your phone via OTP, and complete your profile (name, email, phone). As an organisation, a single point of contact (SPOC) first registers as an "Explorer," then initiates org onboarding with one supporting document (GST, PAN, CIN, UDYAM, or TAN); on approval the SPOC becomes Organisation Admin.
2. Browse the Datasets page and download. Open datasets are downloadable by any registered user; Restricted datasets are visible to all but need the contributor's explicit approval to download; Private datasets are contributor-only.
3. To contribute your own dataset: verified contributors use the "Upload Dataset" button on the Datasets page, fill in the metadata, and set the access level. There is also an open Expression of Interest (EOI) process for contributing datasets and AI artefacts, announced on indiaai.gov.in.

3.4 — IndiaAI Application Development Initiative

This pillar funds the "last mile" of the Mission: AI applications addressing real problem statements submitted by Central Ministries, State Departments, and other public institutions, with the explicit goal of catalysing large-scale socio-economic transformation rather than funding AI research for its own sake. Source: IndiaAI Mission documentation, 2026.

Approved use cases under this pillar include AI4Pragati, aimed at enhancing governance and public-service access, and Citizen Connect 2047, a citizen-services initiative. As of the Mission's most recent public reporting, roughly 30 AI applications have been approved for development under this pillar. Source: IndiaAI Mission, PIB, 2026. The pillar's structural bet is that AI adoption inside government — not just inside startups — is itself a meaningful outcome, and one that Indian citizens will interact with directly.

HOW TO ENGAGE — Build a solution against a real government problem statement (periodic — currently between calls)

Portal: indiaai.gov.in/hub/indiaai-application-development-initiative

This pillar runs on problem statements sourced from Central Ministries, State Departments, and public institutions, released as periodic challenges rather than a standing application window. The most recent, the IndiaAI Innovation Challenge 2026 (in partnership with the Ministries of AYUSH and MSME), closed on 22 February 2026 — shortlisted teams received ₹25 lakh at the pilot stage, with selected teams eligible for a two-year contract of up to ₹1 crore. As of mid-2026 the challenge is closed; watch the portal for the next round. If you are a builder looking for a well-scoped, real-world problem with a government adoption path attached, this is the pillar to monitor.

3.5 — IndiaAI FutureSkills

Of all seven pillars, FutureSkills has produced the largest raw participation numbers — and arguably the clearest evidence that the Mission's "Make AI Work for India" half of its mandate is reaching beyond a small circle of well-funded startups.

As of February 2026, 2.62 million candidates have registered on the FutureSkills PRIME platform, with 16.65 lakh (1.665 million) enrolled in or having completed training across AI, big data analytics, IoT, and cybersecurity courses. Source: Storyboard18, February 2026. More than 290 fellowships have been awarded under the initiative, and the Mission has identified 31 AI labs and 174 ITIs/Polytechnics — many in tier-2 and tier-3 cities — as sites for hands-on AI skilling. Source: Storyboard18; IndiaAI FutureSkills documentation, 2026.

Union Minister Shri Ashwini Vaishnaw has publicly cited enrolment figures for this platform as evidence of the Mission's reach beyond metro-city talent pools. Source: IndiaAI.gov.in, "8.6 Lakh Candidates Have Already Enrolled in the Future Skills Platform," 2025 (an earlier snapshot than the February 2026 figures above, useful for showing the platform's growth trajectory).

HOW TO ACCESS IT — Courses, fellowships, and AI labs

Portals: The IndiaAI FutureSkills hub (indiaai.gov.in/hub/indiaai-futureskills) is the entry point; skilling courses run through the FutureSkills PRIME platform (a MeitY–NASSCOM programme). Fellowships for postgraduate and PhD students, and the network of 31 AI labs across ITIs and polytechnics, are announced through the same hub. If you are a student, faculty member, or working professional wanting structured AI training — including in tier-2/tier-3 cities — this is the most accessible of the seven pillars, with no funding application or compute proposal required to begin.

This is the pillar most directly aligned with AIVidhya4Sarvam's own mission — the idea that AI capability should not remain concentrated in five metro cities and a handful of elite institutions. It is also, on the evidence, the pillar where the Mission's money appears to be reaching furthest beyond that circle.

3.6 — IndiaAI Startup Financing

This pillar is conceptually straightforward: streamline access to funding for deep-tech AI startups building "futuristic" AI products, addressing the well-documented gap between what Indian deep-tech founders need (patient, technically-informed capital) and what much of India's venture ecosystem is set up to provide (faster-return, application-layer bets). Source: IndiaAI Mission documentation, 2026.

Public reporting on this specific pillar's disbursement numbers — how many startups funded, at what ticket sizes, through what instrument — is thinner than for Compute or Innovation Centre, a pattern that recurs across several of the Mission's less headline-grabbing pillars and one that itself supports the broader disbursement-transparency critique raised in Part IV.

It is worth distinguishing this pillar from an adjacent, larger, and better-publicised scheme it is sometimes confused with: the Startup India Fund of Funds 2.0, a separate ₹10,000 crore initiative managed by SIDBI (the Small Industries Development Bank of India), which supports early-stage and deep-tech startups broadly — AI included, but not AI-exclusive — through SEBI-registered Alternative Investment Funds. SIDBI had committed roughly ₹11,808 crore to such funds as of December 2025. Source: IncorpX; BusinessConnect India, 2026. The two schemes are complementary — a deep-tech AI startup can plausibly draw on both — but they are not the same programme, and coverage that treats the larger Fund of Funds number as if it were the IndiaAI Mission's own Startup Financing pillar overstates what this specific pillar has actually committed.

HOW TO ACCESS IT — Funding routes for a deep-tech AI startup (periodic cohorts — currently between rounds)

Portal: indiaai.gov.in/hub/indiaai-startup-financing. This pillar too runs in windows, not as a standing fund: for example the IndiaAI Startups Global international acceleration programme (with Station F and HEC Paris) opened its Cohort 2 application window from 15–28 June 2026 — offering 10 startups up to ₹5 lakh, investor connections, and a Paris residency — and closed. A DPIIT startup recognition (via startupindia.gov.in) is the near-universal prerequisite; it unlocks self-certification and eligibility across most government startup schemes, including this one and the separate SIDBI Fund of Funds. Deep-tech AI founders should treat these as distinct applications to pursue when their windows open, and watch the portal (and the gm-startups@indiaai.gov.in contact) for the next cohort.

3.7 — Safe & Trusted AI

The seventh pillar funds the Mission's responsible-AI infrastructure: bias mitigation, deepfake detection, and the establishment of an IndiaAI Safety Institute. As of February 2026, 13 Responsible AI projects have been cleared for funding under this pillar. Source: Storyboard18, February 2026.

This pillar receives less public attention than Compute or Innovation Centre, but its existence at all is notable in comparative terms: not every national AI strategy funds safety and trust infrastructure as a named, budgeted pillar in its own right, rather than treating it as a compliance afterthought layered on top of capability-building pillars.

HOW TO ENGAGE — Responsible-AI project funding (periodic EoI rounds — check for the current window)

Portal: indiaai.gov.in/hub/safe-trusted-ai. Access runs through Expression of Interest (EoI) rounds, not a standing application. Round 1 selected 8 projects; a second EoI round was subsequently opened, targeted at Indian academic institutions, on themes including watermarking and labelling, ethical-AI frameworks, AI risk assessment, stress-testing tools, and deepfake detection. There is also an ongoing Call for Partnerships for the IndiaAI Safety Institute. Researchers and institutions working on responsible AI should watch this hub for the next EoI window.

---

Part IV — The Honest Number: Budget Approved vs. Money Actually Moved

Here is the part of the Mission's story that gets the least attention, and arguably matters the most for judging whether the seven pillars above are living up to their mandates.

₹10,371.92 crore is the total approved outlay over five years — a ceiling, set at Cabinet level in March 2024, not an annual budget. Source: PIB Press Release ID 2012355. The actual money moves through annual Union Budget allocations, and those have told a more complicated story than the headline number suggests.

For the fiscal year prior to 2026–27, ₹2,000 crore was allocated to the Mission. Actual utilisation came in significantly lower — reported at roughly ₹800 crore, well under half the allocated amount. Source: AICerts News, "India Budget 2026 Trims AI Outlay but Amps Up Infrastructure," 2026. Reporting from Medianama in April 2026 put the cumulative total released across the Mission's first two years at only around ₹400 crore — a small fraction of the ₹10,371.92 crore five-year outlay. Source: Medianama, "IndiaAI Mission: Only Rs 400 Crore Released in Two Years," April 2026.

Set against that utilisation record, the Union Budget for 2026–27 allocated ₹1,000 crore to the Mission — down from the ₹2,000 crore allocated (and substantially underspent) the year before. Source: ThePrint; BusinessToday; Angel One, February 2026. This arrived despite mid-year signals, discussed in Part III, of a far larger "Mission 2.0" expansion — with some industry voices, including Yotta CEO Mr. Sunil Gupta, publicly arguing that the Mission needed to scale toward ₹20,000 crore to keep pace with surging AI compute demand. Source: BusinessToday, "Budget 2026 Must Scale IndiaAI Mission to Keep Pace with Surging AI Compute Demand," January 2026. That scale-up did not appear in the final budget.

None of this means the Mission has failed — the Compute Capacity and Innovation Centre pillars, in particular, show real, measurable, internationally-noticed output (38,000+ GPUs deployed; a homegrown AI unicorn trained on subsidised compute). But there is a meaningful gap between the ₹10,371.92 crore figure that gets quoted in almost every article about the Mission, and the roughly ₹400 crore that had actually reached the ground two years in. That gap is not a reason for cynicism about the Mission's direction — it is a reason to track disbursement, not just announcements, when judging its progress from here.

4.1 — The Question Nobody Publishes: Are the Resources Actually Being Used?

There is a second, subtler gap hiding underneath the disbursement gap — and it points a finger not only at the government, but at the research community itself.

Notice what gets announced and what does not. The supply side of the compute pillar is published loudly and often: 38,000+ GPUs deployed, three times the original 10,000 target, with the government now targeting 100,000 GPUs by end of 2026. Source: AICerts, TechDodo, 2026. The demand side — how many of those GPU-hours are actually being consumed, how many approved projects genuinely ran, what the utilisation rate of that expensive silicon really is — is conspicuously absent from public reporting. IndiaAI has said it will "publish hourly utilisation reports" only from a coming quarter. Source: AICerts, 2026. In other words: we are told, in detail, how much capacity was created. We are not yet told how much was used. That asymmetry is itself a finding.

Three proxies are all the hard data currently available, and each points the same way:

• Budget utilisation: of the ₹2,000 crore allocated in the prior fiscal year, only ~₹800 crore was actually spent — roughly 40%. Source: AICerts, 2026. Money set aside but not deployed is, indirectly, capacity offered but not taken up.
• The FutureSkills conversion gap: 2.62 million people registered on FutureSkills PRIME, but 1.665 million actually enrolled or trained — meaning roughly 36% signed up and then did not convert into actual usage. Source: Storyboard18, February 2026. Even when access is free, registration is not the same as use.
• The global backdrop: across enterprises worldwide, average GPU utilisation sits near 5% — roughly 95% of provisioned accelerator capacity goes unused. Source: Cast AI, "State of Kubernetes Optimization Report," 2026. India's subsidised fleet is a different context, but the global number is a sobering reminder that "GPUs deployed" and "GPUs usefully used" are very different metrics.

The honest reading is that the bottleneck may not be only the government's disbursement discipline — it may also be uptake. The government has, on the evidence, built more compute capacity, faster, than it originally promised. Whether India's researchers, faculty, startups, and students are registering for and actually using that capacity at the same pace is the question the utilisation reports — when they finally arrive — will answer. Until then, the most useful thing any individual reader can do is move themselves from the "available but unused" column into the "actually using it" one. The registration links in Part VII are, quite literally, how that column gets changed.

---

Part V — Beyond the Money: What Experts Say Is Still Missing

The disbursement gap in Part IV is a funding-execution critique. A separate, more structural critique comes from researchers and policy analysts who argue that even fully-disbursed money would not, by itself, close India's most important AI gaps — because those gaps are in data quality and research talent, not primarily in GPU count.

One frequently-cited formulation of this critique, from AI-sector commentary in 2026, is blunt: "50,000 H100s will not save you if you have a thin training corpus for Indic language." Source: explainx.ai, "India Sovereign AI Status 2026," 2026. The IndiaAI Mission directs a substantial share of its budget toward compute infrastructure — the Compute Capacity pillar's 38,000+ GPUs are the Mission's single most visible achievement — while the Datasets Platform pillar, by comparison, has produced far less public reporting on its output, funding, or timeline. Whether that imbalance reflects a genuine strategic priority or simply reflects which pillar is easier to measure and announce is, on the public record, difficult to say with confidence.

A related critique concerns research talent rather than infrastructure. Analysts note that India faces a shortage of frontier AI researchers working specifically on new model architectures, data curation methods, and model interpretability — specialised research capacity that broad-based AI literacy programmes, however large their enrolment numbers, are not designed to produce. Source: explainx.ai, 2026. The FutureSkills pillar's 2.62 million registrations are a genuine achievement in breadth of AI literacy; they are a different achievement from building the depth of researchers capable of designing next-generation model architectures.

This connects to a second, longer-running structural problem: brain drain. India's most technically skilled AI researchers disproportionately end up at Google, Meta, Microsoft, or OpenAI in the United States — driven by compensation gaps, by research-culture concentration in the US and UK, and by visa pathways that make emigration for top AI talent comparatively frictionless. Source: American Bazaar reporting on Stanford AI Index findings; theprint.in, 2026. The Stanford AI Index itself has flagged India as leading globally in AI talent concentration growth while simultaneously flagging brain drain and talent anxiety as live concerns — the two facts sit side by side in the same report. Source: ThePrint, "India Leads in AI Talent, But Also Brain Drain & Anxiety, Says Stanford's AI Index Report," 2026. The Mission's startup grants and compute subsidies help at the margins — a founder with subsidised compute access has one more reason to build in India rather than relocate — but analysts are clear that these incentives do not yet reverse the structural pull toward Silicon Valley research labs. Source: explainx.ai, 2026.

The Carnegie Endowment for International Peace has framed this most directly, describing data, talent, and R&D as "the missing pieces in India's AI puzzle" and arguing that without dedicated attention and resourcing behind these three enabling elements specifically, India risks falling short of its own stated AI ambitions — regardless of how well the Compute Capacity pillar performs. Source: Carnegie Endowment for International Peace, "The Missing Pieces in India's AI Puzzle: Talent, Data, and R&D," February 2025.

There is also a nearer-term, more operational version of this same gap, visible in employer surveys: Indian employers report AI and data-competency gaps in the 38–42% range among their existing workforce, with demand for advanced AI specialists rising faster than the supply of experienced professionals able to fill those roles. Source: WION News; industry workforce surveys, 2026. This is the FutureSkills pillar's central challenge in miniature — registering 2.62 million people is a strong top-of-funnel number, but converting registration into the kind of advanced, employer-ready competency that closes a 40% skills gap is a materially harder problem than the enrolment figures alone suggest.

None of this amounts to a case that the Mission's architecture is wrong. Compute, models, data, applications, skills, startup capital, and safety are, on paper, a reasonably complete list of what a national AI strategy needs to fund. The expert critique is narrower and more specific: that the balance across those seven pillars — how much attention, disbursement, and public accountability each one receives — currently tilts toward what is easiest to measure and announce (GPUs deployed, models released) and away from what is hardest but arguably most decisive over a ten-year horizon (data depth, research talent retention, and applied R&D capacity).

---

Part VI — How India's Bet Compares Globally

Placed next to the AI budgets of the other major AI powers, India's Mission is, in absolute terms, small — and understanding why requires being honest about what is actually being compared.

China invested an estimated ¥890 billion (~$125 billion) in AI in 2026 — 18% year-over-year growth, and roughly 38% of global AI investment by some estimates. Government funding made up an estimated ¥345 billion of that (around 39% of the total). Beyond that, China announced a $138 billion National Venture Capital Guidance Fund at the March 2025 National People's Congress, structured to mobilise capital over 20 years through local governments and private partners. Source: Second Talent; FourWeekMBA; Axis Intelligence, 2026.

The United States saw $285.9 billion in private AI investment in 2025, with annual AI venture investment running at $60–70 billion. Source: PatentPC; Quantumrun, 2026. Much of this is private capital rather than direct federal budget allocation — a structurally different model from India's Cabinet-approved government mission.

The European Union has committed roughly €110 billion through its InvestAI initiative aimed at closing the gap with the US, with France alone announcing a €109 billion (~$120 billion) national AI investment plan in February 2025 — the largest single government AI commitment in Europe. EU-wide annual AI venture investment, however, sits at only about $7–8 billion, four to ten times smaller than the US figure. Source: Euronews; IW Köln; Atlantic Council, 2026.

Against all three, India's ₹10,371.92 crore (~$1.25 billion) Mission outlay looks like a rounding error. The honest caveat that belongs alongside that comparison: China's and the US's figures blend enormous private capital markets with state funding, while India's number is a pure government-mission figure — not a measure of India's total AI investment, which includes separate and substantial private venture activity of its own. Comparing India's government mission budget to other countries' total AI investment (public and private combined) is not quite apples to apples.

Even adjusting for that, the honest reading is that India is not trying to out-spend China or the US on raw capital. Its stated bet — visible across the FutureSkills and subsidised-Compute pillars in particular — is closer to what industry commentators have called "frugal AI": stretching a comparatively small public budget by subsidising access rather than attempting to fund capability from the ground up, and prioritising reach (2.62 million FutureSkills registrations; ₹65–92/hour compute access) over raw capital intensity. Source: Rest of World, "India's Frugal AI Startups," 2026. Whether that bet compounds into durable capability over the next five years — or whether the disbursement gap documented in Part IV becomes the more decisive story — is the genuinely open question.

---

Part VII — Your Access Guide: How to Actually Use the Mission

This is the part that matters most if you are a researcher, faculty member, student, founder, or simply a curious citizen. Every pillar above is not just a policy line item — several are live facilities you can register for and use today. This table consolidates the direct entry points; the "How to Access It" boxes earlier in Part III give the step-by-step for each.

First, the single most important distinction — because it decides whether you can act today or must wait for a window:

• ✅ Continuously open (register and use right now): Compute Capacity, Datasets (AIKosh), and FutureSkills. These are standing facilities — no call to wait for.
• 🟡 Periodic calls (apply only when a window is open — most are currently closed): Innovation Centre, Application Development, Startup Financing, and Safe & Trusted AI. For these, the action today is to watch the hub and prepare, not to apply.

Table 4: Quick-Start Access Guide to the IndiaAI Mission's Seven Pillars — with honest open/closed status

The honest planning note that belongs alongside this table: four of the seven pillars are periodic — you cannot apply to the Innovation Centre, Application Development, Startup Financing, or Safe & Trusted AI on demand; you apply when a window opens, and as of mid-2026 those windows are largely closed. That makes the three continuously-open pillars — Compute, AIKosh, and FutureSkills — the ones that actually matter for most readers today. The single most useful first step for almost everyone reading this is the same: register on the compute portal and on AIKosh, and start a FutureSkills course. All three are free, all three are open right now, and all three build exactly the track record that a future Innovation Centre or Startup Financing proposal would rest on when its window reopens.

---

Conclusion

India's AI Mission is usually reduced to a single number and a single fact: ₹10,371.92 crore, approved in March 2024. That shorthand is not wrong, but it flattens a structure that is genuinely more interesting — and more uneven — than the headline suggests.

Two of the seven pillars, Compute Capacity and the Innovation Centre, have produced measurable, internationally-noticed results: 38,000+ subsidised GPUs, and a homegrown AI unicorn trained substantially on public compute. FutureSkills has reached over 2.6 million registrants, a genuine claim to the Mission's "Make AI Work for India" half of its mandate. Datasets, Applications, Startup Financing, and Safe & Trusted AI are moving more quietly, with thinner public reporting on their disbursement specifics.

And underneath all seven pillars sits the number that gets the least attention of all: roughly ₹400 crore actually released against a ₹10,371.92 crore five-year ambition, two years in — with the FY 2026–27 budget cutting the annual allocation rather than scaling it up, despite mid-year talk of a much larger "Mission 2.0."

None of this settles whether the Mission will ultimately succeed. What it does is replace a single misleading number with seven pillars' worth of specific, checkable facts — and, more usefully, with a set of doors you can actually walk through.

Because here is the point that a policy audit alone would miss: the three most useful pillars for an individual are not locked behind a grant committee at all. The four funding pillars — Innovation Centre, Application Development, Startup Financing, Safe & Trusted AI — are periodic, and most of their windows are currently closed. But the three access pillars are standing and open: the compute portal takes registrations now, AIKosh's datasets are free to download now, and the FutureSkills courses ask for nothing but your time. The single biggest waste would be for India's researchers, faculty, and students to read about the Mission as news — a thing that happens to a country — rather than as infrastructure that is already sitting there, open, waiting to be used. The gap between announcement and disbursement is real. The gap between what is open today and who actually registers for it is one that every reader of this note can personally close.

Seven pillars. Three open doors you can walk through this week — and four more worth watching for their next window. Which of the three open ones will you use first? Don't just track the Mission. Use what's already open.

---

References

Government & Official Sources

• Cabinet Approves IndiaAI Mission with Outlay of Rs. 10,371.92 Crore. PIB Press Release ID 2012355, March 2024. pib.gov.in
• Cabinet Approves Ambitious IndiaAI Mission to Strengthen the AI Innovation Ecosystem. Prime Minister of India, March 2024. pmindia.gov.in
• Cabinet Approval — Allocation of Over Rs 10,300 Crore for IndiaAI Mission. PIB Press Release ID 2012375, March 2024. pib.gov.in
• IndiaAI Compute Capacity Portal. indiaai.gov.in/hub/indiaai-compute-capacity
• IndiaAI Application Development Initiative. indiaai.gov.in/hub/indiaai-application-development-initiative
• IndiaAI Startup Financing. indiaai.gov.in/hub/indiaai-startup-financing
• IndiaAI FutureSkills Hub. indiaai.gov.in/hub/indiaai-futureskills
• IndiaAI Compute Portal (registration & GPU access). compute.indiaai.gov.in
• IndiaAI Compute Portal — End-User Policy (PDF). indiaai.s3.ap-south-1.amazonaws.com
• AIKosh Datasets Platform — Home, Datasets, User Manual & FAQs. aikosh.indiaai.gov.in
• Now Open: Expression of Interest (EOI) to Contribute Datasets and AI Artefacts to AIKosh. indiaai.gov.in, 2026.
• Call for Proposals for Building India's Foundational AI Models — Date Extended (30 April 2025). indiaai.gov.in, 2025. (Innovation Centre call — closed; no new round open as of mid-2026.)
• IndiaAI Innovation Challenge 2026 (Application Development; deadline 22 Feb 2026). indiaai.gov.in, 2026.
• IndiaAI Startups Global: International Acceleration Program — Cohort 2 (15–28 June 2026). indiaai.gov.in, 2026.
• Expression of Interest for Safe & Trusted AI Projects; Second EoI Round; Call for Partnerships — IndiaAI Safety Institute. indiaai.gov.in, 2025–2026.
• Startup India (DPIIT recognition portal). startupindia.gov.in
• "8.6 Lakh Candidates Have Already Enrolled in the Future Skills Platform": Shri Ashwini Vaishnaw. indiaai.gov.in, 2025.
• AI@Work: Driving Productivity, Jobs, and Innovation. PIB Press Note ID 157310. pib.gov.in
• National Strategy for Artificial Intelligence (#AIforAll). NITI Aayog, June 2018. niti.gov.in

Academic & Expert Commentary

• The Missing Pieces in India's AI Puzzle: Talent, Data, and R&D. Carnegie Endowment for International Peace, February 2025. carnegieendowment.org
• India Leads in AI Talent, But Also Brain Drain & Anxiety, Says Stanford's AI Index Report. ThePrint, 2026. theprint.in
• India's AI Problem: Gaps in Compute, Hardware, Infrastructure, Chips Affect Readiness. WION News, 2026. wionews.com
• National Strategy for Artificial Intelligence — Key Takeaways. Ikigai Law, 2018. ikigailaw.com

Budget & Economic Reporting

• Budget 2026 Allocates Rs 1,000 Crore for IndiaAI Mission, Pushes Data Centres and AI Upskilling. ThePrint, February 2026. theprint.in
• India Budget 2026 Trims AI Outlay but Amps Up Infrastructure. AICerts News, 2026. aicerts.ai
• Union Budget 2026: India Allocates ₹1,000 Crore to Bolster AI via IndiaAI Mission. Angel One, February 2026. angelone.in
• IndiaAI Mission: Only Rs 400 Crore Released in Two Years. Medianama, April 2026. medianama.com
• Budget 2026 Must Scale IndiaAI Mission to Keep Pace with Surging AI Compute Demand: Yotta CEO Sunil Gupta. BusinessToday, January 2026. businesstoday.in
• Union Budget 2026 Bets Big on AI as IndiaAI Mission Gets Rs 1,000 Crore Push. BusinessToday, February 2026. businesstoday.in
• Decoding the Budget for India's AI Priorities. Takshashila Institution, February 2026. takshashila.org.in
• India to Add 20,000 GPUs as AI Mission 2.0 Expands. EE Times, 2026. eetimes.com

Industry & Company Sources

• IndiaAI Mission: 34,000 GPUs at Rs 150/Hour for Startups. Abhishek Gautam Blog, 2026. abhs.in
• India's AI Compute Race: Will Subsidised GPUs Close the Capability Gap? Business Standard, June 2026. business-standard.com
• IndiaAI GPUs Access 2026: Rs 67/Hour Compute for Startups. TechDodo, 2026. techdodo.in
• Make in India Chips: How Subsidized GPUs & Tata Fab Empower Startups. DQIndia, 2026. dqindia.com
• India GPUs: National Mission Hits 38,000 Accelerators. AICerts News, 2026. aicerts.ai
• India's AI Compute Infrastructure Accelerates with GPU Surge (utilisation reports forthcoming). AICerts News, 2026. aicerts.ai
• Is India's AI Startup Race Hitting a GPU Wall? NewsDrum, 2026. newsdrum.in
• State of Kubernetes Optimization Report (average enterprise GPU utilisation ~5%). Cast AI, 2026. cast.ai
• India's Sovereign LLM. Sarvam AI Blog, 2025. sarvam.ai
• Centre Funds 12 AI Projects for Sovereign Models; BharatGen 4x the Next-Highest Allocation. Medianama, April 2026. medianama.com
• IndiaAI Mission Shortlists 12 Teams for Indigenous AI Models: Minister Updates Rajya Sabha. DD News, 2026. ddnews.gov.in
• IndiaAI Mission: Soket AI, Gnani.ai & Gan.ai to Develop Indigenous AI Models. Inc42, 2026. inc42.com
• 12 Made-in-India Foundation Models Powering the India AI Impact Transformation. Varindia, 2026. varindia.com
• Startup Funding in India 2026: Government Grants, Loans, Seed Funding & How to Raise Capital. IncorpX, 2026. incorpx.io
• Top 6 Government Schemes for Startups in India 2026. BusinessConnect India, 2026. businessconnectindia.in
• IndiaAI Mission Funds Sarvam AI to Develop Sovereign LLM. Medianama, April 2025. medianama.com
• Sarvam AI (2026): India's $1.5 Billion Sovereign AI Unicorn. Founderpin, 2026. founderpin.com
• Govt Could Get a Minority Stake in Sarvam Through IndiaAI Mission Support. Medianama, June 2026. medianama.com
• What Does Sarvam's Unicorn Status Mean for India's Sovereign AI Push? Forbes India, June 2026. forbesindia.com
• From Sovereign AI to Unicorn Status: Sarvam's Rise and What Comes Next. Business Standard, June 2026. business-standard.com
• Sarvam AI to Open-Source IndiaAI Mission's Foundational LLMs. Outlook Business, 2026. outlookbusiness.com
• India Sovereign AI Status 2026: IndiaAI Mission, Sarvam Models, Gaps & Geopolitics. explainx.ai Blog, 2026.
• 2.62 Million Register on FutureSkills PRIME; 13 Responsible AI Projects Cleared Under IndiaAI Mission. Storyboard18, February 2026. storyboard18.com
• India's Frugal AI Startups. Rest of World, 2026. restofworld.org

International Comparison Sources

• Top 50+ Chinese AI Investment Statistics 2026. Second Talent, 2026. secondtalent.com
• China's Net Investment Lead Is the Macro Foundation Under Every AI Chip Story Right Now. FourWeekMBA, 2026. fourweekmba.com
• China AI Statistics 2026: Market Size, Investment & Global Competitive Position. Axis Intelligence, 2026. axis-intelligence.com
• The AI Market in China vs. USA: Growth, Investments, and Market Share. PatentPC, 2026. patentpc.com
• AI Investment By Country Statistics 2026. Quantumrun, 2026. quantumrun.com
• AI Power Play: Can Europe Catch Up With the US and China? Euronews, January 2026. euronews.com
• What Drives the Divide in Transatlantic AI Strategy? Atlantic Council, 2026. atlanticcouncil.org
• AI Competitiveness: How the EU Compares to the US and China, Fact Sheet #13. IW Köln, 2026. iwkoeln.de

---

RESEARCH METHODOLOGY & AI ASSISTANCE

This AI Note was developed through research across government press releases, official Mission documentation, budget reporting, and industry/business news sources. AI-assisted tools, including Claude, supported research synthesis, drafting, editorial refinement, and language improvement. The overall structure, analytical framing, and interpretations reflect the author's independent judgement and are intended to encourage informed discussion rather than present definitive policy positions. Figures marked with a single source should be treated as best-available public reporting rather than officially audited data, given the IndiaAI Mission's own disbursement reporting — as this note documents — remains less granular than its announcement reporting.

---

About the Author

Bhaveshkumar Choithram Dharmani is the Founder of AIVidhya4Sarvam and works as an AI mentor, researcher, and ecosystem builder. His focus is on AI education, mentorship, and building the conditions for meaningful AI participation across India — in institutions, organisations, and communities that are not yet well-served by the current AI education ecosystem.

AIVidhya4Sarvam (aividhya.in) is an AI mentorship, innovation, and transformation organisation. It works with students, professionals, startups, and institutions to build AI capability with rigour and purpose.

Issue #02 of the India AI Journey series by AIVidhya4Sarvam.