The next DPI — how India can commoditise AI

1. At a Glance

• India built identity (Aadhaar), payments (UPI), and data-sharing (DEPA/Account Aggregator) as free/near-free digital public infrastructure (DPI); the argument is that AI/compute is the next layer to commoditise the same way [S4].
• India's claim to DPI leadership rests on the integration of all three rails, not any single component — no other country has built identity + payments + consented data-sharing at this scale [S4].
• The IndiaAI Mission (₹10,371.92 crore, approved 7 March 2024) is the government's vehicle to subsidise AI compute access the way UPI subsidised payments [S1].
• Relevant for GS-II (governance, DPI) and GS-III (technology, infrastructure, economy).

2. Why in the News

• Article "The next DPI — how India can commoditise AI" by Srivatsa Krishna, IAS, published in The Hindu Business Line, 31 July 2026, Chennai edition, p.12 [S4].
• Recent milestone: IndiaAI Mission's common compute facility crossed 38,000 GPUs onboarded via the AI Compute Portal, offered to startups/academia at subsidised rates (as low as ₹115.85–150/GPU-hour, effectively under ₹100/hour post-subsidy) [S2].
• Frames 2026 as the moment AI compute is being pushed toward the same "utility" status as data and payments [S4].

3. Background & Evolution

• DPI lineage: Aadhaar (identity) → UPI (payments) → DEPA/Account Aggregator (consented data-sharing) — sequential rails built over roughly a decade [S4].
• Aadhaar: enrolled 1.4 billion people, converted identity verification from a costly paper process to a low-cost API call [S4].
• UPI: processes roughly 20 billion transactions/month at near-zero marginal cost [S4].
• Data affordability: cost of 1 GB fell from ~$4 (Sept 2016) to under 30 cents (2019), among the cheapest globally; enabled by market structure allowing one player to absorb fixed costs, not subsidy [S4]. Result: ~500 million new internet users in under five years [S4].
• IndiaAI Mission: Cabinet-approved 7 March 2024 under MeitY, with a ₹4,563.36 crore compute pillar (of the ₹10,371.92 crore total) dedicated to a national shared GPU facility [S1][S2].
• Union Budget 2024-25 allocated over ₹550 crore initially to the Mission [S2].

4. Core Static Facts

5. Multi-Dimensional Analysis

• Economic: Subsidised compute lowers entry barriers for AI startups, mirroring how cheap data fuelled the 2016-19 digital economy boom [S4]; risk of fiscal burden if compute subsidy is not self-sustaining like UPI's private-sector-absorbed model [S2].
• Technological: Compute is treated as a public utility (like electricity/bandwidth) rather than a purely private-market good; national GPU pool of 38,000+ units is shared infrastructure [S2].
• Governance/Administrative: MeitY-run compute allocation via a portal introduces a new administrative function — provisioning and pricing GPU-hours for startups/academia [S2].
• Social/Equity: Rationale is inclusion — making AI access available beyond well-funded Big Tech-adjacent firms, replicating UPI's democratisation of payments [S4].
• Comparative/Geopolitical: Article explicitly benchmarks India against Estonia (identity only), Brazil's Pix (payments only), Singapore's Singpass/SGFinDex, and Europe's open banking — arguing India's edge is stack integration, not any single piece [S4].

6. Recent Developments (last 12-18 months)

• 7 March 2024: Cabinet approves IndiaAI Mission (₹10,371.92 crore) [S1].
• 30 May 2025: National compute crosses 34,000 GPUs [S1].
• 2026 (recent): GPU count rises past 38,000; lowest subsidised bids of ₹115.85/hr (standard) and ₹150/hr (high-end) reported [S2].
• June 2026: PIB publishes "Digital India 11 Years of Transformation" report referencing DPI-AI convergence [S1].
• 31 July 2026: Op-ed proposing AI/compute as "the next DPI" published in The Hindu Business Line [S4].

7. Prelims Hooks

• IndiaAI Mission was approved by the Cabinet on 7 March 2024.
• Total outlay of IndiaAI Mission: ₹10,371.92 crore over five years.
• Nodal ministry for IndiaAI Mission: MeitY (not MoS or DST).
• Compute pillar of IndiaAI Mission: ₹4,563.36 crore.
• India's national compute capacity crossed 34,000 GPUs by 30 May 2025, later exceeding 38,000.
• Subsidised GPU rates under IndiaAI Mission: as low as ₹115.85/hour (standard GPU).
• Aadhaar has enrolled 1.4 billion individuals.
• UPI processes approximately 20 billion transactions per month.
• Cost of 1 GB of data in India fell from ~$4 in September 2016 to under $0.30 by 2019.
• DEPA (Data Empowerment and Protection Architecture) works through the Account Aggregator framework for consented data-sharing.
• India's DPI trio commonly cited: Aadhaar (identity), UPI (payments), DEPA/AA (data).
• Comparator DPI systems cited internationally: Estonia (e-ID), Brazil (Pix), Singapore (Singpass/SGFinDex), EU (open banking).
• Union Budget 2024-25 gave an initial allocation of over ₹550 crore to IndiaAI Mission.

8. Mains Relevance

• GS-II: Government policies and interventions; e-governance applications, models, successes, limitations.
• GS-III: Science and technology developments and their applications; infrastructure (digital); Indian economy — growth, employment.
• Possible question stems: 1. "Digital Public Infrastructure (DPI) transformed India's identity and payments landscape. Examine whether AI compute can be commoditised on similar lines. What are the risks of treating GPU compute as a public utility?" (GS-III) 2. "Discuss how the integration of Aadhaar, UPI, and DEPA gives India a distinct advantage in digital governance compared to other countries' DPI efforts." (GS-II) 3. "Critically evaluate the IndiaAI Mission's compute subsidy model as a strategy for democratising access to Artificial Intelligence in India." (GS-III)

9. Related Topics to Study Next

• Aadhaar Act, 2016 — legal backbone of India's identity DPI.
• Unified Payments Interface (UPI) & NPCI — payments-rail case study for "free" infrastructure.
• DEPA / Account Aggregator framework — RBI-regulated consented data-sharing model.
• IndiaAI Mission pillars (compute, foundational models, datasets platform, application development, FutureSkills, safe & trusted AI) — full scheme architecture.
• India AI Governance Guidelines (Feb 2026) — regulatory framework accompanying the Mission.
• Digital India Programme — parent umbrella initiative since 2015.
• Semiconductor Mission / India Semiconductor Mission — hardware dependency behind GPU/compute ambitions.
• Global DPI comparisons (Brazil's Pix, Estonia's e-ID, Singapore's Singpass) — for comparative-governance answers.

10. Common Errors / Trap Areas

• Confusing IndiaAI Mission (2024, MeitY) with the earlier National AI Strategy/#AIforAll (NITI Aayog, 2018) — different bodies and years.
• Attributing India's cheap data prices to government subsidy — the article clarifies it was a private-market structure (one player absorbing fixed costs), not state subsidy.
• Mixing up the total Mission outlay (₹10,371.92 crore) with the compute-specific outlay (₹4,563.36 crore).
• Assuming DEPA is a statute — it is an architecture/framework implemented via RBI-licensed Account Aggregators, not a standalone Act.
• Treating GPU counts (34,000 vs 38,000) as fixed — these are point-in-time figures that keep rising; always cite the reporting date.

11. Sources

• [S1] Cabinet Approves Ambitious IndiaAI Mission — PIB — https://www.pib.gov.in/PressReleaseIframePage.aspx?PRID=2012355 — (tier: 1)
• [S2] IndiaAI Mission Expands AI Ecosystem with Affordable Compute and Startup Support — PIB — https://www.pib.gov.in/PressReleasePage.aspx?PRID=2245069&reg=3&lang=1 — (tier: 1)
• [S3] Union Budget 2024-25 allocates over 550 crores to the IndiaAI Mission — indiaai.gov.in — https://indiaai.gov.in/article/union-budget-2024-25-allocates-over-550-crores-to-the-indiaai-mission — (tier: 1)
• [S4] "The next DPI — how India can commoditise AI" by Srivatsa Krishna, The Hindu Business Line, 31 July 2026 — https://www.thehindu.com/todays-paper/2026-07-31/th_chennai/articleGP4GB1BO7-15752722.ece — (tier: 4)