According to Anthropic's India Country Brief, India accounted for 5.8% of global Claude.ai conversations in November 2025, ranking second worldwide behind the United States. However, when adjusted for working-age population, India ranked 101st out of 116 surveyed countries.
Public discussions frequently conflate consumer chat applications, administrative fraud detection, weather forecasting systems, and data annotation labour into a single adoption metric, despite significant differences in underlying data sources, methodology, and measurement criteria.
Tap to enlargeFor detailed budget, GPU allocation, and fellowship data regarding the IndiaAI Mission, see The IndiaAI Mission.
Absolute counts and per-person ranks
The IAMAI and Kantar Internet in India Report 2025, presented on 29 January 2026, estimated 958 million active internet users across the country. Among respondents in the 87,000-person sample, 44% reported using AI-enabled features including voice search, image search, chatbots, and photo filters. Reported usage stood at 57% among users aged 15–24 and 52% among those aged 25–44. The broad survey definition encompasses casual product features alongside dedicated generative AI tools.
In contrast, an April 2025 Google-Kantar study covering 8,000 respondents across 18 urban centres found that 60% of participants described themselves as unfamiliar with AI, while 31% had tried a generative tool. Because the survey definitions, target demographics, and sampling methodologies differ substantially between these reports, their findings cannot be combined into a single unified adoption percentage.
In media interviews with the Economic Times, WhatsApp head Will Cathcart and Meta CFO Susan Li identified India as Meta AI's largest global market by monthly active users, though the company has not published absolute user headcounts for the region.
Google reported that the Gemini application surpassed 1 billion monthly active users globally by August 2026, supported by expanded Indian-language capabilities and regional features like Personal Intelligence and Gemini Spark. Google has not released India-specific breakdown figures for this global user base.
Distribution partnerships also shape surface-level reach. In July 2025, Airtel offered customers a complimentary one-year subscription to Perplexity Pro across its subscriber base. TRAI subscription data recorded 1,028.61 million total internet subscribers in December 2025 and 1,087.50 million broadband subscribers by June 2026. These telecom connection tallies and promotional bundle eligibilities reflect network access rather than active AI user engagement.
In enterprise consumer applications, Swiggy's FY25 annual report details algorithmic recommendation engines and Pyng, an automated professional booking interface. Similarly, a joint report by Meta and the Retailers Association of India found that 72% of surveyed digital product discovery occurred over WhatsApp. These metrics reflect automated commerce infrastructure rather than explicit consumer use of standalone AI assistants.
Tap to enlargeAnthropic's telemetry data indicates that four states—Maharashtra (15.5%), Tamil Nadu (13.2%), Karnataka (12.7%), and Delhi (10.5%)—account for more than half of India's Claude.ai traffic, closely aligning with major IT services and software hubs. Software development tasks represented 45.2% of Indian Claude interactions (the highest country share in Anthropic's brief), and work-related queries accounted for 51.3% (compared to a 46% global average). This sample reflects one week of consumer Claude.ai web interactions and excludes API usage, enterprise deployments, and alternative model ecosystems.
Public-sector deployments and backend systems
A substantial portion of AI implementation in India operates through institutional backend workflows rather than consumer-facing chat interfaces.
In a Lok Sabha submission on 5 December 2025, the Ministry of Health and Family Welfare stated that the Clinical Decision Support System (CDSS) within eSanjeevani facilitated standardised clinical data capture for 282 million teleconsultations between April 2023 and November 2025. The ministry also noted that roughly 12 million consultations received direct algorithmic differential diagnosis suggestions, alongside specialised screening tools for diabetic retinopathy and abnormal chest radiographs. Final diagnostic and prescription authority remains with the licensed medical practitioner.
The National Health Authority's 2025–26 reporting states that its National Anti-Fraud Unit used automated claim auditing algorithms to identify and block ₹678.47 crore in suspicious transactions, resulting in ₹114.06 crore in penalties across 2,842 facilities, 2,003 hospital de-empanelments, and 839 suspensions. These agency metrics do not include public assessments of false-positive claim rejections.
In agriculture, the Ministry of Agriculture reported in an April 2025 PIB release that the Kisan e-Mitra multilingual conversational query system had handled over 92 lakh interactions, with daily query volumes exceeding 20,000 across 11 languages. Follow-up parliamentary data in December 2025 placed total processed queries above 93 lakh. The companion National Pest Surveillance System was deployed to more than 10,000 agricultural extension workers covering 61 crops and 400 agricultural pests.
In judicial translation, the Supreme Court's SUVAS machine translation tool processed 36,344 judgments into Hindi and 47,439 into other regional languages by March 2025, producing 83,783 translated rulings on the e-SCR portal (PIB note, April 2025). Official guidelines confirm that automated translations undergo human judicial verification prior to publication.
In meteorology, the Ministry of Earth Sciences reported in a July 2026 Lok Sabha response that the Bharat Forecast System operates numerical weather prediction at a 6 km grid resolution, delivering an estimated 30% accuracy improvement for extreme rainfall events over prior research baselines. Concurrently, a collaborative initiative involving Google Research and the University of Chicago integrated NeuralGCM forecasting models with IMD historical data to deliver targeted monsoon SMS alerts to roughly 38 million farmers in 2025.
In transportation, Civil Aviation data from mid-2026 recorded over 10.75 crore biometric passenger journeys through DigiYatra facial recognition gates across 39 operational airports. In telecommunications, Airtel reported that its network-level machine learning spam detection system had flagged 27.5 billion suspected spam calls between September 2024 and April 2025.
In financial services, the Reserve Bank of India's FREE-AI committee published a governance framework outlining seven operational principles and six structural pillars for financial sector AI. In banking operations, State Bank of India leadership reported automating approximately 25% of routine cheque clearing and underwriting nearly ₹1 trillion in MSME loans using algorithmic risk assessment models in FY26 (FIBAC coverage). HDFC Bank's FY26 annual report noted that its proprietary analytics platform, Neev, reduced eligibility verification latency by approximately 50% for pre-approved retail credit limits.
| Domain | Latest public record | Operational context |
|---|---|---|
| Telemedicine | 282M eSanjeevani consultations with CDSS data capture to Nov 2025 | Standardised clinical record capture; doctor validates treatment |
| Health insurance | ₹678.47 Cr suspicious claims blocked in FY26 reporting | Automated fraud auditing; false-positive rates unpublished |
| Agriculture | 93L+ cumulative Kisan e-Mitra farmer queries | Multilingual voice/text query volume |
| Courts | 83,783 vernacular judgment translations by Mar 2025 | SUVAS machine translation followed by judicial review |
| Weather | BharatFS at 6 km resolution; SMS alerts to ~38M farmers | Physics-AI hybrid model run by IMD and research partners |
| Airports | DigiYatra ~10.75 Cr passenger journeys | 1:1 facial recognition authentication at gate checkpoints |
| Spam filtering | 27.5B suspected calls flagged by Airtel by Apr 2025 | Telecom carrier packet/call metadata analysis |
Enterprise adoption and IT services revenue
Survey data indicates high adoption of generative AI tools among Indian white-collar employees, alongside varying rates of structured enterprise workflow integration.
The Microsoft and LinkedIn 2024 Work Trend Index reported that 92% of surveyed Indian knowledge workers used generative AI tools at work, compared to a 75% global average. In BCG's 2026 AI at Work survey, 97% of Indian managers reported regular AI tool usage, with 70% of frontline users estimating weekly productivity savings of one workday. Microsoft's 2025 report found that 93% of Indian enterprise leaders planned to deploy autonomous AI agents within 12–18 months, with 59% reporting active pilot automation.
These figures primarily capture self-reported feedback from salaried knowledge workers in corporate environments rather than the broader labour market.
NASSCOM's AI Adoption Index 2.0, published with IndiaAI.gov.in, evaluated 500 enterprises across seven core economic sectors, assigning India an aggregate maturity score of 2.47 on a 4-point scale in 2024 (compared to 2.45 in 2022). Roughly 87% of surveyed organizations scored within the intermediate "Enthusiast" or "Expert" bands. Salesforce's Agentic Workplace Study India (July 2026) observed that while Indian desk workers were 45% more likely than global peers to integrate AI into daily tasks, 38% of respondents reported unsuccessful enterprise AI pilots in the preceding year (versus 28% globally), with 34% attributing pilot failures to insufficient business domain context.
Tap to enlargePublicly listed IT service providers have begun reporting dedicated AI contract revenues. TCS reported annualised AI-related services revenue of $2.3 billion in Q4 FY26. Infosys stated that AI engagements accounted for 8.2% of total revenue in Q1 FY27. HCLTech posted annualised Advanced AI revenue of $620 million in FY26, while Wipro disclosed that several hundred production AI agents were deployed across its global delivery pipelines.
Tap to enlargeIn research talent metrics, the Stanford AI Index 2026 ranked India second globally for identified AI authors and inventors (50,460 in 2025), behind the United States. Relative skill penetration among Indian LinkedIn profiles stood at 3.0 times the global baseline, while student adoption of generative AI learning tools reached 84%.
Data annotation and human-in-the-loop labour
India remains a major global supplier of data labelling, multimodal annotation, and reinforcement learning human feedback (RLHF) services.
Industry projections by NASSCOM estimate that the domestic data annotation market could reach $7 billion by 2030, employing up to 1 million full-time and part-time workers. Trade estimates cited by the Economic Times indicate approximately 20,000 full-time annotation specialists working in formal facilities and over 50,000 independent contractors operating across international crowdsourcing platforms.
Field reporting from Tier-2 and Tier-3 centres highlights the operational mechanics of this sector. Accounts by AFP and TechTimes from Karur, Tamil Nadu, document Objectways technicians using wearable head cameras and smart glasses to record physical motion data for training embodied robotics models, with compensation rates around $2.62 per hour. A related investigation by the New York Times detailed how distributed annotation workforces in non-metropolitan centres generate ground-truth training datasets for international foundation models.
A labour market analysis by Nomura (Times of India, India Today) identified approximately 83,100 AI-related job additions against 31,921 AI-attributed job displacements in India between 2022 and August 2026. Hiring was predominantly concentrated in software engineering and cloud implementation roles, while workforce reductions occurred primarily within legacy customer support and routine business process outsourcing operations.
Government infrastructure initiatives and digital public goods
Official frameworks published for the February 2026 AI Impact Summit outline the national India AI stack across compute capacity, datasets, foundational models, application interfaces, and energy infrastructure (PIB India AI Stack Overview, Significance of AI for India). Comprehensive compute allocations and project expenditures are tracked in The IndiaAI Mission. Additional dataset catalogues on the AIKosh platform and sanctioned GPU hours were detailed in an August 2026 PIB release and a Lok Sabha submission on 5 August 2026.
In education and literacy, Ministry of Education submissions to the Lok Sabha recorded 7.91 lakh secondary and 50,343 senior-secondary CBSE students enrolled in formal AI curriculum electives for the 2024–25 academic year, with AI skill subjects introduced in 18,839 CBSE schools. Under the national YUVA AI for All digital literacy initiative, over 26.5 lakh individuals completed introductory modules by August 2026 (TechObserver parliamentary coverage).
Private-public infrastructure commitments have also expanded. In October 2025, Reliance and Google announced a strategic partnership providing Gemini model distribution across Jio platforms, Google Cloud TPU access, and enterprise solutions via Reliance Intelligence. Multi-billion-dollar hyperscale cloud and data centre investments by Microsoft and Amazon encompass broad enterprise infrastructure allocations alongside dedicated AI training and inference capacity.
Across citizen services, regional language translation relies heavily on open digital public goods, notably the Bhashini language repository. Voice-enabled agricultural advisories, multilingual land record lookups, and vernacular court document systems depend on these shared linguistic and acoustic models to function across regional languages.
Summary of adoption dynamics
Understanding AI adoption in India requires analyzing separate, non-overlapping segments of the technology ecosystem:
- Consumer applications: High aggregate conversation volumes driven by WhatsApp distribution (Meta AI) and intensive coding usage among software engineers (Claude.ai), contrasted with relatively low per-capita penetration across the broader population.
- Institutional public systems: Broad-scale automated data capture and administrative rule filtering—such as eSanjeevani CDSS (282 million consultations), NHA anti-fraud screening, and DigiYatra biometric gates—operating within structured institutional workflows.
- Enterprise deployment: Widespread self-reported productivity tool usage among corporate knowledge workers, balanced against measured enterprise maturity scores and ongoing challenges in business context integration.
- Global supply chain participation: A substantial and geographically distributed workforce in Tier-2 and Tier-3 centres providing computer vision labelling, RLHF feedback, and motion capture data for international foundation model developers.