{
 "_about": "The single source of every number and fact bharath.ai shows. build.js renders all pages from this file; tools/check.js validates it. Edit here, never in the generated HTML. Each figure carries its own date and source.",
 "edition": {
  "quarter": "Q4 2026",
  "refreshed": "2026-10-10"
 },
 "contact": {
  "domainEmail": "lionofpnw@protonmail.com"
 },
 "stats": [
  {
   "id": "papers",
   "kicker": "Research",
   "value": "7.55%",
   "shortTitle": "India’s share of the world’s AI papers",
   "title": "India’s share of the world’s AI research papers in computer science",
   "context": "In 2024, a shade above the United States (7.29%) and behind China (17.76%). 39% of the papers in the dataset carry no country, so read the ranking with care.",
   "asOf": "2024-12",
   "asOfLabel": "Data for 2024",
   "source": {
    "name": "Stanford HAI, AI Index Report 2026, chapter 1 (figure 1.6.6)",
    "url": "https://hai.stanford.edu/assets/files/ai_index_report_2026_chapter_1_research_development.pdf"
   },
   "verified": true
  },
  {
   "id": "patents",
   "kicker": "The gap",
   "value": "0.40%",
   "shortTitle": "India’s share of AI patents granted worldwide",
   "title": "India’s share of the AI patents granted worldwide",
   "context": "In 2024. China held 74.24% and the United States 12.06%, about 30 times India’s share, although the two countries publish similar shares of AI papers.",
   "asOf": "2024-12",
   "asOfLabel": "Data for 2024",
   "calc": "Our calculation: 12.06 ÷ 0.40 = 30.15",
   "source": {
    "name": "Stanford HAI, AI Index Report 2026, chapter 1 (figure 1.7.2)",
    "url": "https://hai.stanford.edu/assets/files/ai_index_report_2026_chapter_1_research_development.pdf"
   },
   "verified": true
  },
  {
   "id": "reliance",
   "kicker": "The biggest commitment",
   "value": "$110B",
   "shortTitle": "Reliance’s seven-year AI infrastructure plan",
   "title": "Reliance’s seven-year plan to build AI infrastructure in India",
   "context": "₹10 lakh crore (about $110 billion), announced at the India AI Impact Summit in February 2026. A commitment, not money already spent.",
   "asOf": "2026-02-19",
   "source": {
    "name": "TechCrunch, 19 February 2026",
    "url": "https://techcrunch.com/2026/02/19/reliance-unveils-110b-ai-investment-plan-as-india-ramps-up-tech-ambitions/"
   },
   "verified": true
  },
  {
   "id": "gpus",
   "kicker": "Compute",
   "value": "45,000+",
   "shortTitle": "GPUs in the IndiaAI Mission’s shared pool",
   "title": "GPUs of shared AI compute built up under the IndiaAI Mission",
   "context": "The graphics processors that train and run AI models, pooled for shared use, as of June 2026.",
   "asOf": "2026-06",
   "source": {
    "name": "Press Information Bureau factsheet, Semiconductor and AI Revolution, 13 August 2026",
    "url": "https://www.pib.gov.in/FactsheetDetails.aspx?id=150860&NoteId=150860&ModuleId=16&reg=3&lang=1"
   },
   "verified": true
  },
  {
   "id": "investment",
   "kicker": "Private money",
   "value": "$4.09B",
   "shortTitle": "Private investment in Indian AI companies, 2025",
   "title": "Private investment in Indian AI companies in 2025",
   "context": "Sixth in the world, behind the United States, China, the United Kingdom, France and Canada. The United States drew $285.88 billion, about 70 times as much.",
   "asOf": "2025-12",
   "asOfLabel": "Data for 2025",
   "calc": "Our calculation: 285.88 ÷ 4.09 = 69.9",
   "source": {
    "name": "Stanford HAI, AI Index Report 2026, chapter 4 (figure 4.2.8, data from Quid)",
    "url": "https://hai.stanford.edu/assets/files/ai_index_report_2026_chapter_4_economy.pdf"
   },
   "verified": true
  },
  {
   "id": "startups",
   "kicker": "New companies",
   "value": "108",
   "shortTitle": "Newly funded AI companies in India, 2025",
   "title": "Newly funded AI companies in India in 2025",
   "context": "Fourth in the world, after the United States (1,953), the United Kingdom (172) and China (161).",
   "asOf": "2025-12",
   "asOfLabel": "Data for 2025",
   "source": {
    "name": "Stanford HAI, AI Index Report 2026, chapter 4 (figure 4.2.9, data from Quid)",
    "url": "https://hai.stanford.edu/assets/files/ai_index_report_2026_chapter_4_economy.pdf"
   },
   "verified": true
  },
  {
   "id": "mission",
   "kicker": "Government",
   "value": "₹10,372 cr",
   "shortTitle": "The IndiaAI Mission budget",
   "title": "The IndiaAI Mission’s budget, approved by the Union Cabinet",
   "context": "₹10,371.92 crore over five years, approved on 7 March 2024.",
   "asOf": "2024-03-07",
   "source": {
    "name": "Press Information Bureau, Cabinet approves IndiaAI Mission, 7 March 2024",
    "url": "https://www.pib.gov.in/PressReleasePage.aspx?PRID=2012355"
   },
   "verified": true
  },
  {
   "id": "google",
   "kicker": "Big tech",
   "value": "$15B",
   "shortTitle": "Google’s first AI hub in India",
   "title": "Google’s investment in its first AI hub in India, at Visakhapatnam",
   "context": "Approximately $15 billion over 2026 to 2030 for a gigawatt-scale AI hub: Google’s largest investment in India to date.",
   "asOf": "2025-10-14",
   "source": {
    "name": "Google India blog, 14 October 2025",
    "url": "https://blog.google/intl/en-in/company-news/our-first-ai-hub-in-india-powered-by-a-15-billion-investment/"
   },
   "verified": true
  },
  {
   "id": "datacentres",
   "kicker": "Data centres",
   "value": "1,789 MW",
   "shortTitle": "Data-centre capacity in India’s top seven markets",
   "title": "Data-centre capacity running in India’s seven largest markets",
   "context": "After 178 MW was added in the first half of 2026. Another 3,860 MW is planned or under construction for 2030.",
   "asOf": "2026-06",
   "source": {
    "name": "Cushman & Wakefield, India Data Centre H1 2026 Update, 23 September 2026",
    "url": "https://www.cushmanwakefield.com/en/india/news/2026/09/india-data-centre-h1-2026-update"
   },
   "verified": true
  },
  {
   "id": "fab",
   "kicker": "Chips",
   "value": "₹91,000 cr",
   "shortTitle": "India’s first commercial chip fab",
   "title": "Investment in India’s first commercial chip fab, being built by Tata Electronics at Dholera",
   "context": "About ₹91,000 crore, with Taiwan’s Powerchip (PSMC), for up to 50,000 wafer starts a month. IT Minister Ashwini Vaishnaw said in July 2026 that commercial operations are expected in mid-2028.",
   "asOf": "2026-07-15",
   "source": {
    "name": "Business Standard (Bloomberg), 17 July 2026",
    "url": "https://www.business-standard.com/amp/companies/news/tata-electronics-to-launch-india-s-first-chip-fab-using-older-technology-126071700161_1.html"
   },
   "verified": true
  },
  {
   "id": "languages",
   "kicker": "Languages",
   "value": "22 of 22",
   "shortTitle": "Indian languages in one open translation model",
   "title": "Scheduled Indian languages covered by one open translation model",
   "context": "IndicTrans2, from AI4Bharat at IIT Madras, was the first model to support all 22 languages of the Eighth Schedule, and its models are openly licensed.",
   "asOf": "2023-12",
   "source": {
    "name": "Gala et al., IndicTrans2 (arXiv 2305.16307; TMLR, December 2023)",
    "url": "https://arxiv.org/abs/2305.16307"
   },
   "verified": true
  },
  {
   "id": "jobs",
   "kicker": "Jobs",
   "value": "+20%",
   "shortTitle": "Growth in AI job postings on Naukri",
   "title": "Growth in AI and machine-learning job postings on Naukri, year on year",
   "context": "In September 2026, while postings in IT and software services fell 4%. The index counts job postings and recruiter searches, not hires.",
   "asOf": "2026-09",
   "source": {
    "name": "Naukri JobSpeak, September 2026 (Info Edge)",
    "url": "https://www.infoedge.in/pdfs/News_Events_pdfs/Naukri-Jobspeak-September-2026.pdf"
   },
   "verified": true
  }
 ],
 "world": {
  "source": {
   "name": "Stanford HAI, AI Index Report 2026 (chapters 1 and 4)",
   "url": "https://hai.stanford.edu/ai-index/2026-ai-index-report"
  },
  "note": "One source, one basis per row. The source never publishes an EU-27 figure: where it publishes \"Europe\" (which includes non-EU states such as the UK) that is shown and marked *, and where it publishes only individual countries the cell reads \"Not published\". A 2026 report carries 2024 data for patents and publications.",
  "rows": [
   {
    "metric": "AI patents granted (share of world)",
    "india": "0.40%",
    "china": "74.24%",
    "us": "12.06%",
    "europe": "2.95%*",
    "year": "2024",
    "insight": "The four shares sum to exactly 100.00%, the firmest internal check in this table. *Stanford HAI publishes “Europe”, a geographic grouping that includes non-EU states, and never an EU-27 figure."
   },
   {
    "metric": "AI publications in computer science (share of world)",
    "india": "7.55%",
    "china": "17.76%",
    "us": "7.29%",
    "europe": "11.05%*",
    "year": "2024",
    "insight": "India publishes a larger share of computer-science AI research than the United States while holding 0.40% of granted patents, by far the widest publication-to-patent gap of the four."
   },
   {
    "metric": "Private AI investment, single year (USD)",
    "india": "$4.09B",
    "china": "$12.41B",
    "us": "$285.88B",
    "europe": "Not published",
    "year": "2025",
    "insight": "The source lists individual European countries (France $4.36B, Germany $3.89B) but never an EU total, so none is shown here rather than one summed from member states."
   },
   {
    "metric": "Private AI investment, cumulative (USD)",
    "india": "$15.39B",
    "china": "$131.83B",
    "us": "$757.27B",
    "europe": "Not published",
    "year": "2013-25",
    "insight": "Cumulative over thirteen years, so not comparable with the single-year row above."
   },
   {
    "metric": "Newly funded AI companies",
    "india": "108",
    "china": "161",
    "us": "1,953",
    "europe": "Not published",
    "year": "2025",
    "insight": "India funded two-thirds as many new AI companies as China on roughly a third of China’s private investment."
   },
   {
    "metric": "Notable AI models released",
    "india": "Not listed",
    "china": "35",
    "us": "59",
    "europe": "2*",
    "year": "2025",
    "insight": "India does not appear in Epoch AI’s list, which runs down to a count of one. The report calls it a select group, so this reads as not listed rather than zero."
   }
  ]
 },
 "money": [
  {
   "who": "Reliance Industries",
   "usd": "$110B",
   "inr": "₹10,00,000 Cr",
   "what": "AI Infrastructure & Data Centers",
   "when": "2026-33 (7-yr)",
   "status": "Committed",
   "note": "Announced at the India AI Impact Summit; figure is Reliance’s own.",
   "url": "https://techcrunch.com/2026/02/19/reliance-unveils-110b-ai-investment-plan-as-india-ramps-up-tech-ambitions/"
  },
  {
   "who": "Amazon",
   "usd": "$48B",
   "inr": "",
   "what": "Cloud + AI Infra (whole India business; more than $21B is AI/cloud)",
   "when": "2026-30",
   "status": "Committed",
   "note": "Revised upward from an earlier $35B commitment in June 2026.",
   "url": "https://www.aboutamazon.in/news/company-news/andy-jassy-modi-amazon-investment-48-billion"
  },
  {
   "who": "Adani Group",
   "usd": "$100B",
   "inr": "",
   "what": "Renewable-powered AI Data Centers",
   "when": "by 2035",
   "status": "Planned",
   "note": "Announced two days before Reliance’s programme.",
   "url": "https://www.prnewswire.com/news-releases/adani-group-to-invest-100-billion-in-renewable-powered-ai-data-centres-by-2035-302689648.html"
  },
  {
   "who": "Microsoft",
   "usd": "$17.5B",
   "inr": "",
   "what": "Azure AI + Data Centers",
   "when": "2026-29",
   "status": "Committed",
   "note": "Microsoft’s own announcement.",
   "url": "https://news.microsoft.com/source/asia/2025/12/09/microsoft-invests-us17-5-billion-in-india-to-drive-ai-diffusion-at-population-scale/"
  },
  {
   "who": "Google",
   "usd": "$15B",
   "inr": "",
   "what": "AI Hub, Visakhapatnam (gigawatt-scale)",
   "when": "2026-30",
   "status": "Committed",
   "note": "India’s first Google AI hub; construction began Apr 2026.",
   "url": "https://blog.google/intl/en-in/company-news/our-first-ai-hub-in-india-powered-by-a-15-billion-investment/"
  },
  {
   "who": "AM Intelligence",
   "usd": "$8B+",
   "inr": "",
   "what": "Compute-as-a-Service (India, US, Finland, Malaysia)",
   "when": "multi-year",
   "status": "Committed",
   "note": "Hyderabad factory (30 MW, 9,000 Nvidia Vera Rubin systems) is the first tranche of this 1 GW programme, not the whole of it.",
   "url": "https://analyticsindiamag.com/ai-news/am-intelligence-orders-9000-nvidia-rubin-gpus-for-hyderabad-ai-factory"
  },
  {
   "who": "HCLTech",
   "usd": "",
   "inr": "₹14,257 Cr",
   "what": "AI Data Center, Bhubaneswar (with Sarvam)",
   "when": "multi-year",
   "status": "Committed",
   "note": "Joint outlay with Sarvam and the Government of Odisha; not a single-company figure.",
   "url": "https://www.hcltech.com/en-us/press-releases/hcltech-announces-ai-data-center-bhubaneswar-partnership-sarvam-and-government"
  },
  {
   "who": "Tata Electronics",
   "usd": "",
   "inr": "₹91,000 Cr",
   "what": "Semiconductor Fab, Dholera (with Powerchip)",
   "when": "commercial operations mid-2028",
   "status": "Building",
   "note": "50,000 wafer starts/month target. The IT minister said in July 2026 that commercial operations are now expected in mid-2028, against an earlier end-2026 target, starting at 90nm (TrendForce, citing Bloomberg; COR-157).",
   "url": "https://www.tata.com/newsroom/business/first-indian-fab-semiconductor-dholera"
  },
  {
   "who": "IndiaAI Mission",
   "usd": "$1.25B",
   "inr": "₹10,371.92 Cr",
   "what": "Compute, datasets, sovereign models (government)",
   "when": "2024-29 (5-yr)",
   "status": "Committed",
   "note": "Cabinet-approved outlay; common compute pool had more than 38,000 empanelled GPUs onboarded as of 25 March 2026, per the IT Ministry’s own Lok Sabha reply (up from 34,333 in May 2025).",
   "url": "https://www.pib.gov.in/PressReleasePage.aspx?PRID=2245069"
  },
  {
   "who": "Sarvam AI",
   "usd": "$234M",
   "inr": "",
   "what": "Sovereign foundation models (funding round)",
   "when": "2026",
   "status": "Active",
   "note": "First close of a $300M Series B at a $1.5B valuation, led by HCLTech.",
   "url": "https://www.hcltech.com/en-us/press-releases/sarvam-raises-234-million-first-close-300-million-series-b-15-billion-valuation"
  },
  {
   "who": "Neysa",
   "usd": "$600M equity",
   "inr": "",
   "what": "AI Infrastructure Platform (funding round)",
   "when": "2026",
   "status": "Active",
   "note": "Blackstone-led consortium; company intends to secure a further $600M of debt, together described as up to $1.2B.",
   "url": "https://www.blackstone.com/news/press/blackstone-leads-funding-of-over-1-billion-to-neysa-to-work-towards-building-indias-leading-ai-infrastructure-platform/"
  },
  {
   "who": "Wipro",
   "usd": "$1B",
   "inr": "",
   "what": "AI training, R&D, platforms (capex commitment)",
   "when": "2023-26 (3-yr)",
   "status": "Committed",
   "note": "Wipro AI360; a 2023 capital commitment, not 2025-26 services revenue.",
   "url": "https://www.wipro.com/newsroom/press-releases/2023/wipro-launches-wipro-ai360-commits-to-investing-1-billion-in-ai-over-the-next-three-years/"
  },
  {
   "who": "Infosys",
   "usd": "~$1B (annualized)",
   "inr": "",
   "what": "AI services revenue, not capital investment",
   "when": "FY27",
   "status": "Active",
   "note": "Infosys reports AI as roughly 8% of quarterly revenue; this row is a revenue estimate, not a pledge.",
   "url": "https://www.prnewswire.com/in/news-releases/infosys-ai-revenues-at-8-2-in-q1-resilient-operating-margin-of-21-1-302833364.html"
  }
 ],
 "made": [
  {
   "kind": "Speech",
   "date": "Mar 2026",
   "title": "VAANI: Capturing the language landscape for an inclusive digital India",
   "who": "ARTPARK and Dept. of Electrical Engineering, IISc Bangalore (with Google DeepMind)",
   "result": "31,255 hours of speech, 2,043 hours of it transcribed, and 289,838 images across 105 languages and 165 districts (Table 1).",
   "url": "https://arxiv.org/abs/2603.28714"
  },
  {
   "kind": "NLP",
   "date": "Mar 2026",
   "title": "Sarvam 30B and Sarvam 105B",
   "who": "Sarvam AI, trained in India on IndiaAI Mission compute",
   "result": "The company’s own benchmark page reports MMLU 90.6, GPQA Diamond 78.7 and Math500 98.6 for Sarvam 105B. These are self-reported by the lab, not an independent evaluation.",
   "url": "https://www.sarvam.ai/blogs/sarvam-30b-105b"
  },
  {
   "kind": "Multimodal",
   "date": "Feb 2025",
   "title": "Chitrarth: Bridging Vision and Language for a Billion People",
   "who": "Krutrim AI, Bangalore",
   "result": "Scores 67.90 on English and 51.5 on Hindi in BharatBench LLaVA-Bench, and 87.63 on POPE English (Table 1).",
   "url": "https://arxiv.org/abs/2502.15392"
  },
  {
   "kind": "NLP",
   "date": "Feb 2025",
   "title": "Krutrim LLM: Multilingual Foundational Model for over a Billion People",
   "who": "Krutrim AI, Bangalore (Ola Krutrim)",
   "result": "Reports 0.89 BERTScore on IndicCOPA Bengali against GPT-3.5’s 0.77 (Table 4).",
   "url": "https://arxiv.org/abs/2502.09642"
  },
  {
   "kind": "Speech",
   "date": "Sep 2024",
   "title": "IndicVoices-R: Unlocking a Massive Multilingual Multi-speaker Speech Corpus for Scaling Indian TTS",
   "who": "AI4Bharat, Dept. of CSE, IIT Madras",
   "result": "1,704 hours from 10,496 speakers across all 22 scheduled languages (Table 3). Published at NeurIPS 2024 Datasets and Benchmarks.",
   "url": "https://arxiv.org/abs/2409.05356"
  },
  {
   "kind": "NLP",
   "date": "Mar 2024",
   "title": "IndicLLMSuite: A Blueprint for Creating Pre-training and Fine-Tuning Datasets for Indian Languages",
   "who": "AI4Bharat Nilekani Centre, IIT Madras (with Sarvam AI)",
   "result": "The Sangraha corpus totals 251,321 million tokens: 64B verified, 24B unverified and 162B synthetic (Table 1).",
   "url": "https://arxiv.org/abs/2403.06350"
  },
  {
   "kind": "Speech",
   "date": "Mar 2024",
   "title": "IndicVoices: Towards building an Inclusive Multilingual Speech Dataset for Indian Languages",
   "who": "AI4Bharat, IIT Madras (with Sarvam AI)",
   "result": "7,348 hours from 16,237 speakers across 145 districts and 22 languages (Table 1).",
   "url": "https://arxiv.org/abs/2403.01926"
  },
  {
   "kind": "NLP",
   "date": "Nov 2023",
   "title": "Mukhyansh: A Headline Generation Dataset for Indic Languages",
   "who": "LTRC, KCIS, IIIT Hyderabad",
   "result": "3.39 million article-headline pairs across 8 languages; the best baseline averages 31.43 ROUGE-L (Table 2).",
   "url": "https://arxiv.org/abs/2311.17743"
  },
  {
   "kind": "NLP",
   "date": "May 2023",
   "title": "IndicTrans2: Towards High-Quality and Accessible Machine Translation Models for all 22 Scheduled Indian Languages",
   "who": "AI4Bharat Nilekani Centre, IIT Madras",
   "result": "Gains 5.2 chrF++ over IndicTrans1 on English-to-Indic and 6.3 on Indic-to-English, and beats the best commercial and open baselines on English-to-Indic (Table 12, IN22-Gen). Published in TMLR, December 2023.",
   "url": "https://arxiv.org/abs/2305.16307"
  }
 ],
 "languages": {
  "note": "All 22 scheduled languages have IndicTrans2 translation coverage (462 direct Indic-to-Indic directions) and IndicVoices-R speech data. Speaker counts are first-language speakers in the 2011 Census (Table C-16).",
  "sources": [
   {
    "name": "Census of India 2011, Table C-16 (speaker counts, as cited on the site since September 2026)",
    "url": "https://sourcelatam.thesourcemediaassets.com/2024/02/C-16_2011.pdf"
   },
   {
    "name": "Census of India 2011, Language Atlas",
    "url": "https://censusindia.gov.in/nada/index.php/catalog/42561"
   },
   {
    "name": "IndicTrans2, AI4Bharat (arXiv:2305.16307)",
    "url": "https://arxiv.org/abs/2305.16307"
   },
   {
    "name": "IndicVoices-R, AI4Bharat (arXiv:2409.05356)",
    "url": "https://arxiv.org/abs/2409.05356"
   }
  ],
  "rows": [
   {
    "lang": "Hindi",
    "speakers": "528,347,193",
    "translation": true,
    "speech": true,
    "note": "Census’s ‘Hindi’ grouping includes 56 related mother tongues (Bhojpuri, Rajasthani, Chhattisgarhi, Magahi, Haryanvi and others); the mother tongue named ‘Hindi’ alone is 322,230,097."
   },
   {
    "lang": "Bengali",
    "speakers": "97,237,669",
    "translation": true,
    "speech": true,
    "note": ""
   },
   {
    "lang": "Marathi",
    "speakers": "83,026,680",
    "translation": true,
    "speech": true,
    "note": ""
   },
   {
    "lang": "Telugu",
    "speakers": "81,127,740",
    "translation": true,
    "speech": true,
    "note": ""
   },
   {
    "lang": "Tamil",
    "speakers": "69,026,881",
    "translation": true,
    "speech": true,
    "note": ""
   },
   {
    "lang": "Gujarati",
    "speakers": "55,492,554",
    "translation": true,
    "speech": true,
    "note": ""
   },
   {
    "lang": "Urdu",
    "speakers": "50,772,631",
    "translation": true,
    "speech": true,
    "note": "First public open-source TTS data for this language (IndicVoices-R)."
   },
   {
    "lang": "Kannada",
    "speakers": "43,706,512",
    "translation": true,
    "speech": true,
    "note": ""
   },
   {
    "lang": "Odia",
    "speakers": "37,521,324",
    "translation": true,
    "speech": true,
    "note": ""
   },
   {
    "lang": "Malayalam",
    "speakers": "34,838,819",
    "translation": true,
    "speech": true,
    "note": ""
   },
   {
    "lang": "Punjabi",
    "speakers": "33,124,726",
    "translation": true,
    "speech": true,
    "note": ""
   },
   {
    "lang": "Assamese",
    "speakers": "15,311,351",
    "translation": true,
    "speech": true,
    "note": ""
   },
   {
    "lang": "Maithili",
    "speakers": "13,583,464",
    "translation": true,
    "speech": true,
    "note": "First public open-source TTS data for this language (IndicVoices-R)."
   },
   {
    "lang": "Santali",
    "speakers": "7,368,192",
    "translation": true,
    "speech": true,
    "note": "First public open-source TTS data for this language (IndicVoices-R)."
   },
   {
    "lang": "Kashmiri",
    "speakers": "6,797,587",
    "translation": true,
    "speech": true,
    "note": "First public open-source TTS data for this language (IndicVoices-R)."
   },
   {
    "lang": "Nepali",
    "speakers": "2,926,168",
    "translation": true,
    "speech": true,
    "note": "First public open-source TTS data for this language (IndicVoices-R)."
   },
   {
    "lang": "Sindhi",
    "speakers": "2,772,264",
    "translation": true,
    "speech": true,
    "note": "First public open-source TTS data for this language (IndicVoices-R)."
   },
   {
    "lang": "Dogri",
    "speakers": "2,596,767",
    "translation": true,
    "speech": true,
    "note": "First public open-source TTS data for this language (IndicVoices-R)."
   },
   {
    "lang": "Konkani",
    "speakers": "2,256,502",
    "translation": true,
    "speech": true,
    "note": "First public open-source TTS data for this language (IndicVoices-R)."
   },
   {
    "lang": "Manipuri",
    "speakers": "1,761,079",
    "translation": true,
    "speech": true,
    "note": ""
   },
   {
    "lang": "Bodo",
    "speakers": "1,482,929",
    "translation": true,
    "speech": true,
    "note": ""
   },
   {
    "lang": "Sanskrit",
    "speakers": "24,821",
    "translation": true,
    "speech": true,
    "note": "First public open-source TTS data for this language (IndicVoices-R)."
   }
  ]
 },
 "glossary": [
  {
   "term": "Neural Network",
   "desi": "Mumbai Local Train Network",
   "definition": "Just as Mumbai’s complex train routes connect millions of people with decision-making at each junction, a neural network is a system of interconnected ‘neurons’ that process information layer by layer, learning patterns along the way.",
   "icon": "🚄"
  },
  {
   "term": "Backpropagation",
   "desi": "Chai Wala’s Feedback Loop",
   "definition": "A street chai walah adjusts his recipe based on customer feedback: more sugar for some, less for others. Similarly, backpropagation sends error signals backward through a neural network, adjusting the ‘recipe’ (weights) to improve predictions.",
   "icon": "📉"
  },
  {
   "term": "Machine Learning",
   "desi": "Learning from Experience",
   "definition": "Like a cricket player improving through practice and experience rather than memorizing rulebooks, machine learning teaches computers to improve from data without explicit instructions for every scenario.",
   "icon": "🎓"
  },
  {
   "term": "Overfitting",
   "desi": "Cramming for Exams",
   "definition": "A student who memorizes exam questions perfectly might fail when asked different questions on the same topic. Overfitting is when an AI model memorizes training data too well and fails on new, unseen data.",
   "icon": "📚"
  },
  {
   "term": "Gradient Descent",
   "desi": "Finding Your Way Down a Hill Blindfolded",
   "definition": "Imagine navigating down a dark hill by feeling the ground slope: you take small steps in the steepest direction. Gradient descent is how AI models ‘walk’ down the error landscape to find the best solution.",
   "icon": "⛰️"
  },
  {
   "term": "Tokenization",
   "desi": "Breaking Hindi Into Words",
   "definition": "Just as Hindi sentences break into meaningful words and phrases, tokenization breaks text into smaller pieces (tokens) that AI can understand and process individually.",
   "icon": "🔤"
  },
  {
   "term": "Embedding",
   "desi": "Mapping People on a Train",
   "definition": "If you map every person in a train by their position, wealth, and age in 3D space, words in AI work similarly: each word gets a unique position in a mathematical space where similar words are close together.",
   "icon": "📍"
  },
  {
   "term": "Attention Mechanism",
   "desi": "Focusing in a Crowded Market",
   "definition": "In a bustling Indian market, your attention shifts to what’s relevant: a vendor’s call if you need produce, a sound if there’s danger. Attention in AI lets models focus on the most relevant information from vast input.",
   "icon": "👁️"
  },
  {
   "term": "Transformer",
   "desi": "A Panchayat, Not a Relay Race",
   "definition": "In a relay race only the runner holding the baton knows where the race stands, and older language models read that way, one word at a time. A transformer works more like a panchayat where everyone hears everyone at once: each word looks at every other word in parallel and weighs how much each matters to it. That parallel reading is what lets transformers learn from vast amounts of text quickly.",
   "icon": "🏃"
  },
  {
   "term": "Fine-tuning",
   "desi": "Adapting a Bollywood Script",
   "definition": "A Bollywood script for Tamil cinema gets adjusted for local humor, references, and culture; the core story remains but details shift. Fine-tuning adapts pre-trained AI models for specific tasks.",
   "icon": "🎬"
  },
  {
   "term": "Large Language Model (LLM)",
   "desi": "A Pandit Who Read Every Book",
   "definition": "An LLM is like a learned pandit who has read millions of texts and learned to predict what comes next in any conversation, having absorbed patterns from vast knowledge.",
   "icon": "📖"
  },
  {
   "term": "Prompt Engineering",
   "desi": "Asking a Google Voice Assistant",
   "definition": "The way you ask ‘Hey Google, what’s the weather?’ vs ‘It’s raining outside, how should I dress?’ changes the answer. Prompt engineering is the art of asking AI precisely what you need.",
   "icon": "🎤"
  },
  {
   "term": "Mixture of Experts (MoE)",
   "desi": "The Indian Cricket Selection Committee",
   "definition": "Like how India’s selection committee assigns different specialists for different conditions (pace bowlers for Australia, spinners for Sri Lanka), MoE routes inputs to specialized sub-networks based on the task. Only a few experts work on each input, so a very large model costs far less to run for every word it reads.",
   "icon": "🏏"
  },
  {
   "term": "Knowledge Graphs",
   "desi": "The Dharavi Social Network",
   "definition": "In Dharavi, everyone knows everyone through connections: the leather guy knows the potter who knows the exporter. Knowledge graphs connect facts through relationships, so a system can answer a question such as which languages are spoken in which state by following the links.",
   "icon": "🕸️"
  },
  {
   "term": "Quantization",
   "desi": "Compressing a Bollywood Film for Mobile",
   "definition": "Just as a 4K Bollywood film is compressed to play on a budget phone without losing the story, quantization stores a model’s numbers at lower precision, for example 4 bits instead of 16 or 32, so it fits and runs on affordable devices.",
   "icon": "📦"
  },
  {
   "term": "Retrieval-Augmented Generation (RAG)",
   "desi": "The Open-Book Exam",
   "definition": "Instead of memorizing everything (like a closed-book exam), RAG lets AI look up relevant information from a knowledge base before answering, like a student who can check their notes during the test. It also lets the model point to the passage its answer came from.",
   "icon": "📚"
  },
  {
   "term": "Hallucination",
   "desi": "The Overconfident Pandit",
   "definition": "When an uncle at a wedding gives you confident, wrong directions to the venue, sure of every turn, that’s hallucination: AI generating plausible-sounding but factually incorrect information with complete confidence.",
   "icon": "🎭"
  },
  {
   "term": "Edge AI",
   "desi": "The Village Doctor vs City Hospital",
   "definition": "Instead of sending every patient to AIIMS Delhi (cloud), an AI model running on a local device (edge) can diagnose basic conditions right in the village: faster, cheaper, and it works offline. It matters most where the network is weak or absent.",
   "icon": "🏥"
  }
 ],
 "states": [
  {
   "state": "Andhra Pradesh",
   "hub": "Visakhapatnam",
   "note": "Google’s first India AI hub: $15B, 2026-30, construction began April 2026."
  },
  {
   "state": "Gujarat",
   "hub": "Dholera",
   "note": "Tata Electronics’ semiconductor fab (with Powerchip): ₹91,000 crore; commercial operations now expected in mid-2028, against an earlier end-2026 target."
  },
  {
   "state": "Karnataka",
   "hub": "Bangalore",
   "note": "Headquarters of Sarvam AI, Infosys and Wipro; home to IISc’s Kotak IISc AI-ML Centre (KIAC), established 2021-23."
  },
  {
   "state": "Maharashtra",
   "hub": "Mumbai",
   "note": "Reliance Industries’ $110 billion, seven-year AI infrastructure programme, the largest single commitment on the ledger; home to IIT Bombay’s Centre for Machine Intelligence and Data Science (C-MInDS)."
  },
  {
   "state": "Odisha",
   "hub": "Bhubaneswar",
   "note": "HCLTech, Sarvam and the state government’s joint ₹14,257 crore AI data centre, announced July 2026."
  },
  {
   "state": "Telangana",
   "hub": "Hyderabad",
   "note": "AM Intelligence’s compute factory, the first 30 MW tranche of a four-country, 1 GW programme, delivering Q1 2027."
  },
  {
   "state": "Tamil Nadu",
   "hub": "Chennai",
   "note": "Home to IIT Madras’ AI4Bharat lab (IndicTrans2 and other open Indic-language models) and its Wadhwani School of Data Science and Artificial Intelligence, launched November 2023."
  },
  {
   "state": "Delhi",
   "hub": "New Delhi",
   "note": "Home to IIT Delhi’s Yardi School of Artificial Intelligence and Jawaharlal Nehru University."
  },
  {
   "state": "West Bengal",
   "hub": "Kharagpur",
   "note": "Home to IIT Kharagpur’s Department of Artificial Intelligence, made a full department in 2024."
  },
  {
   "state": "Uttarakhand",
   "hub": "Roorkee",
   "note": "Home to IIT Roorkee."
  },
  {
   "state": "Assam",
   "hub": "Guwahati",
   "note": "Home to IIT Guwahati’s Mehta Family School of Data Science and Artificial Intelligence, established 2021."
  },
  {
   "state": "Uttar Pradesh",
   "hub": "Kanpur",
   "note": "Home to IIT Kanpur’s Wadhwani School of Advanced Artificial Intelligence and Intelligent Systems, launched April 2025 with the Department of Intelligent Systems."
  }
 ],
 "cricket": {
  "elo": [
   {
    "team": "Royal Challengers Bengaluru",
    "elo": 1620.8
   },
   {
    "team": "Gujarat Titans",
    "elo": 1574.1
   },
   {
    "team": "Sunrisers Hyderabad",
    "elo": 1543.1
   },
   {
    "team": "Delhi Capitals",
    "elo": 1524.9
   },
   {
    "team": "Punjab Kings",
    "elo": 1519.3
   },
   {
    "team": "Kolkata Knight Riders",
    "elo": 1518.5
   },
   {
    "team": "Rajasthan Royals",
    "elo": 1515.1
   },
   {
    "team": "Chennai Super Kings",
    "elo": 1475.7
   },
   {
    "team": "Mumbai Indians",
    "elo": 1464.2
   },
   {
    "team": "Lucknow Super Giants",
    "elo": 1461.7
   }
  ],
  "meta": {
   "source": "Cricsheet ball-by-ball IPL data (cricsheet.org), analyzed locally",
   "matches": 1243,
   "seasons": "2007/08-2026",
   "generated": "2026-07-30",
   "method": "Elo ratings (K=24) computed match by match across all IPL seasons from Cricsheet ball-by-ball data."
  }
 },
 "constitution": {
  "quote": "India, that is Bharat, shall be a Union of States.",
  "where": "Article 1(1), the Constitution of India",
  "url": "https://www.legislative.gov.in/static/uploads/2025/07/ca7ce5c746fa7480804bbdeb6cb704f0.pdf"
 }
}