印度人工智能：明确未来方向

IDENTIFYING FUTURE Directions Anulekha Nandi • Shravishtha Ajaykumar • Amoha Basrur • Prateek Tripathiwith Pranjali Goradia, Anusha Guru, Srijan Jha © 2025 Observer Research Foundation. All rights reserved. No part of this publication may be reproduced ortransmitted in any form or by any means without permission in writing from ORF. Attribution: Anulekha Nandi et al.,AI for India: Identifying Future Directions, August 2025, Observer ResearchFoundation. Editorial and Production Team: Vinia Mukherjee,Editor and Producer;Monika Ahlawat and Meryl Mammen,Assistant Editors;Rahil Miya Shaikh,Design and Layout Executive Summary ArtificialIntelligence(AI)holdstransformativepotential for the Indianeconomy,boostingproductivityandefficiency across sectors. However, realising effectivevalue from AI-driven transformations is determined bythe moving parts of the country’s AI ecosystem. Thisreportsynthesises current trends across AI adoption,investment,and innovation to provide insights forinformed policy action and business strategy. Key Takeaways: 1.AI initiatives have a scaling problem.MostAI initiatives tend to be ad-hoc and project-based,or else at the proof of concept(PoC)stagewith difficulty in scaling beyond that level. This can beattributed to systemic bottlenecks such as issues inintegrating AI with legacy IT systems, inconsistent datastandardsand inadequate data architectures,lack ofrequisite talent, and unclear governance or complianceframeworks.Seventy-fivepercentof500small,medium,and large enterprises surveyed by Nasscomhave a PoC-only strategy and 60 percent have ad-hoc or project-based funding. Meanwhile, only 13percent have a dedicated AI talent pool forexecution;and 18 percent have enterprise-wide data standards. Further, 62 percent oforganisationssurveyed by the InternationalDataCorporation(IDC)and Qlik reporttheneed to improve data governance andprivacy policies. 4.AI innovation should not missopportunities at the bottom of thepyramid. The adoption of indigenous LLMs is not atpar with their foreign counterparts. Despite alow overall adoption rate of 31 percent, Indiais the largest user of the ChatGPT mobileapp globally (13.5 percent of all users) andthethird largest user of the Deepseekmobile app (6.9 percent). On one level, thisunderscores the need to understand India’smobile-first, app-based consumption pattern;on another, it highlights the need to developdownstreaminnovationopportunitiestofacilitatethe development of applicationsforIndia’s critical social sectors and last-mile, underserved multilingual population. 2.India’s growth story suffers from aperception issue. India’srapid economic growth and largemarketsize is perceived predominantlyasa consumption story with investorscontinuingto direct their capital towardscommercial,application-specific,andlate-stageventures with defined marketopportunities.In 2024,early-stage fundingdeclinedby 37 percent from the previousyear.Patient capital for deeptech venturesthat involve commercialisation of research-driven innovation is picking up pace but iscomparatively limited. 5.Edge computing demand is fuellingthe next stage of infrastructuredevelopment. India’sdata centre industryhasbeengrowing at a CAGR of 24 percent since 2019,with its edge data centre market projectedto grow at a CAGR of 19.5 percent between2025-2033,indicating strong demand forprocessingdata close to source.India’sedgedatacentremarketiscurrentlydominatedby IT&Telecommunications aswellas BFSI(Banking,Financial Services,andInsurance).With comparatively high AIadoption in the manufacturing and telecom,andthe media and entertainment sectors,it offers opportunities to diversify to Tier 2and 3 cities close to sectoral demand hubstotake advantage of low costs and landavailability. 3.AI is triggering a change in theinvestment landscape. AI,with its promise of cost and processoptimisation, is forcing investors to rethinktheirinvestment approach.Investors haveexplored integration of AI at scale—e.g., withIndia’s outsourcing sector through strategicbuyouts.Indian IT firms are also fundingstartupsin a bid to pursue intellectualproperty-ledgrowth.Further,with a viewof data as a strategic resource, technologyfirms are consolidating their data capabilitiesthrough mergers and acquisitions. canbe used to inform strategic action.Akey part of public sector transformation canbepreferential procurement for domesticinnovation,thereby providing a boostingeffect across the AI economy as a whole. 6.Talent development is required acrossthe AI ecosystem as a whole. Indiacurrently has a 51-percent demand-supplygap when it comes to the nicheskillsrequired for core AI development.Inaddition to building its talent pool in coreAIdevelopment,India needs to developskills and human capital to address currentecosystembottlenecks with talent in dataengineering, cloud, and compute. 9.Innovation and governance arecomplementary forces. Innovation and governance are complementary,andnot opposing forces,with