人工智能、全球与印度：当前趋势与未来方向

A.I., the World, and India:Present Trends and Future Directions Anirban Sarma and Basu Chandola Introduction I to the tens of billions of devices it has poweredsince 2020, and its expanding user base. BetweenOctober2022 and April 2025,ChatGPT usersalone went from zero to 800 million, and theplatform crossed US$4 billion in revenue,2making as artificial intelligence (AI) and outlines how AI isreshaping economic structures, business models,and user behaviour. AI has driven steep upwardtrends across various indicators—ranging from itsprojected contribution of US$15.7 trillion to theglobal Gross Domestic Product (GDP) by 2030,1 Attribution:Anirban Sarma and Basu Chandola, “A.I., the World, and India: Present Trends and Future Directions,”ORFSpecial Report No. 270, Observer Research Foundation, July 2025. This white paper identifies eight key takeawaysfrom the BOND report: 1.Strengthening AI sovereignty can enhancegeopolitical influence. 7.As AI disrupts industries,present andfuture talent must be equipped for a new 2.As the power and adoption of AI toolsincrease,underlying infrastructure and 8.Satellite internet may drive the next waveof AI participation. 3.The escalating demand for compute createsnew opportunities for innovation. Theseeight phenomena impact India justas India, in turn, influences them. The BONDreport points out, for example, that India hasthe largest share of monthly active users (MAUs)of the ChatGPT mobile app and the third largest 4.The production, supply, and availability ofhigh-performance AI chips will be centralto states’ AI leadership. 5.Data centres are proliferating, but capacityremains uneven between developing andadvanced economies. This paper describes how each of these eighttrends is unfolding globally, and explores theirimplications for the Indian government, regulators,businesses, and markets. It concludes that Indiacouldbecome an AI shaper—defining norms, 6.AI economics show rising revenues but alsosoaring costs, which limit profitability. Takeaways from theBOND Report 1.Strengthening AI sovereignty isa pathway to greater geopoliticalinfluence. lead globally, highlighting a growing platformand regulatory divide.11 Meanwhile, the USdeveloper ecosystem is expanding rapidly—NVIDIA’s base grew sixfold to 6 million overseven years, more than doubling in just the lastfour.12Similarly, Google’s AI ecosystem grew •AI is emerging as a tool for building nationalresilienceand geopolitical influence,6 withthe global AI race increasingly being definedby US–China rivalry. The US leads in modelinnovation and custom silicon, while Chinaadvances through open-source development,state coordination, and scaled infrastructure.7In 2025, China released three major open- •States have become more attuned to theimperativeof producing AI using theircountry’sown infrastructure,data,models,andbusiness networks.14‘Sovereign AI’encompassesbothphysicalanddatainfrastructure—rangingfromlocalcloud •India must strengthen itsAI Missionto buildsovereigncapabilities and reduce externaldependence.This includes developing astrong AI ecosystem that fosters innovationandenables effective regulation.For both 2.As the power and adoption ofAI tools increase, underlyinginfrastructures and talent must befortified. •AI adoption is accelerating across consumers,enterprises,andgovernments,outpacingpast waves of technological change. ChatGPTreached800 million users in 30 months,generatingnearly US$4 billion in revenueand 20 million subscribers.23The number of •Creating foundational LLMs is another coreelement of AI sovereignty. Sarvam AI16 hasbeen selected to build India’s first LLM, butalternative models must also be encouraged.Privateplayers should focus on developingmore localised AI solutions and applications.For instance, AI4Bharat is developing open-source language models for Indian languages,17while SoketAI,18Gnani.ai,19and Gan.ai20arebuildingfoundational models tailored for •AI tools are evolving from assistants to serviceproviders.26 New AI agents can reason, act,andcomplete multi-step tasks,managingworkflowsand executing commands vianatural language. A race has now begun tobuildagentic interfaces,enterprise copilots,autonomous systems, and sovereign models. •India has continued to develop a range of AIuse cases across health, education, agriculture,financialservices,and smart cities,in linewith its 2018 National Strategy for ArtificialIntelligence.28 However,its focus is nowexpandingto include foundational model pointingto 2.3 million new openings by202731 and around 2.73 million by 2028.32Expandingcompute capacity,scaling datacentres,and investing in next-generation 3.The escalating demand forcompute creates new opportunitiesfor innovation. •India now faces opportunities to buildsovereignAI models,grow its AI-readyworkforce, and strengthen its talent pipeline.Itis advancing its capabilities through thedevelopmentofindigenousfoundationalmodels—including LLMs and small languagemodels—sucha