AI could unlock $1.7T in value for India if the country gets the next decade right. In NITI Aayog’s AI for Viksit Bharat framework, India’s baseline GDP reaches $6.6T by 2035. With faster AI-led growth, the figure rises to $8.3T. The gap is large enough to be inspirational, but also to demand uncomfortable questions about execution.
WaterBridge Ventures convened those questions in a closed-door roundtable with Dr. Chintan Vaishnav, an MIT-trained socio-technologist and former Mission Director of the Atal Innovation Mission at NITI Aayog. The session brought together founders, operators, and investors to examine where the national AI roadmap holds, where it strains under operating reality, and what it leaves out.
The policy framework is ambitious and serious. It identifies three levers behind India’s AI opportunity: productivity gains across industries, AI-led research acceleration, and growth in technology services. Its first two levers focus heavily on sectoral transformation, with banking and manufacturing used as detailed test cases. The report estimates that accelerated AI adoption across industries could add $500-600B to GDP by 2035, with financial services alone representing a $50-55B productivity opportunity.
AI adoption is already real at the firm level. One builder described a drastic reduction in workforce size while maintaining the same revenue base, a lived version of the productivity gains the report models. Financial services practitioners confirmed that customer service, fraud detection, and credit underwriting are seeing genuine AI deployment. Data fragmentation came up repeatedly as the central bottleneck across manufacturing, fintech, and wealth management, matching the report’s own concerns around fragmented, inaccessible, and inconsistent data.
Manufacturing exposed the infrastructure gap most clearly. The question is not only which model to use. Most factories are still not instrumented deeply enough for AI to work well. Without IoT sensors, clean process data, and connected machines, AI in manufacturing remains a promise sitting ahead of the data layer required to deliver it. The same pattern appeared in the MSME discussion. Smaller businesses may benefit enormously from AI, but many lack the data systems, awareness, and shared infrastructure needed to begin.
The most important challenge concerned workforce redeployment. The NITI framework assumes that displaced workers are redeployed within the same sector at similar productivity levels. Participants questioned whether this is how AI will actually move through knowledge work. Entry-level roles in startups, venture capital, services, and operations may be compressed before a clean redeployment pathway exists. While that may not weaken the productivity thesis, it definitely changes the social arithmetic beneath it.
Banking’s role as a lead sector also drew scrutiny. The report is right that financial services has significant AI potential, but practitioners argued that it is also one of India’s most regulated and constrained domains. Personalised financial advice cannot simply be automated without licensing and liability clarity. Legacy core systems make middle-office integration slower than use-case maps suggest. Data aggregation across multiple bank accounts, apps, and financial products remains unresolved for many consumers. Banking may be a large opportunity, but size should not be confused with ease of adoption.
The sovereignty question added another layer. If Indian firms adopt AI through foreign model providers, India may capture operational efficiency while the model-layer economics accrue elsewhere. Stanford HAI’s AI Index shows how concentrated frontier model production remains, with U.S.-based institutions far ahead of other regions in notable model output. India’s compute push through the IndiaAI Mission is meaningful, but compute access alone does not close the full gap. The country also needs datasets, model capability, application depth, procurement demand, and pricing power.
Market access surfaced as another missing bridge. Productivity gains in manufacturing or auto components only translate into GDP value if firms can sell into higher-value markets. A component maker that improves design cycles through AI still needs certification, buyer access, export channels, and trust from global OEMs. The NITI report does identify certification and market-access risks in specific sectors, but the roundtable made the point more general: AI infusion is necessary, while commercial conversion requires a broader industrial strategy.
Consumer AI may be India’s most immediate mass-adoption opportunity. Several participants pointed to vernacular-language apps, tools for micro-entrepreneurs, and UPI-style distribution as the place where Indian AI could scale first. A sectoral framework built around banking and manufacturing can miss this because it begins from GDP blocks rather than user behaviour. India’s strongest AI products may emerge from messy, multilingual, low-ticket, high-frequency usage before they become formal enterprise infrastructure.
The Indian founder was another missing variable. Participants argued that India’s AI advantage may come less from top-down policy design and more from ground-level founders who are resourceful, cost-effective, and close to real constraints. This is not romanticism. It is an operating thesis. In markets where customers pay less, infrastructure is uneven, and workflows are informal, founders are forced to build lighter, cheaper, and more adaptive products. That frugality can become a global advantage if paired with distribution and pricing power.
Even the innovation-index discussion needed precision. India is no longer at the lower rank sometimes cited in older conversations. It ranked 39th in the Global Innovation Index in 2024 and 38th in 2025, a major rise from 81st in 2015. The deeper point remains valid: India’s economic weight, talent base, and startup activity still need to translate into stronger research, IP, product depth, and frontier technical capability. The question is no longer whether India has improved. It is whether the improvement is fast and deep enough for the AI decade.
Can data systems be cleaned and connected fast enough? Can factories be instrumented before AI roadmaps overpromise? Can MSMEs access shared AI infrastructure without needing enterprise-grade teams? Can financial-sector AI move responsibly within regulation? Can displaced junior workers find new ladders into productivity rather than falling out of the pipeline? Can Indian AI products earn global revenue without being forced into cheaper-faster domestic pricing? Can India build enough model-layer capability to retain more of the value it creates?
These are not mere objections to the AI for Viksit Bharat vision, rather they are the conditions under which the vision sees the light of day.
WaterBridge’s roundtable mattered because it created a bridge between policy architecture and operating truth. The top-down framework supplies direction, scale, and urgency. The bottom-up voices supply friction, sequencing, and consequence. India needs both. A country cannot build an AI economy from founder anecdotes alone. It also cannot build one from GDP models that assume the hard parts will resolve themselves.
The clearest output from the day was an honest inventory. India’s AI ambition is not in question. The next task is execution infrastructure: data readiness, compute access, liability frameworks, enterprise literacy, MSME enablement, market access, sovereign capability, and a more realistic view of labour transition.
The $1.7T question is therefore not only whether AI can lift India’s GDP. It is whether India can build the institutional, commercial, and technical rails required to capture that value at home.