AI Investors Shift Focus From Models to Moats as Startup Failures Rise; 18 Indian AI Startups Shut in 16 Months โ VCs Now Backing Applied AI With Real Differentiation Over Thin-Layer Foundation Model Wrappers
The Shutdown Wave โ What's Happening
18 Indian AI startups shut down in the last 16 months โ signalling sector stress
Companies that have ceased operations include Builder AI, Neuropixel, Dotagent, Zeda. io, LEGOAI and MoneyAI
Even Ola's AI venture Krutrim has reportedly pulled back its consumer assistant from app stores and the web as it pivots towards AI cloud infrastructure
Shutdowns reflecting a broader reassessment of the AI startup landscape after aggressive funding in 2023 and 2024 when many companies raised capital on expectations of rapid growth that proved difficult to sustain
Why Startups Are Failing โ The Core Problem
Large global technology and LLM players have very high interest in building horizontal AI capabilities into their own stack โ rapidly eroding startup moats
Moat can "shrink quickly when one of them decides to move into your space" โ Satyakam Mohanty, Wyser Capital
Concern particularly acute for startups building products as a thin layer on top of existing foundation models without underlying intellectual property or deeper customer integration
Advances in underlying technology stack can quickly reduce the relevance of such offerings
Pilot-to-revenue conversion remains a key bottleneck โ many AI startups secured pilots but struggled to convert them into recurring commercial contracts
What VCs Want Now โ The Shift in Focus
Investors increasingly moving away from generic AI offerings and focusing on businesses with stronger differentiation and clearer commercial outcomes
VC firms showing greater preference for applied AI models where technology solves specific business problems and becomes embedded into customer workflows
Preference for AI that directly improves productivity, reduces costs or creates measurable efficiency gains across manufacturing, healthcare, fintech infrastructure and enterprise workflows โ Apoorva Ranjan Sharma, Venture Catalysts
Investors wary of "thin-layer" AI built on foundation models
Post-2023 funding boom cooling amid unmet growth expectations
Founder Behaviour โ What's Changing
Instead of building around models first, entrepreneurs now spending more time validating use cases, distribution and operational integration before taking products to market
Founders most drawn to now are those who have spent real time on the operational or distribution layer before spending time on the model itself โ Maanav Sagar, Good Capital
AI startups benchmarked against best global companies for speed of iteration and scaling โ a tall order
Lots of pilots not converting into real commercial traction โ Deepak Gupta, WEH Ventures
Core Theme
India's AI startup ecosystem is undergoing a painful but necessary correction โ after a funding frenzy built on model-first thinking and optimistic growth projections, the market is now demanding real moats, embedded workflows and proven revenue conversion; with 18 shutdowns in 16 months and big tech rapidly commoditising the model layer, the next wave of AI investment will flow to founders who have done the hard work of distribution, operational integration and customer stickiness โ not just those who have built on top of someone else's foundation.