Indian AI Startups Get a Reality Check: Hype, Heuristics, and Cash

When Subtl.ai shut down in 2025, it shattered a thousand dreams. To many, it was the perfect case of a GenAI startup building from India, for India. 

It helped the State Bank of India extract information from a growing pile of files using its RAG (Retrieval-Augmented Generation) system, securely collecting pertinent data. It had set out to build a 'Private Google' for organisations, had procured defence contracts, and worked with at least two airports to implement its document intelligence solution.

However, the startup’s cumulative revenue since 2020 was just Rs 55 lakh—a gap between promise and monetisation that proved fatal.

Co-founder and CEO Vishnu Ramesh blamed misplaced assumptions around investor commitments, the broader funding slowdown, and being stretched thin. “We wanted to do OpenAI-like things for the RAG ecosystem, but I got stuck handling customers from wildly different domains with wildly different use cases,” he had posted on LinkedIn.

Subtl.ai is just one among a herd of Indian AI startups that recently shut shop.

According to Tracxn, 18 India-headquartered AI companies closed in 2025, compared with 6 in both 2024 and 2023.

The seeds of India's startup collapse were sown during the pandemic boom. Between 2020 and 2022, money was cheap, venture capitalists were aggressive, and nearly every pitch deck found a cheque. Startups raised at 100x revenue multiples, burning through crores on ads, discounts, and rapid expansion. Unit economics were an afterthought. Valuation was everything. 

Then the music stopped.

"During the pandemic, many companies capitalised on temporary demand and easy money, but as normalcy returned, cracks in their foundations became evident," Gaurav Gulati, Founder and Managing Partner, Aroa Venture Partners, tells AIM.

Global VC investments dropped by over 40% year-on-year after 2022, according to the Global Innovation Index. In India, new startup launches fell from roughly 9,600 annually in the pre-COVID-19 years to just 5,264 in 2024, according to the Department for Promotion of Industry and Internal Trade. 

For companies already running on fumes, the follow-on rounds simply never came.

The Funding Picture

India's equity funding landscape for startups has been under serious strain. 

According to Tracxn, Indian startup funding fell to $10.5 billion in 2025, down 17% from $12.7 billion in 2024 and $11 billion in 2023. The share of AI startups was a mere $643 million, a fraction of the $121 billion that US startups raised.

Moreover, year-to-date figures for 2026 show that Indian startups have raised only $3.48 billion across 365 equity funding rounds—a 33% drop.

The story is more nuanced than a simple funding drought. VCs are investing—but far more selectively. The 'wrapper' era is over. 

"AI wrappers and application-layer tools built quickly on commoditised models, without deep defensive moats, are facing the sharpest correction,” Dori Yona, CEO and Co-founder, SimpleClosure, noted in the 2025 State of Startup Shutdowns report. The company assists venture-backed startups in winding down efficiently.

VCs are no longer funding startups that simply put a user interface on top of existing Western AI models. Top-tier funds are aggressively hunting for outcome-based B2B AI agents that automate enterprise workflows, replacing traditional per-seat SaaS pricing.

"Building primarily on the application layer is not a weakness; it is pragmatic. The real danger lies in mistaking wrappers for businesses, and confusing early traction with defensibility,” Ninad Karpe, Founder and Partner, 100X.VC explains.

Exit pressure is also creating a structural mismatch with AI's long innovation cycles. Pre-IPO rounds and secondaries have become the flavour for larger funds, but it has also created tunnel vision around exits. Startups are now raising significant rounds without an eye on valuation, promising an IPO in the next two years.

"Engagement and retention are the two core metrics VCs are looking at with AI startups—not growth and acquisition. What will happen when another business comes up with the same solution? What is the reason customers will stick with you?” Arpit Mittal, CEO and Co-founder of edtech startup SpeakX, ponders.

Subtl.ai also met its end when it ran out of cash runway, unable to raise any follow-on rounds after raising ₹1.8 crore.

The AI Graveyard: When Hype Met Reality

Many AI startups fail to move beyond the initial proof-of-concept stage.

Take Tune AI, which focused on helping enterprises and developers leverage generative AI for their workflows and claimed hundreds of thousands of users and integrations with popular AI models. But as major cloud providers began bundling similar tooling at lower cost, Tune AI's margins and differentiation collapsed. Many of its users were free developers, not paying enterprise clients. The startup quietly shut down last year.

This is what industry insiders now call 'pilot purgatory'—the phase in which AI projects get caught in a continuous loop of testing and small-scale trials, with no clear strategy or leadership buy-in to move them into production. According to a 2025 MIT study, roughly 95% of enterprise generative AI pilots fail to deliver measurable financial returns. 

"In the 2026–27 window, 'real PMF' won't be a demo or pilot count. It will look like AI moving into production with measurable business outcomes. Investors will increasingly distinguish between experimentation and systems that run real work,” notes Deepak Dhanak, Co-founder and COO of vibe-coding platform Rocket.

AI startups are also victims of their own hype.

Take AI health startup Nintee, which launched in 2022 to much fanfare as a project of Wingify founder Paras Chopra. Though backed by Peak XV Partners and CRED founder Kunal Shah, the company folded in 2024 and had to return the money to its investors. Chopra attributed the closure to user retention and scaling, noting that even a pivot to an education-led model failed to revive the platform's user numbers. 

“Our original hypothesis was to use AI to help people build better habits to transform their lives. This attracted a passionate niche, but we couldn’t build conviction that it could be a VC-scale business,” Chopra explained in a blog post. 

And then there was Builder.ai, the Microsoft-backed AI startup once valued at $1.5 billion that promised to make app-building easy. It filed for bankruptcy in 2025 amid revelations about inflated revenue figures and AI whitewashing, forever becoming a cautionary tale for the industry.

The Talent Deficit

A lot of this can be traced back to the AI talent problem.

India mass-produces engineers but lacks specialists who can build and deploy production-grade AI systems in domain-specific verticals. According to TeamLease Digital, the demand-supply gap for AI talent stands at 51%, and there’s only one qualified engineer for every 10 open GenAI positions. 

There’s high demand for specialised roles such as prompt engineers, LLMOps specialists, MLOps engineers, AI product engineers, and data engineers. GenAI engineers now command salaries exceeding Rs 60 lakh annually, an 18% year-on-year increase, according to the HRtech platform’s report. 

"AI is reshaping how we hire and develop talent, but the real unlock isn't just adopting AI; it's making it work for the business. Too often, companies pour resources into AI tools without the right talent to unlock their full potential, turning a game-changing opportunity into a missed one,” said Ruchee Anand, Vice President, Talent Solutions at LinkedIn, during a report launch last year.

Pivot or Perish

Indian AI startups, plagued by a lack of resources and intense competition, remind one of the early internet companies. While there was a massive consolidation, hundreds of thousands of companies either went bankrupt, liquidated, or were acquired cheaply during the dot-com bubble crash. 

"This year, we are witnessing a natural weeding out of unsustainable models, which is crucial for building stronger, more resilient startups. Shutdowns are still happening, often due to business models lacking long-term viability or the ability to adapt to evolving market dynamics,” Karpe enlightens.

The defining survival strategy for startups that have made it this far is the pivot. Roughly 63% of Indian GenAI startups pivoted their model or focus in the past year, largely toward vertical SaaS and application-focused models, according to NASSCOM. Startups with fewer legacy constraints are rebuilding workflows around agent-driven models and seeing measurable efficiency gains.

"We've seen adoption jump almost overnight once we localised our models,” says Aman Goel, Founder, GreyLabs AI, which specialises in agentic voice AI and speech analytics for Hindi in the BFSI sector.

The resilience is real, but analysts and investors are increasingly framing the shutdowns as an inevitable consequence of a boom built on cheap capital rather than durable fundamentals.