Big bets on AI, but the exit doors look small

Artificial Intelligence (AI) is witnessing one of its biggest funding surges yet. AI startups raised $47.3 billion in Q2 2025, the second-highest quarterly total on record, buoyed by mega-deals such as Meta’s $14.8 billion minority stake in Scale, according to recently released CB Insights. Deal volume edged up 2.5% from the previous quarter, with median deal size rising to $4.6 million, a four-year high. Overall, 2,772 AI deals have been sealed this year, pulling in $116 billion in global investments.

But behind the enthusiasm lies an increasingly pressing question: Is this momentum grounded, or are we in the early stages of an AI bubble? The rapid surge in capital contrasts sharply with the still-uncertain ability of many AI companies to scale revenues and prove sustainable commercial models.

Speaking to Business Today, Anil Joshi, Founder & Managing Partner at Unicorn India Ventures and member of IVCA’s VC Council, stresses that exit visibility remains uncertain. “Exit is a very relevant concern. AI today is in a zero-to-one phase where success depends heavily on execution and timing. If the broader AI wave moderates, fallback or exit paths may be limited as many models are still being validated for commercial scalability rather than profitability.”

He adds that recovering capital from businesses where models are still being validated can be difficult, particularly in cases where valuations have already run ahead of fundamentals. The sector continues to command premium pricing because of its transformative promise and the belief that AI will eventually permeate every sector of the economy.

On the other hand, Arjun Malhotra, General Partner at Good Capital, argues that the current boom isn’t comparable to hype cycles of the past. “The capital committed to AI by any major tech company represents multi-year deployment timelines. These are infrastructure investments with three-five year build-out periods,” he says. According to him, “there will absolutely be winners, losers, and companies that go nowhere. But this isn't unique to AI — it's true of every technology wave. The fintech boom, the edtech surge, SaaS adoption — same pattern.” He insists investment momentum remains intact: “Honestly, we don't see a slowdown coming — at least not in the foreseeable investment horizon that matters for our fund cycle.”

While optimism runs high, Joshi explains that assessing fundamentals in AI startups is a uniquely challenging task. “In normal circumstances, a balanced view between promise and proof works. But in the case of AI, the space is changing very fast and hence becomes tricky,” he says. Investors are closely scrutinising revenue streams, customer stickiness, proprietary technology, competitiveness, and unit economics, along with a team’s ability to rapidly evolve. He points out that “most startups are still in experimentation or the initial phase, so unit economics and scale hinge on how efficiently the underlying AI models are being productized and monetised.” Yet, he adds, “valuation at times seems to run ahead of fundamentals, driven more by global sentiment and scarcity premium rather than operational metrics.”

Malhotra notes that more than half of Good Capital’s Fund II portfolio consists of companies that meaningfully integrate AI into their core, not just as “window dressing,” but enabling their primary value creation. Meanwhile, Joshi underscores a key learning emerging from the sector’s growth so far: “AI’s value proposition doesn’t come from the technology itself, but from the meaningful outcome it delivers. If the value proposition isn’t clear, then even the most advanced AI solution quickly loses relevance.”

Globally, nearly 72% of AI deals in 2025 so far are flowing to early-stage startups, especially in segments like AI agents and voice-driven interaction technologies — signalling a shift toward more autonomous and intuitive applications. But such innovation is largely pre-scale, making it vulnerable to macro corrections. Joshi warns that Indian AI startups, still at an early stage relative to global counterparts, are inevitably exposed. “Given that much of the capital flow, valuation benchmarks, and sentiment are globally driven, any downturn will certainly impact the Indian ecosystem as well,” he says.

The money isn’t slowing, but neither are the questions. With valuations racing ahead and revenues still finding their footing, the AI boom may be entering its most critical test: transforming investor conviction into commercial reality.