The market saw a 60% revenue jump in India after a free trial ended. Most would call this growth. I call it a stress test of retention.
Perplexity AI, the answer-engine startup, reported that its Indian revenue surged 60% following the conclusion of a promotional bundling deal with telecom giant Airtel. The free-period ended, and instead of the usual churn cliff, the revenue line climbed. This is the kind of data point that sends venture capitalists into a frenzy. But as someone who has spent years dissecting the structural integrity of digital assets—from ICO tokenomics in 2017 to DeFi liquidity mining in 2020—I know that headline growth often masks the deeper fault lines.
Context: The AI Search Architecture
Perplexity is not a foundational model builder. It is an application-layer innovator that stitches together third-party LLMs (GPT-4, Claude, and its own Sonar series) with a real-time retrieval-augmented generation (RAG) pipeline. Each query requires crawling, re-ranking, multi-step reasoning, and citation mapping. The unit cost per search is significantly higher than a standard chatbot interaction. This cost structure makes Perplexity acutely sensitive to both infrastructure spending and per-user economics.
The Airtel partnership was a classic telecom bundling play: free Pro subscription for a set period, then a conversion to paid at a localized Indian price (roughly INR 200-300 per month, about one-third of the US price). The expectation was that price-sensitive Indian users would churn after the free ride. Instead, revenue grew 60% post-promotion.
Core: The Silence Between the Candlesticks
Let me be clear: this 60% is a retention signal, not a growth signal. The number of new users (downloads) remained low, but the revenue per existing user expanded. The math is simple: a small base of users converted at a high rate, and many of them upgraded to higher tiers (like Pro Max). This is not a story of mass adoption; it is a story of deep, engaged cohorts.
But watch the silence between the candlesticks. The real metric is not revenue but the gross margin of that revenue. If Perplexity’s cost per query in India—given the same inference cost but lower pricing—is negative, then every additional dollar of revenue is a dollar of loss. The 60% growth becomes a liability, not an asset. Based on my experience auditing 40+ ICO whitepapers in 2017, I learned that top-line growth without unit profitability is a ticking time bomb. The LUNA collapse in 2022 taught me that structural fragility can be masked by a rising curve.
Furthermore, the low download numbers hint at a dependency on a single channel: Airtel. This is a classic single-point-of-failure in distribution. Cross-chain bridges have been hacked for over $2.5 billion cumulatively, yet the industry still depends on them. Perplexity’s dependence on Airtel is a similar structural paradox. If Airtel renegotiates revenue share terms or partners with a competitor (like Jio with ChatGPT), the growth evaporates.
Contrarian: The Decoupling Myth
The prevailing narrative is that this validates the “AI search” vertical in emerging markets. I disagree. The contrarian angle is that this is a one-time signal from a carrier-subsidized, price-sensitive cohort. The 60% growth likely includes a spike from users who purchased Pro at the end of the free period to lock in the discounted price—a one-off revenue swell. The next quarter will almost certainly see a regression.
Moreover, the real competitive threat is not ChatGPT or Google Gemini, but Google’s AI Overviews embedded directly into the search mainline. Google owns the distribution monopoly in India. If Google integrates similar RAG capabilities into its core search product, the need for a standalone Perplexity app collapses. The 60% growth is a lighthouse, but it is shining on a rock.
Takeaway: The Liquidity Harvest
Harvesting the liquidity that others overlook requires patience. The pattern emerges from the chaos of noise. For Perplexity, the next six months will reveal whether the Indian experiment is a sustainable model or a blip. The key is not the 60% growth, but the gross margin and the repeatability of the telecom bundling in other emerging markets. If the unit economics are positive and the channel can be replicated, then this is a genuine paradigm shift. If not, it is a footnote.
As a digital asset fund manager, I see a parallel to early DeFi yield farming: the first mover gets the liquidity, but the second mover gets the rug. Perplexity has the first-mover advantage in carrier-bundled AI search. But the structural fragility of single-channel dependence and negative unit economics is a fault line that cannot be ignored. The silence between the candlesticks is growing louder.
