Over the past quarter, Perplexity AI reported a 60% revenue increase in India after ending a free trial with Airtel. At first glance, this is a growth story. But for those of us who watch order flow, it's a positioning signal. The market interprets this as a validation of AI search. I see a more subtle truth: a test case for distribution models that could redefine how crypto protocols acquire users in price-sensitive markets.
Context: The Airtel Experiment
Perplexity is not a blockchain company. It is an AI search engine. Its product integrates real-time web retrieval, citation tracing, and multiple LLM backends. In 2025, the company partnered with Airtel, India's second-largest telecom operator, to offer a free trial of Perplexity Pro to Airtel subscribers. The trial ended. Revenue in India climbed 60%. App downloads remained low. This combination—low downloads, high revenue growth—is the anomaly that demands attention.
In crypto, we see similar patterns. When a DeFi protocol launches a liquidity mining program, TVL spikes. When incentives end, TVL collapses. The 2022 bear market taught me this lesson personally. I held positions in Curve and Lido, watching TVL drop 40% as rewards dried up. Holding the line when the world screams to sell was my only discipline. Perplexity's retention after trial expiry suggests a different dynamic: the product delivered enough value to convert users even without further subsidies. That is rare.
Core: The Retention Signal
India is a price-sensitive market. AI subscriptions are priced at roughly one-third of US rates. ChatGPT and Gemini are free. Yet users paid. Why? The answer lies in the product's unique value: real-time, verifiable answers with citations. For a market where information asymmetry is high, this is a structural advantage. The 60% revenue growth implies that a meaningful portion of Airtel's trial users converted to paid subscribers. This is not a spike from a single promotion. It is a signal of latent demand.
I have seen this before. In 2024, during the Bitcoin ETF approval, I executed 15 trades based on institutional volume spikes. The data told me where the smart money was flowing. Here, the data tells me that AI search has found a product-market fit in a non-Western market. The distribution channel—telecom bundling—is the key. Airtel already has billing relationships with hundreds of millions of users. For crypto projects, this model is a blueprint. Instead of relying on App Store organic downloads, a protocol can partner with a local telecom to distribute a wallet or a staking app. The retention data from Perplexity suggests that if the product provides verifiable utility, users will stay.
The Contrarian Angle: Small Base, Fragile Economics
But the 60% growth is from a low base. Perplexity's India downloads are tiny compared to ChatGPT. The absolute revenue is likely still modest. High conversion on a small funnel does not equal market dominance. Moreover, the unit economics are unclear. AI search has high inference costs. Each query requires retrieval, reranking, multi-step reasoning, and citation generation. At Indian pricing, margins may be thin or negative. Growth that sells at a loss is not sustainable. I have seen this in crypto: projects that buy users with high token incentives, only to watch them leave when rewards stop. Perplexity's retention is better, but without cost data, we cannot call it a victory.
Another risk: the telecom partnership may be a one-off. Airtel subsidized the trial. If Perplexity faces higher revenue share demands or if competitors like Google or OpenAI strike similar deals, the advantage vanishes. The structure of the deal matters. I always look for structural integrity in a protocol's design. If the foundation is weak, the tower falls. In this case, the foundation is a single partnership in a single market. That is not a moat.
Takeaway: What This Means for Crypto
The crypto market is currently sideways. Chop is for positioning. Perplexity's India story offers a framework for evaluating projects that combine AI with distribution. I am watching for protocols that replicate this model—partnerships with local telecoms or banks to distribute AI-powered crypto tools. The key metric is not initial download volume but retention after the free trial ends. Noise is expensive. Silence is profit. The quiet signal here is that users in emerging markets will pay for verifiable information. That is a truth that applies to blockchain-based data markets as well.
Discipline is the only edge that compounds. Structure reveals truth, volume confirms it. I will hold my positions until the data forces a change. The chart doesn't speak either. But the order flow does. Perplexity's 60% is a whisper. I am listening.