We mined liquidity while the code slept. That's what I thought when I read the Crypto Briefing report on Perplexity AI's Indian operations. A 60% revenue jump after the Airtel free promotion ended—sounds like a breakout, right? But I've been a battle trader long enough to know that top-line numbers without unit economics are just noise. In 2022, I watched Terra's algorithmic stablecoin collapse because everyone ignored the mechanics behind the growth. This is no different.
Let me step back. Perplexity AI is an AI search engine—a 'answer engine' that combines large language models with real-time retrieval and citation. In 2025, they partnered with Airtel, India's second-largest telecom, to offer free Pro subscriptions to millions of users. The typical outcome? Freebie hunters flood in, churn when the party ends. But here, revenue didn't drop—it climbed 60% post-promotion. That's a strong retention signal. Yet the article also notes low app downloads. The implication: a small, high-quality user base converted through a targeted channel, not mass adoption.
Core Analysis: The Conversion Chemistry
From my experience running a copy-trading community, I've learned that conversion rates are the real metric. In DeFi, a 10% deposit-to-stake conversion is considered healthy. Perplexity's 60% revenue growth on a small base suggests conversion rates far above industry norms. The Airtel channel is a precision distribution tool—like a liquidity pool that only admits strategic capital. Telecom operators in India have billing relationships with hundreds of millions of users. They've already sold Netflix and Spotify bundles. Now AI subscriptions. The key is that Perplexity's product—search with verified citations—solves a real pain point for Indian users who rely on mobile data for information. It's not a chatbot toy; it's a utility.
But here's the technical catch. Every AI search query costs significantly more than a standard chatbot query due to retrieval, reranking, multi-step reasoning, and citation generation. Perplexity uses a mix of third-party models (GPT-4, Claude) and its own Sonar series. In India, they price subscriptions at about one-third of US rates (₹200-300 per month). That's a razor-thin margin. If the cost per query exceeds the revenue per user, then 60% growth is just 'selling more at a loss'—a classic trap I saw in 2020 when Uniswap V2 liquidity miners competed for yields that were net negative after impermanent loss.
The Contrarian Angle: Fragile Growth
Everyone is celebrating the revenue surge. I'm not. Low downloads mean weak organic growth. Perplexity's user acquisition is almost entirely dependent on the Airtel deal. What happens when Airtel renegotiates the revenue share? Or when Reliance Jio strikes a similar deal with Google or OpenAI? The competitive moat is not the product—it's the distribution partnership. And that's a lease, not ownership.
Worse, the unit economics remain unverified. The article doesn't disclose absolute revenue numbers, conversion rates, or profit margins. As a pre-mortem risk engineer, I always ask: 'What if the cost of serving an Indian user is higher than the subscription price?' In 2021, I audited a DeFi protocol that had 200% APY but was paying out from the treasury—it collapsed within three months. Perplexity's growth could be a similar illusion if the cost structure is unsustainable.
Then there's Google. AI Overviews are already being embedded into the dominant search engine in India. Once Google fully integrates AI search into its free product, Perplexity's main differentiator—real-time, cited answers—becomes a commodity. The 60% growth might be a temporary spike before the market consolidates.
Ethical and Regulatory Landmines
India's media ecosystem is litigious. Perplexity has already faced copyright complaints from US publishers. In India, local language news outlets are increasingly protective of their content. If Perplexity's crawlers are scraping Indian news sites without authorization, legal battles are imminent. And the Indian IT rules impose strict liability on intermediaries for user-generated content. AI-generated answers that provide incorrect medical or financial advice could lead to lawsuits. I've seen this in crypto: projects that ignore regulatory signals end up delisted or fined. Perplexity's risk is similar.
Takeaway: The Telecom-AI Model as a New Paradigm
This episode validates a viable go-to-market strategy for AI services in emerging markets: partner with a telecom giant, offer a free trial, and convert high-intent users. But the model is fragile. It depends on channel exclusivity, sustainable unit economics, and the ability to fend off incumbents. For the crypto world, there's a parallel. We saw how mobile money (M-Pesa) enabled crypto adoption in Africa. Now AI subscriptions may follow the same path. But as I wrote in my paper on 'Regulatory-Proof Yield,' the last human decision—the one that checks the bottom line—cannot be automated. Perplexity's 60% growth is a signal, but without the full dataset, it's just a number in a sea of noise. We rode the wave until it broke our boards. Let's see if Perplexity can surf the next one without sinking.
Liquidity is just trust, digitized and leveraged. In this case, trust is on the line.
