Hook
Over the past quarter, Perplexity AI reported a 60% revenue increase in India following the expiration of its Airtel promotional period. Most analysts read this as a simple growth story — a validation of AI search in a price-sensitive market. I see a structural signal: a narrative arbitrage in the subscription model that mirrors the early DeFi liquidity mining cycles. The market fixates on raw user numbers; I fixate on the conversion funnel. The data reveals a hidden layer: the low download count coupled with rising revenue suggests that the true value accrual is not from broad adoption but from high-quality channel-driven retention. This is not a growth story — it is a retention story, and the implications for AI subscription economics are far more significant than the headline percentage suggests.
Context
Perplexity AI is an AI-powered answer engine that aggregates real-time web data via retrieval-augmented generation (RAG) and cites sources. Unlike general chatbots, it prioritizes verifiability and structured answers. Its business model is subscription-based, with a free tier plus Pro plans at $20/month in the US, and a localized Indian pricing at roughly INR 200–300/month (1/3 to 1/2 of US price). In early 2025, Perplexity partnered with Airtel, India’s second-largest telecom operator, to offer free Pro subscriptions to select Airtel users. The promotion ran for a limited period. After the promo ended, Perplexity’s India revenue jumped 60%.
To understand the significance, one must recognise the context: India is a market where free AI tools (Google Gemini, ChatGPT Free, DeepSeek) dominate. Telecom bundles have historically been used for streaming services, not for AI. The 60% figure is a single data point, yet it carries the weight of a thesis — if the mechanics hold, it could redefine how AI services penetrate emerging markets. But as with any early-stage signal, the devil is in the unit economics, not the top-line percentage.
Core: The Retention Signal and the Hidden Mechanics of Subscription Arbitrage
Let me be direct: the 60% revenue growth after a promotional period is a classic sign of a high-conversion, low-churn user base. In crypto terms, it is akin to a DeFi protocol that retains 90% of its liquidity after a liquidity mining program ends. The market treats such events as bullish, but the real alpha lies in understanding the mechanism that drove the retention.
First, the low download count. The article notes that Perplexity’s app downloads are relatively low in India. Yet revenue grew. This implies that many users are accessing Perplexity via desktop web or mobile browser — not through the app. In crypto, we often see a similar pattern: the on-chain activity is high, but the mobile app traffic is low. This suggests that the product is used for utility, not for casual browsing. The real value is in the search experience, not in the brand stickiness. Users are coming back because the answer engine provides a better outcome than Google or ChatGPT for specific queries — likely those requiring real-time, cited information (e.g., news, research, financial data).
Second, the Airtel channel. Telecom partnerships are not new, but they are notoriously difficult to monetize for AI. Airtel gave away free subscriptions as a customer retention tool. The fact that revenue grew after the promo ended means that a significant fraction of those free users converted to paid. This is a stronger retention signal than any organic growth metric because the initial user acquisition was not self-selected — it was a mass distribution. The users who stayed did so because they found the product indispensable. In crypto, this is akin to a project that sees airdrop recipients actually using the protocol instead of dumping the token.
Third, the unit economics. The Indian Pro subscription is priced lower than the US version, but the cost per query for an AI search engine is also lower due to smaller model inference (Potentially Sonar series) and the use of caching. However, the margin may still be thin. The 60% revenue growth could be masking a scenario where volume increases but profitability per user declines. The article does not disclose gross margin, but based on my experience auditing DeFi protocols during the 2020 liquidity mining craze, I have learned that growth without margin is a time bomb. The key question is: does Perplexity’s per-query cost in India allow for a 30%+ gross margin at the current pricing? If not, the 60% growth is merely a precursor to a future margin squeeze.
Fourth, the hidden information: the download-revenue divergence. The article correctly points out that low downloads could mean desktop web usage. But there is another layer: the Airtel partnership might have driven users to the web version via the Airtel app or browser. This means Perplexity is not dependent on the App Store for distribution — a structural advantage. In crypto, we see this when a protocol attracts users through a partner dApp rather than through a direct frontend. The takeaway: the distribution moat is not the app but the channel. Airtel’s billing relationship with 300 million users is a far stickier distribution layer than any ad campaign.
Fifth, the competitive positioning. The article argues that Perplexity competes not with ChatGPT but with Google AI Overviews. I agree. The 60% growth likely comes from users who need real-time, verifiable answers — the kind that Google’s AI Overviews often fail to provide (e.g., for recent news, niche queries, or multi-step research). In a market where Google’s monopoly is strong, any sign of a viable alternative is a significant narrative shift. The growth is a signal that a segment of users is willing to pay for a better search experience, even when free alternatives exist.
Contrarian Angle: The Fragility of the Signal
Now, the contrarian view — because the market loves a narrative, but the data reveals the cracks.
First, the base effect. A 60% increase from a small base is not impressive in absolute terms. If Perplexity’s India revenue was $100,000 per quarter, 60% growth brings it to $160,000 — still negligible compared to the US market. The article acknowledges this, but many readers will extrapolate too quickly. In crypto, a 100% increase in TVL for a small protocol is often noise, not trend.
Second, the sustainability of the Airtel channel. Airtel likely subsidized the initial free subscriptions, but the revenue growth after the promo may include a one-time catch-up effect: users who wanted to keep the service bought a subscription immediately after the promo ended. The next quarter’s growth could be flat or negative. Without month-over-month data, we cannot extrapolate a trend.
Third, the margin problem. As I mentioned, the unit economics are unclear. AI search is expensive to run. Perplexity’s “Sonar” model and other smaller models reduce cost, but they also reduce quality. If the retention is driven by the ability to use GPT-4 or Claude through Perplexity, then the cost is higher. The article hints that some users may be subscribing to access advanced models, not the search feature itself. This is a value proposition risk: if users are paying for the LLM access, then Perplexity is just a wrapper, and its moat is thin. In crypto, this is like a DeFi aggregator that depends on third-party protocols — if the underlying protocols become better, the aggregator loses its edge.
Fourth, the regulatory and ethical risks. The article’s fifth dimension highlights copyright and hallucination risks. In India, the media landscape is fragmented, and local language search quality is often poor. If Perplexity’s answers cause harm (e.g., medical misinformation), the liability could be significant. The Indian IT rules impose strict content moderation requirements on intermediaries. Perplexity, as a search-like tool, may be considered a “significant social media intermediary” and face compliance burdens. This could force them to invest in costly local servers and moderation, squeezing margins further.
Fifth, the competitive response. Google is not asleep. Google AI Overviews will improve, and Google may also partner with telecom operators. If Jio (India’s largest telecom) inks a deal with ChatGPT or Google, Perplexity’s channel advantage will vanish. In crypto, we see similar dynamics: a first-mover advantage in a specific distribution channel is quickly eroded when larger players enter the same channel with better products.
Takeaway: The Data Reveals the Path, but the Next Quarter Will Tell the Story
Perplexity’s 60% revenue growth in India is a high-signal, low-albatross event. It confirms that there is a viable market for paid AI search in emerging economies, but it does not guarantee that Perplexity is the winner. The real test will come when the next quarter’s data is released: if the growth rate sustains or accelerates, then the retention signal is robust. If it reverts, then the spike was a one-time catch-up from the promo end.
From a crypto analyst’s perspective, this is a textbook case of narrative arbitrage: the market will likely overinterpret the 60% headline, while the smart players will dig into the unit economics, the conversion rate, and the channel dependency. The lesson is: yield is the lie; retention is the truth. Perplexity has shown that it can retain users, but the structure of its revenue — low margin, high channel dependency, and regulatory risk — suggests that the growth may not be as alpha-rich as it seems.
Pivot not panic: The data reveals the path. The next narrative to watch is not Perplexity’s absolute growth, but the evolution of the telecom-AI bundling model. If Airtel expands the partnership, or if other telecoms compete, the market will reward the infrastructure providers of this distribution layer. Perplexity is the frontrunner, but the real prize is the channel itself.
Auditing the code, not the charisma. The article’s analysis is sound, but it lacks the granular data to confirm the thesis. Until we see the conversion rate, the gross margin, and the churn curve, the 60% growth remains a promising signal — not a confirmation.
Floor prices bleed, but structure remains. The structure of high-retention, channel-driven growth is a durable foundation. If Perplexity can maintain this structure, it will survive the coming competitive storm. But the market is fickle, and the next quarter’s data will tell us whether this is a real trend or a mirage.
Narrative follows logic, never precedes it. The logic is clear: the 60% growth is a retention signal, not a growth story. The market will eventually catch up to this distinction. The question is whether you will be positioned correctly before the narrative shifts.