Is Perplexity the first AI unicorn to fail? (medium.com)

At a recent San Francisco AI conference, founders and investors voted Perplexity—once a poster child of AI search—as the most likely billion-dollar startup to fail, ahead of even OpenAI. Perplexity (founded 2022) hit a $20B valuation by September 2025, claiming 780M queries/month, 30M+ active users and nearly $1.5B raised amid frequent financings. But critics call it a “wrapper”: it packaged third‑party LLMs into a search product and rode early novelty rather than owning the stack. That edge evaporated after ChatGPT added web search (Feb 2025) and OpenAI launched ChatGPT Atlas (a browser with integrated AI search), while Google pushed Gemini 2.5 across Chrome/Android—leverage Perplexity lacks. The company’s distribution and unit-economics problems are acute. A high-profile Airtel deal offered 360M Indian users a free one‑year Perplexity Pro (normally ₹17,000/year), inflating user counts but likely yielding poor paid-conversion in a price‑sensitive market. Perplexity’s reported interest in buying Chrome for $34.5B (unrealized) underscores the desperation for reach. For the AI/ML community this is a warning: model ownership and durable distribution matter as much as UX. Wrappers dependent on third‑party models and vanity metrics face rapid disruption once platform owners integrate AI natively, and valuations based on transient usage can collapse when conversion and margins are tested.