Perplexity's Bet on Search Data

Perplexity is trying to win the part of the stack that compounds with every search, not the part that gets leapfrogged every few months. A better model can be bought, licensed, or swapped in, but a live search index, clickstream, follow up query history, and direct habit with users create feedback loops that make the product better with use. That is why Perplexity has pushed beyond answer generation into consumer search, vertical search, and enterprise distribution.1 2 3 4 5

• Perplexity is built around cited, multi sentence search answers, and by late 2024 it was already being framed as more product company than model company. That matters because the core workflow is not just generating text, it is deciding what to fetch, rank, summarize, and show next based on real user behavior.1 2
• The competitive pressure comes from labs with giant model budgets and consumer reach. OpenAI has paired its models with a massive consumer business, reaching an estimated $50B annualized revenue by September 2026, while Anthropic reached an estimated $65B by July 2026. Against players that strong, Perplexity needs a moat outside raw model quality.3 4
• Perplexity has been scaling that wedge fast, from an estimated $63M annualized revenue at the end of 2024 to $148M by June 2025, then $750M by September 2026. The through line is owning more of the user session, first through search, then through shopping, enterprise search, browser, and agent workflows.2 5 1

The next phase is a race to own the interface where research turns into action. If Perplexity can keep turning search sessions into persistent habits and higher value workflows, its data advantage and user relationships become the asset that survives model turnover, even as OpenAI and Anthropic keep improving the underlying engines.5 3 4