Perplexity Shifts to AI Agents, Boosts Revenue

Perplexity’s revenue accelerated sharply after refocusing from an AI-powered search/chat interface toward agentic products and usage-based pricing: estimated annual recurring revenue climbed about 50% in one month and crossed roughly $450 million as of March 2026. The company attributes the lift to new agent tooling and a product-and-pricing reframe.

Technical and product context Perplexity began as a conversational search challenger in 2022 and has layered increasingly agentic capabilities. In 2024 it launched Comet, an AI-assisted web browser that can act on users’ behalf. In February 2026 the company released an agent product called Computer and added Model Council, which lets users compare outputs across models. To reflect the higher compute and orchestration costs of agents, Perplexity moved toward a usage-based pricing model that issues credits to premium users and charges for overage — shifting economics away from flat subscriptions toward consumption billing.

• •Estimated ARR rose ~50% in one month to more than $450 million as of March 2026. (Financial Times/Reuters coverage)
• •Perplexity reports more than 100 million monthly active users and “tens of thousands” of enterprise customers.
• •Subscription tiers range in the reported market coverage from roughly $20 to $200 per month; usage-based credits now top up those tiers for heavy consumption.
• •The company’s market valuation is reported near $20 billion, but it remains smaller in scale than OpenAI, Anthropic and other large model providers.

Perplexity’s results illustrate two operational realities for AI product teams: agentic features can accelerate monetization but materially increase infrastructure and customer-cost complexity; and usage-based pricing can unlock revenue growth quickly but reduces revenue predictability and increases billing/observability requirements. For ML engineers and platform teams, this means prioritizing cost measurement, fine-grained telemetry, model selection for cost-per-task tradeoffs, and tooling for credit/usage accounting.

Monitor retention metrics as usage-based billing widens; watch how Perplexity manages model mix and routing to control inference costs; and observe competitive responses from larger model providers on agent UX and pricing.