The startup everyone thought would get swallowed is now worth more than most Fortune 500 companies—and burning through a business model pivot in real time.
The Summary
- Nvidia is in talks to invest in Perplexity at a $30 billion valuation, up 50% from $20 billion a year ago, with annualized revenue hitting $750 million (3x since January 2026)
- The company pivoted from pure search to AI agents via its "Computer" product, switching to usage-based pricing instead of flat subscriptions
- Traffic data tells a different story than revenue: the valuation is 40x sales, and actual user visits remain unclear despite the hockey stick growth narrative
The Signal
Perplexity went from answer engine to agent platform in a year, and the financials suggest the pivot worked. Revenue jumped from under $250 million in January to $750 million annualized by the time Nvidia came knocking. The Financial Times reported a 50% revenue spike in a single month after launching Computer, the AI agent product that does work on your behalf. By March, annual recurring revenue hit $450 million, powered partly by "tens of thousands" of enterprise customers.
This is the Web4 playbook in motion: start with a thin wedge product (AI-powered search), build trust, then sell the thing people actually want (agents that complete tasks, not just summarize information). The usage-based pricing model is critical here. Instead of charging a flat $20/month like everyone else, Perplexity gives you credits for agentic work. You pay for what the agent does, not what the interface looks like. That aligns incentives: if the agents work, revenue scales. If they don't, customers churn before the next billing cycle.
"At $750 million in annualized revenue, a $30 billion valuation is roughly 40 times sales."
But here's the friction: Similarweb data shows Perplexity traffic peaked at 219 million visits per month, then the trail goes cold in the source material. For a company supposedly tripling revenue in nine months, we'd expect to see traffic numbers screaming upward. Either enterprise customers are doing heavy lifting behind login walls (likely), or the revenue growth is coming from higher per-user spend via the credit system (also likely), or both. The murkiness matters because it signals where the real value lives: not in consumer eyeballs, but in B2B seats where agents run workflows at scale.
Nvidia's interest makes sense beyond the obvious (they sell the picks and shovels). Perplexity is stress-testing the agent economy at revenue scale. If usage-based pricing works here, every AI company will copy it within six months. If enterprise customers actually pay $750 million a year for AI that does work instead of AI that chats, that's the proof point the entire agents thesis needs.
The Implication
Watch the pricing model, not the valuation. If Perplexity's usage-based credits become the standard, we're looking at a fundamental shift in how AI companies monetize. Flat subscriptions assume you're selling access. Credits assume you're selling outcomes. The company that figures out how to price agentic work wins the category.
For anyone building in this space: enterprise is where the money is, but only if your agents do something worth metering. Consumer buzz doesn't pay the bills anymore. The era of "AI that does things" just got its first $30 billion data point.