Product1 distinct publisher3 min readPublished
The Information puts Perplexity's annualized revenue above $750m, up from under $250m in January, and the app user counts grew nowhere near that fast, which makes metered credits and enterprise seats the thing a buyer is actually pricing.
The Product Desk · Product desk
Compiled by The Product DeskSomething wrong?How this is made
The person who has to explain the invoice on Friday is looking at a credit balance that empties in nine days when three people on the team run agent tasks, then sits almost untouched until the reset, not a valuation. That credit balance is the unit the revenue story is built on, not a headcount of readers.
Run the two reported numbers against each other. More than $750m annualized divided by the 37.9 million monthly app users Similarweb counted in February is about $20 per app user per year [15]. The same reported valuation over the same base is roughly $790 per app user [16]. Revenue grew at least threefold over 2026 [14] while the app base grew about 45% across six months [10], which implies revenue per app user roughly doubled [19]. The windows do not line up cleanly, the revenue figures run from January to now and the user figures from August 2025 to February 2026, and the traffic numbers are a third-party estimate rather than a company disclosure [10].
What sits inside that per-user increase is partly visible. By March, annual recurring revenue was $450m with some of it coming from what The Information described as tens of thousands of enterprise customers [8], so more than $300m has been added since that mark [17]. The Financial Times reported a 50% revenue increase in the month after Computer shipped [7][5]. Neither figure splits consumer subscriptions from enterprise contracts [20], and the reported Nvidia money is an equity round, not a revenue line [1]. So the growth is priced work rather than more readers. Whether the paying core is procurement departments or heavy individual users is simply not in the reporting.
That matters for who this product is for. The web surface peaked in October 2025 and has been retreating [9], which is the surface most rollouts train people on. The growing surface is the app, and the growing revenue is agentic credits. A team that bought this as a better search box is holding the part that is shrinking.
Private investors are paying roughly 40 times sales [4] in a half-year that Bloomberg reported set records for AI private capital volume [13]. CEO Aravind Srinivas told CNBC in early June that the company still plans to go public in 2028 whatever happens to the Anthropic and OpenAI listings [11]. Fast Company's Mark Sullivan, who wrote in 2024 that Perplexity was most likely an acquisition target rather than a survivor [18], is the one now calling the traffic picture muddled [9].
The forcing function for a buyer is a table with two columns and one month of data. Column one: seats you would renew if every agent feature vanished tomorrow. Column two: credits consumed on tasks that actually finished and got used, counted per task, not per login. Column two is the only evidence of value you own; column one is what you pay for optionality. The open question is which column the vendor's $750m sits in, broken out by segment. If the answer is a total rather than a split, price the renewal as if the growth might belong to somebody else's use case, because on the published evidence it might.
Ranked by verification strength, evidence, and original report placement.
The Information reported that Perplexity's annualized revenue is now more than $750 million, up from under $250 million at the start of 2026.
At $750 million in annualized revenue, a $30 billion valuation is roughly 40 times sales.
In February, Perplexity launched Computer, a product that deploys AI agents to do work on behalf of users.
Perplexity switched to a usage-based pricing plan in which users get a certain amount of credits for agentic work every month.
The Financial Times reported that Perplexity's revenue increased 50% in a month after the launch of Computer.
By March, Perplexity's annual recurring revenue had risen to $450 million, with some of that revenue coming from tens of thousands of enterprise customers.
Distinct publishers with included, body-backed reporting in this cluster.
1 article · September 3, 2026
Follow any of these and your For You feed starts watching them — no settings page required.
invest
Nvidia's Perplexity talks move its money one layer further from its own chips1 distinct publisher
invest
Nvidia's pre-earnings re-rating is a credit story, not a chip story1 distinct publisher
product
Nvidia's talks to price Perplexity above $30bn are a demand contract with equity attached2 distinct publishers
product
Baidu's AI line grew 25 percent and still lost the arithmetic1 distinct publisher
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
One relay, five borrowed sources
Not a single figure in this story was gathered by the outlet publishing it. The Information supplies the revenue, the Financial Times the post-launch jump, Similarweb the traffic and app counts, CNBC the listing quote, Bloomberg the market backdrop — and the $30 billion valuation that anchors everything is introduced with 'reportedly', with no named source and no word from Nvidia or Perplexity. Fast Company is a careful relay and labels each hand-off, which is why this scores as low rather than empty.
Real usage, one-sided disclosure
This is the sturdiest part of the file. There is a shipped product, a pricing change, a revenue path from under $250 million to $450 million by March to more than $750 million, and an app audience Similarweb puts at 37.9 million monthly. Enterprise demand exists but is sized only by the phrase 'tens of thousands'. What no one has published is the retention behind the credit plan or how much of the total is enterprise contracts rather than metered consumer spend — and web visits have been falling since October, so the demand curve is not pointing one way.
Multiple ahead of the measurable audience
The overstatement is not in the prose — the headline hedges, and Sullivan puts the circular-financing objection in his own copy — it is in the arithmetic. A 40x sales multiple is calculated from a valuation that exists only in talks, and the audience it prices grew 45% on mobile while the website shrank. The per-user math makes the stretch visible: about $20 of annualized revenue per app user supporting roughly $790 of valuation per app user, on figures measured months apart.
The chip supplier funds its own customer
The central incentive is structural and Sullivan names it: Nvidia putting money into a company that will spend money on Nvidia compute, in the same pattern it has with OpenAI. Everyone quoted or cited has a position in the outcome — the startup raising, the supplier underwriting, and a private market Bloomberg describes as setting record deal volumes with no slowdown expected. A secondary pull runs through the byline: the author is grading his own 2024 prediction. High score here means strong incentive pressure on the claims, not proven distortion.
Directionally credible, structurally unverified
The direction of travel is believable — a shipped agent product, metered pricing, revenue climbing fast — because three separate outlets are pointed the same way. Confidence stops there. One publisher carries the story, no primary document appears anywhere in it, the round is unsigned, and the audience data ends in February while the piece runs in September. Treat the trend as likely and every specific number as provisional.