Kagi's Revenue Is a Guess. Its 50,000 Paying Members Aren't.

Quick answer (2026): Kagi, the ad-free paid search engine founded by Vladimir Prelovac in 2018, does not disclose its revenue. Third-party estimates for 2026 diverge: about $7M (Growjo) and $9.1M (ZoomInfo). The hard numbers Kagi actually publishes are member counts: 210 paying customers on day one (June 2022), 20,000 by early 2024, and 50,000 by June 2025, at which point the company said it had been break-even profitable since mid-2024 (Kagi blog, June 2025). It is funded by roughly $3M from the founder (2018 to 2023), a $670K raise from its own users (June 2023), and about $2.5M from around 93 community angels, with no venture capital and no ads. The stated goal: 5 million customers, which the founder says would put Kagi at $1B in revenue.

The one number Kagi refuses to sell

Most search engines have a revenue number because revenue is the whole point of the product. Google made roughly $76B in US ad revenue in 2023 and, by Kagi's own math, extracts about $277 a year from every user by selling their attention to advertisers (Kagi docs). You are not Google's customer. You are its inventory.

Kagi inverted that. It takes no ads, no tracking, and no data monetization, and it charges the person doing the searching. That single decision is why "kagi revenue" is a genuinely hard question. There is no ad line to estimate, no S-1 to read, and no investor deck leaking to a reporter. What Kagi does publish, in unusual detail, is how many people pay it. So the honest way to answer the revenue question is to start with members and work outward, and to be clear about which numbers are stated and which are guessed.

Kagi revenue by year (well, by members): a sourced ledger

Here is every figure with a name attached to it, tagged by date and by source type. Where a number is a third-party estimate rather than a company statement, it says so.

Why the estimates disagree, and which one to trust

The two public estimates sit about 30% apart: Growjo says roughly $7M, ZoomInfo says $9.1M. Neither company has access to Kagi's books; both model it from headcount, web traffic, and pattern-matching against comparable SaaS. The useful move is not to pick one, it is to sanity-check both against the number Kagi does confirm.

Kagi had 50,000 paying members in June 2025. Its plans, per its own pricing page, run at $5 a month (Starter, 300 searches), $10 a month (Professional, unlimited), and $25 a month (Ultimate, which bundles the Kagi Assistant and access to premium language models) (Kagi pricing). If the blended average lands somewhere around $10 to $14 a month, 50,000 members implies roughly $6M to $8.4M in annual run-rate. That bottom-up band brackets Growjo's $7M almost exactly and sits just under ZoomInfo's $9.1M. So the defensible read is a business with an annual run-rate in the high single-digit millions in 2026, most credibly around $7M to $8M, with ZoomInfo's figure plausible only if the pricier Assistant tier is pulling the blended average up.

The founder's own math points the same way. Vlad frames the target as "5 million customers, which would get Kagi to $1bn in revenue," a target he puts 5 to 9 years out (Kagi blog). That implies a blended figure near $200 a customer per year, or about $16 a month, which is where the $25 Assistant tier has to do the heavy lifting. It is a useful reminder that in a subscription business the revenue number is downstream of exactly two things you can count: how many people pay, and how much each one pays.

The economics of charging for search: $0.166 a query

The reason ad-free search is rare is not ideology, it is unit cost. Every query Kagi runs has a real cost, because it pays other indexes and its own infrastructure per call, with no ad impression to cover it. On the $5 Starter tier at 300 searches, one Hacker News commenter did the arithmetic during Kagi's 20,000-member milestone: about $0.166 per query, "pretty slim margins," made worse by the heavier tiers (HN, Jan 2024).

That is the whole business in one line. When the customer is the customer, every heavy user is a cost, not a windfall, and every AI feature bolted on (the Assistant, premium models) is a fresh COGS line rather than free engagement. It is the mirror image of the ad model, where your most active users are your most valuable inventory. Kagi's margin discipline is why break-even took until 2024, six years after founding, and why the founder talks about capping headcount rather than chasing it.

Funded by $3M, then by its own users, never by VC

The funding story is the part that makes Kagi genuinely unusual, and it is worth getting right. Prelovac bootstrapped the company from 2018 to 2023 on roughly $3M of his own money (Ars Technica, Aug 2025). The first outside money was not a venture round; it was $670K raised from Kagi's own users in June 2023 (Kagi blog). A larger community round followed, roughly $2.5M from about 93 angel investors drawn largely from the user base (Crunchbase), and in early 2024 the company reincorporated as a Public Benefit Corporation.

Add it up and Kagi has raised something on the order of $6M total, almost none of it from institutional venture capital, to build a profitable search engine now serving 50,000 paying members. For comparison, that is roughly what a single seed round buys a pre-revenue AI startup in 2026. The absence of a VC on the cap table is not a footnote here; it is the product strategy. A venture-backed search engine would have been pushed toward the only model that returns a fund at search scale, which is ads. Being funded by users instead of funds is what lets Kagi keep charging users instead of selling them.

What operators can take from the Kagi model

The transferable lesson is not "reject VC." It is that your funding source quietly writes your product roadmap, and Kagi is the cleanest live example of choosing the source on purpose.

• Whoever funds you is who you are built to serve. Ad money means you optimize for advertisers; venture money means you optimize for the growth curve an exit requires; user money means you optimize for the user, because they can cancel. Kagi picked the constraint it wanted before it picked the product.
• Profitability can be the radical feature. In a category defined by "free," being the one you pay for is the entire positioning. Kagi did not reach break-even by accident at 50,000 members; the paid model forced it, the way killing a free plan can force a smaller SaaS to finally price for value.
• Count members, not press-release revenue. Kagi's disclosures are almost entirely member counts and profitability status, not headline ARR, which is exactly the honest way to run a subscription narrative. Revenue is a lagging output; paying members and per-user cost are the inputs you actually control.

None of this makes the $7M-to-$9M question resolvable to a single figure, and that is the point. For a private, ad-free, user-funded company, the third-party estimates are educated guesses; the only hard numbers are the 50,000 people paying, the mid-2024 break-even, and the roughly $0.166 it costs to answer a single search. Kagi is the search engine you can see the economics of, precisely because it decided you should be the one paying for them.

This is an editorial estimate assembled from public sources, not a company disclosure. Kagi does not publish revenue figures; where numbers are third-party estimates they are labeled as such above.

Keep reading

• Plausible Analytics revenue: $1M ARR built against Google: the same paid-and-private-versus-Google playbook, one layer down the stack.
• Fathom Analytics: the revenue number they refuse to publish: another privacy-first team that keeps its ARR deliberately dark.
• Zoho revenue: the $1.4B number built on zero VC: what "no venture capital" looks like at 100x Kagi's scale.