ElevenLabs Got Famous on Celebrity Voices — Now the Businesses Pay the Bills

You've seen the viral clips: a movie star's voice, cloned by ElevenLabs, saying something ridiculous on a phone. That is the brand almost everyone knows — a goofy consumer toy that makes celebrities talk. Which is why the headline lands like a contradiction: ElevenLabs now takes more money from businesses than from the consumers who made it famous.

The flip matters even if you can't buy the stock yet (it's private, though there's secondary access through brokers like Robinhood Ventures). Because the whole investment question — whether this thing is worth $11 billion, or the $22 billion its insiders are reportedly angling for — turns on which side of the house is actually paying.

The money is real and it's accelerating

Let's get the size first. ElevenLabs was founded in 2022 by two Polish ex-Google engineers and hit roughly $100 million in annual recurring revenue after about 20 months, then $200 million ten months later, then around $300 million within five months of that. By the close of 2025 the CEO was quoting about $330 million. Growth did not stall: the company added over $100 million in net new ARR in the first quarter of 2026 alone, crossed $500 million by mid-spring, and by midsummer the run rate was being cited near $600 million.

The valuation moved with it. A $500 million Series D led by Sequoia in February 2026 put a $11 billion price on the company — a better than threefold jump from the $3.3 billion round a year earlier. By July there were reports of early talks to sell employee shares at around a $22 billion mark. That's the number a retail investor actually has to come to terms with: roughly 35 to 40 times a ~$600 million run rate. Rich by any standard, but AI hypergrowth companies are priced on growth, not current multiples.

Where the revenue actually lives

Here's where the headline does its real work. The consumer product is the famous sliver. Analysts who track the privately-held firm estimate the business roughly as follows: enterprise voice-agent contracts about a third of revenue, the developer API about a quarter, dubbing and voice licensing the rest — with consumer and creator subscriptions, the $22-per-month tier and up, the remainder. The company doesn't publish an exact split, and other estimates see business and consumer closer to half and half. But every tracker agrees on the direction and the driver: enterprise deployment is what's pulling the revenue, and consumer is the runway that feeds it.

Forty-one percent of Fortune 500 companies use ElevenLabs in some capacity. Customers have deployed more than two million conversational AI agents that have handled tens of millions of conversations. The use cases read like a call-center revolution: Revolut putting voice agents on customer support for over four million customers and reporting eight times faster resolution, Klarna doing the same for its U.S. base at up to ten times faster. Enterprises sign contracts worth up to two million dollars a year.

Why businesses pay

The consumer lens makes no sense of this. A lone YouTuber paying $22 a month doesn't build a $600 million business. What does is replacing a human being on the phone, which is why the per-character API pricing — the plumbing — sits at the heart of the model. Charging per unit of voice, rather than per seat, means the price scales with how much value a business actually extracts. Eight times faster resolvement, tens of millions of conversations: that is a labor-cost story, and it justifies enterprise spend that a consumer subscription never could.

Pricing power shows up in the same place. ElevenLabs reportedly charges up to three times more than American rivals for its models because the quality is better, and enterprise customers have pushed its effective revenue per API call up by about 20% as they migrate to higher-value voice-agent workloads. When a company can raise its unit price and see customers follow, that's a real moat signal in a sector otherwise sliding toward commoditization.

The deciding question

The consumer-versus-business mix is really a question about durability, and it points straight at the biggest risk. ElevenLabs builds the application, the "conversation layer," but it relies on foundation models from the very labs it competes with — OpenAI, Anthropic, Google. Those are competitors with effectively unlimited capital, and voice is a large enough prize that they will chase it. ElevenLabs' bet, stated plainly by its own CEO, is that owning the conversation beats owning the model: build the layer businesses actually plug into, stay model-agnostic so customers aren't locked to a single vendor, and build the data moat from one-thousand-plus labeling contractors and a deep integration ecosystem.

That's the "Adapt or Die" wager in real time. If the layer wins — if businesses keep choosing ElevenLabs as the interface regardless of whose model is underneath — then the enterprise-heavy mix is a durable, high-margin software business, and the multiple is defensible. If a frontier lab ships a voice layer that's good enough, the enterprise contracts at $2 million apiece become contestable, and a 35-to-40-times revenue valuation leaves no room to be wrong.

Forbes, in a late-2025 profile, framed the same trade in even starker terms: the company was reportedly profitable, a rarity among AI peers, generating an estimated $116 million of net income on about $193 million of trailing revenue — while staring at rivals with "virtually unlimited resources".

What to watch

Don't judge this business by the celebrity voice clips. Those are marketing dressed as product — top-of-funnel awareness that converts enterprise leads. Judge it like a B2B software company: watch whether the enterprise voice-agent contracts renew and expand, watch whether the per-unit pricing holds against OpenAI and Google, and watch what gross margins the company eventually discloses — it hasn't yet, and in a business built on expensive model inference, that blank is the most important number a future public offering could reveal. The shift from consumer to enterprise is the story because it tells an investor where the value actually got built. Whether it stays there is the question no valuation multiple can answer yet.