20 minutes with the CEO of ElevenLabs, now reportedly valued at $22B
TechCrunch Connie Loizos
ElevenLabs says it’s hit $600M in annual revenue and is now reportedly valued at $22B. Its voice tech is already on customer-service lines, government calls, and creator tools.
Based on reporting by TechCrunch, Connie Loizos — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
ElevenLabs has quietly become one of the default voices of AI. Its models turn text into speech that sounds human, and that’s exactly what users hear when they call support lines run by companies like Klarna, Deutsche Telekom, Cisco, and Adobe. Governments are on the list too. So are creators using it for audiobooks, dubbing, and music.
The company says it’s now pacing at $600 million in annual recurring revenue, and its backers reportedly value it at $22 billion. That’s a big number for a company that says it’s only four years old, especially in a field where customers are starting to overlap with the vendors they buy from. Decagon, for example, trained its voice product on ElevenLabs and now competes with it.
I spoke with co-founder and CEO Mati Staniszewski at Nrth in Toronto, the conference formerly known as Elevate. He’s still betting that voice will stay technically hard for a while. Audio models may get cheaper and more similar over the next three to five years, he said, but there’s still a meaningful quality gap today. His bigger goal is more ambitious: getting to conversational AI that can pass a Turing test by handling not just information, but emotion, pacing, and tone.
That practical streak runs through the rest of the conversation. On enterprise, Staniszewski said a little more than half of ElevenLabs’ business comes from classic enterprise customers, with the rest spread across small and medium businesses, developers, builders, and creators. On margins, he was fine with them getting squeezed if that helps the company win more market share. On disclosure, he said businesses should tell people when they’re talking to an AI, at least for now.
The company also seems comfortable with a world where customers pick from different model types depending on the job. Informational customer service can lean on open source. Financial services, where there’s no room for error, still pushes people toward frontier models. And in public-sector deployments, ElevenLabs says it adapts to local requirements, including data residency, as in a healthcare case in Poland where agents call patients to remind them about appointments.
My take — AI-written commentary, not fact-checked reporting
ElevenLabs is making the sensible bet while everyone else debates philosophy: own the boring infrastructure, sell to enterprises, and let the hype crowd fight over chatbot poetry. The real tell is that it’s happy to let margins shrink if that buys share; that’s how a platform wins before the market gets polite. And yes, telling people they’re talking to an AI should still be the rule, not the twist ending.
Read more about this at: TechCrunch