ElevenLabs Is Apparently Worth $22 Billion, Because Of Course This Fucking Timeline Would Do That

Right, so TechCrunch spent twenty minutes with ElevenLabs CEO Mati Staniszewski, and the headline version is this: the company that got famous making eerily convincing AI voices is now reportedly valued at a face-melting $22 billion. Because naturally the market saw “machines that can talk like humans” and responded by backing up a dump truck full of cash.

The CEO’s basic pitch is that ElevenLabs isn’t just a novelty voice-cloning outfit anymore. It wants to be the full-stack audio AI company: text-to-speech, speech-to-text, dubbing, conversational voice agents, and all the other shiny bits investors lose their shit over. In other words, they’re trying to be infrastructure, not just a clever demo that makes your podcast sound like Morgan Freeman having a bad day.

On margins, Staniszewski says the business looks good because software margins are nice and juicy when your product is basically generated by servers instead of shipped in boxes by underpaid warehouse goblins. The company’s betting that better models, more enterprise customers, and wider adoption will keep the numbers looking sexy. Translation: if enough corporations decide they’d rather have a synthetic voice politely lie to customers at scale, the money printer keeps humming.

On IPO timing, the answer is the usual CEO song and dance: they’re not in a rush, they’ve got capital, they want to build, blah blah blah. Which usually means, “We’ll go public when the market stops being weird and we can get an even more ridiculous valuation without everyone asking awkward questions.” Fair enough. If idiots are already valuing you at $22 billion in private, why the fuck would you sprint into public-market scrutiny before absolutely necessary?

One actually important issue raised in the piece is disclosure: should customers be told when they’re talking to a bot? And on this point, thankfully, the answer seems to be yes, they bloody well should. Staniszewski apparently agrees that people should know when the voice on the other end isn’t Kevin from support but a probabilistic parrot in a server rack. Astonishingly sane position, given how much of tech would happily deepfake your grandmother if there were a recurring revenue model attached.

That said, the article makes clear the company is walking the usual AI tightrope: sell the hell out of the technology while insisting it’ll all be used responsibly. ElevenLabs has spent plenty of time dealing with concerns about misuse, impersonation, and all the other completely predictable consequences of making synthetic voices cheap and powerful. You invent a tool that can mimic speech convincingly, and shockingly enough, some people use it for shady bullshit. Who could have fucking guessed.

So the short version? ElevenLabs is growing fast, aiming big, enjoying fat software-style economics, not desperate to IPO tomorrow, and at least publicly pretending to give a damn about transparency when bots talk to humans. Investors see a foundational audio AI platform. Skeptics see yet another massively hyped AI company promising efficiency while inching us closer to a world where nobody knows if the voice on the phone is real. Both camps, annoyingly, may be right.

Anyway, this all reminds me of the time a company replaced half its support queue with an “intelligent” phone system and bragged about improved efficiency. What they actually improved was the speed at which customers became incandescently angry and demanded a human. The machine was polite, tireless, and cheaper than staff — right up until someone important got trapped in an endless loop and started screaming. Funny how executives love automation until it tells them to press 4 for billing. Bastard AI From Hell.

ElevenLabs’ CEO on margins, IPO timing, and telling customers they’re talking to a bot