News

From High School Friends to Billion-Dollar Startup: The Story of ElevenLabs

For years, voice was one of AI’s weakest links. Machines could process text at scale, but synthetic speech still sounded flat and robotic. Voice assistants were useful, not convincing. According to Iain Martin, who recently wrote about the company for ForbesDealroom has a profile for this one. Try Dealroom → , ElevenLabs entered this gap in 2022 with a narrow ambition: to make AI voices sound human.

The company was founded by Mati StaniszewskiDealroom has a profile for this one. Try Dealroom → and Piotr DabkowskiDealroom has a profile for this one. Try Dealroom →, two Polish engineers who had worked at Palantir and Google. Their insight was partly cultural. In Poland, foreign films are often dubbed by a single monotone narrator, a cost-saving practice that strips performances of emotion. It underscored a simple idea: voice is not a feature—it shapes how people experience content.

From the start, ElevenLabs focused obsessively on expressive speech. Its first text-to-speech model, released in early 2023, stood out immediately. The voices could pause, emote and laugh, and they could be cloned. Adoption spread quickly among authors creating audiobooks, YouTube creators translating videos, and app developers embedding natural speech into their products.

As usage grew, so did revenue. Roughly half of ElevenLabs’ business now comes from enterprises such as Cisco, Twilio and AdeccoDealroom has a profile for this one. Try Dealroom →, which use its technology for customer service and recruiting. The other half comes from creators and media companies. Over the last twelve months, the company generated about $193 million in revenue and, unlike most AI startups, remained profitable, with Forbes estimating net income of around $116 million.

That product quality translated into pricing power. ElevenLabs charges significantly more than many rivals, yet continues to win customers while competing directly with Google, Microsoft, Amazon and OpenAI. Independent tests have shown its models making fewer errors than competing voice systems, allowing the company to position itself as a premium provider rather than a low-cost alternative.

Pricing power also came with infrastructure demands. Early on, ElevenLabs chose to buy its own GPUs and install them in a self-operated data center in Poland. Dabkowski selected a location several hours from Warsaw to keep energy and real estate costs low. As the company expanded beyond voice into music and video, it pushed further, investing heavily in capacity and spending $50 million on a data center project in Oregon.

Today, ElevenLabs is broadening its ambitions. It has launched an AI music generator, plans to introduce video avatars, and is positioning itself as a platform for building and managing AI agents. The bet is that voice will become a core interface for AI systems—and that controlling high-quality speech, backed by owned infrastructure, can remain a durable advantage even as competition intensifies.

Source:
Forbes PL
Forbes

Image Source:
Forbes

A.M. & D.D.

More top stories