Two Polish Boys Hated How Their Movies Sounded — So They Built a $11 Billion Voice AI Company. The Story of ElevenLabs.

Growing up in Poland in the 1990s, watching American movies meant listening to a specific, peculiar experience. Polish television and cinema had a tradition called "lektor" dubbing — a single male narrator reading all the dialogue of all the characters, male and female, hero and villain, child and elder, in a flat, emotionless monotone over the original English audio. No character voices. No acting. No emotional range. Just one voice, droning over everyone else's, while the original performances — the actual human voices of the actors on screen — murmured underneath, barely audible.

Mati Staniszewski and Piotr Dąbkowski grew up with this. They thought it was terrible. They thought AI could obviously do better. And that thought, planted in childhood, sat quietly in both of them for years before it became the founding insight of one of the fastest-growing AI companies in history.

The two had met as teenagers at Copernicus High School in Warsaw — a school whose name, honouring the great Polish astronomer who reframed humanity's understanding of the universe, would prove fitting for founders who would go on to reframe how the world thinks about artificial voice. They became friends through a shared passion for technology and problem-solving. Their paths then diverged: Mati studied mathematics at Imperial College London before joining Palantir Technologies as a deployment strategist, working with governments and large enterprises to implement cutting-edge technology. Piotr went to Oxford, then Cambridge, focusing on AI and machine learning, published an AI-based image detection thesis at NeurIPS — one of the world's leading machine learning conferences — and then built his engineering career at Opera Software, Google, and Tessian.

They stayed in touch. They experimented together. They built an accent-detection app. They built a recommendation engine. And they kept returning to the same frustration: existing text-to-speech technology was robotic, lifeless, and fundamentally incapable of the emotional range that made human voice communication powerful.

In April 2022, they founded ElevenLabs.

A Year in Stealth, One Name Rooted in Pride

The company's name carries a specific, deliberate meaning. "Eleven" refers to 11 November — Poland's National Independence Day, the eleventh day of the eleventh month. "Labs" indicates the company's research and development identity. The name was a quiet declaration: two Poles, building something ambitious, choosing to honour their country's independence in the very title of what they created.

After founding in April 2022, Mati and Piotr spent the next twelve months in stealth mode. While most AI startups in the 2022 wave were rushing to build interfaces on top of large language models and launch as quickly as possible, ElevenLabs made the opposite choice. They focused entirely on the quality of the underlying voice synthesis models — on solving the hard technical problem of making AI voice sound genuinely human rather than building the minimum viable product that would demonstrate a concept.

The technical challenge they were attacking was not simple. Human speech is not just words. It is pauses, intonation, emotion, laughter, hesitation, breath. A voice that can read text correctly is not the same as a voice that can perform it. Every existing text-to-speech system — from Siri to Alexa to Google's offerings — had solved the former and largely ignored the latter. ElevenLabs wanted to solve both.

Their initial prototypes stood out because they replicated human elements that no other system had attempted: natural pauses, conversational fillers, the subtle emotional colouring that distinguishes excitement from boredom from grief. The models were trained to preserve not just linguistic content but emotional texture — the thing that makes voice communication feel like communication between people rather than between a person and a machine.

They raised a quiet seed round of $2 million. They kept building.

January 2023: The Beta That Changed Everything

In January 2023, ElevenLabs launched its public beta. The response was immediate and extraordinary.

Within five months of launching, ElevenLabs had attracted over one million users. The quality of the voice synthesis — the naturalness, the emotional range, the multilingual fidelity — was so far ahead of anything else publicly available that it spread through the creator, media, and developer communities with the kind of organic velocity that no marketing budget could have purchased.

Content creators discovered they could generate narration for their videos that sounded like a professional voice actor. Publishers discovered they could convert written articles into audio at a quality level that made audio journalism genuinely listenable. Game developers discovered they could give characters voices without recording studios. Audiobook producers discovered they could produce full-length narration from text in hours rather than weeks.

The company moved rapidly from seed funding to venture backing. In July 2023, ElevenLabs raised $19 million in a Series A round led by Andreessen Horowitz — one of Silicon Valley's most respected venture capital firms — with participation from Nat Friedman, former CEO of GitHub, and other prominent technology investors. The company was valued at $99 million.

The growth that followed was unlike almost anything the AI sector had seen. Annual recurring revenue hit $25 million by early 2024. By October 2024, it had reached $90 million. By August 2025, it had crossed $200 million. By May 2026, ElevenLabs had surpassed $500 million in annual recurring revenue — a number that most software companies take a decade to reach. ElevenLabs built it in under three years from public launch.

From Text-to-Speech to Full-Stack Audio AI

ElevenLabs began as a text-to-speech company. It has not remained one.

The platform has expanded systematically into every dimension of AI-powered audio. Voice cloning — the ability to replicate a specific person's voice from a sample — became one of ElevenLabs' most commercially significant capabilities, used by media companies, brands, and creators who wanted to produce audio content at scale in their own voice or the voices of their talent. Speech-to-text through the Scribe model joined the platform, making ElevenLabs a complete audio processing solution rather than a one-directional voice generator.

AI dubbing — the capability that most directly addressed the childhood frustration that sparked the company — became a flagship product. ElevenLabs' dubbing system can take a piece of content produced in one language and produce a genuinely high-quality dubbed version in another, preserving not just the words but the voice characteristics, emotional tone, and speaking style of the original performers. The lektor problem — one flat voice over everyone else's — became something ElevenLabs was specifically built to eliminate.

Sound effects generation joined the catalogue. AI music generation followed through Eleven Music. Voice agents — real-time conversational AI systems that could interact with users through natural-sounding voice — became the ElevenAgents product line. The platform was organised into three distinct product lines: ElevenCreative for content and media producers, ElevenAgents for conversational and enterprise voice applications, and ElevenAPI for developers integrating voice AI into their own products and services.

By 2026, the company had 400 employees, headquartered in London with operations globally. Its February 2026 Series D round raised $500 million at an $11 billion valuation — more than tripling the company's valuation within a single year.

The Marketing Strategy Built on Quality, Community, and Creators

ElevenLabs has built one of the most distinctive marketing approaches in the AI industry — defined less by what it spends on advertising and more by how it has structured its relationship with the people who use its technology.

Quality as the viral mechanism. ElevenLabs' most powerful marketing vehicle has consistently been the product itself. When someone hears an ElevenLabs-generated voice for the first time — whether in a YouTube video, a podcast, an audiobook, or an interactive agent — the quality is frequently so far above the expected baseline for AI voice that the listener asks how it was produced. That question is the beginning of every organic conversation that has driven ElevenLabs' growth. No advertising budget produces that kind of word-of-mouth. Only a product that genuinely surprises people can.

The creator economy as distribution infrastructure. ElevenLabs built its initial user base primarily through the creator community — the YouTubers, podcasters, newsletter writers, and independent media producers who needed professional-quality audio without professional-quality budgets. By offering a free tier that gave creators genuine access to the technology, ElevenLabs embedded itself into the workflow of millions of content producers. Every piece of content those creators published using ElevenLabs' voice technology was implicitly a demonstration of what the platform could do — reaching the audiences of those creators without ElevenLabs spending a dollar to reach them.

The creator success flywheel. As Mati Staniszewski has described the strategy: every creator who earns money from their voice on ElevenLabs becomes a permanent advocate for the platform. They share their success stories, bring in more creators, and defend ElevenLabs against competitors. The network of successful creators is not just a customer base — it is a distributed marketing organisation that grows with the company.

Developer access as market expansion. The ElevenAPI product line made ElevenLabs' voice synthesis capabilities available to any developer who wanted to build voice-enabled applications. By giving developers access to the underlying technology through a well-designed API, ElevenLabs multiplied the number of use cases and user touchpoints for its technology exponentially. Every application built on ElevenLabs' API was another distribution channel — another way for the technology to reach users who had never heard of ElevenLabs but were experiencing its capabilities through a product they already used.

Mission-led positioning around accessibility. The founding mission — to make quality content available across all languages — is not marketing language. It is a genuine statement of what the company was built to do, drawn directly from the childhood experience of watching American movies in a country that could not afford to dub them properly. This mission gives ElevenLabs a purpose-driven narrative that resonates particularly strongly with international markets, minority language communities, and educational applications — and that distinguishes it from AI companies whose mission statements feel like aspirational retrofits.

The Sound of Everything That Comes Next

In February 2026, ElevenLabs raised $500 million at an $11 billion valuation. In May 2026, it surpassed $500 million in annual recurring revenue. The same month, it launched ElevenLabs for Government — bringing secure voice AI deployment to public sector organisations in the UK and beyond. Eleven Music entered general availability. Dubbing v2 expanded multilingual capabilities. The company partnered with the UK Government to integrate voice AI into public services.

The two Polish boys who grew up listening to one monotone narrator read everyone's dialogue over American movies built something that will make such compromises obsolete — not just in Poland, but in every language, in every country, for every creator, educator, and organisation that has ever been constrained by the cost, time, or quality limitations of traditional voice production.

ElevenLabs did not build a text-to-speech tool. It built the infrastructure for how the world will hear itself in the age of AI.

The voice of that future is already here. And it sounds genuinely, unmistakably human.

Founded April 2022, London. Beta launched January 2023. 1 million users in 5 months. $11 billion valuation February 2026. $500M+ ARR May 2026. 400 employees. Built by two childhood friends from Warsaw. Still building the voice of AI.