ElevenLabs CEO/Co-Founder, Mati Staniszewski:The Untold Story of Europe’s Fastest Growing AI Startup
Summary
- ElevenLabs says it has crossed $200 million in revenue with roughly 250 employees, after ending 2023 near $35 million. Mati described about 20 months to $100 million and initially “around 10 months” to $200 million, then corrected that second leg to “a bit longer, 15 months maybe.” Enterprise is now the majority of revenue, the largest contract is around $2 million, and Mati cautioned that growth “can also go quickly down.”
- The moat is not a permanent model lead but the product and distribution built during a six-to-12-month research advantage. ElevenLabs concentrates scarce voice talent—Mati estimates only 50-100 researchers operate at the highest level—and turns models rapidly into production workflows. His answer to “Why won’t OpenAI do this?” is that it will do something, but ElevenLabs relies on its exceptional team, focus, execution speed and product layer.
- Voice agents could become a multi-billion-dollar business for ElevenLabs “if we play it right.” The company is moving from speech components into knowledge bases, functions, testing, monitoring and enterprise integrations such as Salesforce, ServiceNow and SIP trunking, with email and WhatsApp potentially extending the platform into omnichannel support. Mati expects routine scheduling and refunds to automate first while high-stakes, domain-heavy work remains human.
- The company’s financing story moved from speaking with 30-50 pre-seed investors to US investors competing through speed and operational help. ElevenLabs raised $2 million at a $9 million post-money valuation in 2022, then $19 million in 2023 after its launch broke out; a later round was priced at $3.3 billion. Mati says American firms are “playing a different game”: they ask how to bet bigger, test whether they can help before investing and, in his reference checks, generally showed stronger evidence of supporting founders through failure than some European investors.
- Owning training infrastructure is an economic and strategic bet, not a vanity project. ElevenLabs calculated that continuously training models and moving large datasets through rented infrastructure would make an owned data center break even over roughly two years; ownership now enables faster experimentation and greater control. Mati concedes that hardware innovation could invalidate the equation, while arguing that a new model should initially optimize for “magic” before cost.
- Mati’s European-company thesis is to build from Europe without building only for Europe. He calls Europe “hard mode,” rejects the idea that Europeans inherently work less hard, and says its ambitious talent is underused. ElevenLabs develops local leaders, supplements them with experienced advisers and operates roughly 20 teams. The tension is that “small and mighty” now means growing from 250 toward 400 people while opening small teams in Brazil, Japan, India and Mexico.
- Liquidity is being used to support employee risk-taking and a longer-duration bet. ElevenLabs has received acquisition approaches but became a “flat no” after learning from the first processes; nearly every financing includes secondary liquidity or a tender for vested employee shares. In the host’s private-market choice among OpenAI at $300 billion, Anthropic at $170 billion and xAI at $120 billion, Mati would buy OpenAI but declined to name a sale, while saying Anthropic is especially compelling in coding and Google is “definitely in the race,” though he is not “super bullish” on it.
Deep dive
1. Poland taught Mati that talent density changes ambition
Growing up outside Warsaw gave Mati a small initial frame of reference, then high school in Warsaw revealed “what’s possible.” Moving from a local public school into a cohort that won competitions made the density of ambitious peers itself motivating—the organizational quality ElevenLabs now tries to reproduce.
His older brother “trailblazed” by studying abroad, while friends and his future co-founder reinforced the belief that elite universities and difficult exams might be reachable. Mati’s account of hunger is therefore less a solitary-founder myth than a cycle of people repeatedly pushing one another to “go after it.”
He still frames company building as climbing successive hills and mountains: each summit reveals how much remains unseen. That posture matters because the company’s stated aspiration is global even though its deepest talent base remains European.
2. A Polish dubbing problem exposed a broader voice market
Before ElevenLabs, Mati worked at Palantir and his co-founder at Google; their hack weekends explored recommendations, a crypto risk analyzer that “didn’t work very well,” and an audio tool that analyzed speaking style. The last project showed them how much unbuilt technology existed around voice.
The catalytic observation arrived in late 2021: in Poland, films were still commonly presented with one flat narrator speaking over every original character, regardless of gender or emotion—“an audiobook reading of a movie.” They believed future dubbing would retain each voice’s emotion, intonation and identity.
Research and customer discovery ran in parallel. Mati scraped and personalized thousands of emails to YouTubers, receiving roughly a 15% reply rate, but the reaction was lukewarm: creators doubted the technology, requested samples and asked how they could operationalize multilingual tracks when YouTube did not support them cleanly.
The sharper demand was simpler than dubbing. Creators wanted to correct a misspoken line, preview how a script would sound, or voice an entire video without recording themselves; meanwhile existing models sat visibly in the “uncanny valley.” ElevenLabs therefore shifted away from the initial dubbing focus and built its own expressive text-to-speech model while continuing toward the broader voice platform.
3. Model leadership buys time; integrated products must capture it
Had today’s architecture existed in 2022, Mati says ElevenLabs would have skipped several single-modality steps. Its more recent generation, Eleven v3, reflects the multimodal direction of combining reasoning and speech to produce a richer voice experience than a dedicated speech model alone.
Harry argued that foundation-model progress appears to be plateauing, with GPT-5 discussion shifting from dramatic new capabilities toward cost and efficiency. Mati agreed that narration is nearing a ceiling—new generations may not change audiobook narration drastically—and said LLM development is flattening to some extent, even as AI adoption accelerates.
Voice remains earlier on its curve, in Mati’s view, but “if you just do research, eventually it will commoditize.” Research therefore provides only a head start: he estimates ElevenLabs leads by roughly six to 12 months depending on the use case, enough time to construct a superior product and ecosystem before competitors catch up.
His answer to OpenAI is deliberately not denial: “They definitely will do something.” ElevenLabs instead relies on focus, perhaps 50-100 truly elite voice researchers worldwide—of whom Mati believes it employs five to 10—and a production layer spanning creative editing, knowledge bases, functions, deployment, testing, evaluation and monitoring.
4. The pre-seed was difficult because investors doubted all three layers
ElevenLabs spoke with between 30 and 50 investors during the pre-seed process, and investors questioned the load-bearing issues: could two founders solve the research, was AI voice too small a market, and could any advantage survive competition from incumbent model companies?
After receiving an offer from a US accelerator other than YC in early 2022, the founders declined and continued independently. Their Google and Palantir savings funded GPUs and the first hires, making the decision progressively riskier as they wanted to accelerate research.
ElevenLabs ultimately raised $2 million at a $9 million post-money valuation in 2022. Mati recalls the first investor’s allocation as exactly 11%—with other investors subsequently layered into the round—and names early believers including Credo Ventures, Concept Ventures and Oxford friend and Polkadot co-founder Peter Czaban.
Capital went into a small data center in Poland and two additional hires before infrastructure expanded into the US. Although the round closed around Q3 2022, ElevenLabs held the announcement until its January 2023 beta because a financing announcement should “have another purpose”: launch a product, establish customers or introduce research.
5. The first real demand signal came from creators, not press
One early blog post demonstrated “the first AI that could laugh.” Newsletters circulated its samples, and roughly 1,000 people joined the waiting list the following day—a qualitatively stronger response than the earlier dubbing outreach.
Among the first hundred testers, an audiobook author copied his manuscript through a tweet-sized text box roughly 500 times, downloaded every clip and stitched the result together. At the time, AI content was banned, but it passed through as human content, received strong reviews and prompted him to invite other authors.
That was enough to show users loved the product, but Mati resists declaring instantaneous product-market fit. His stricter test is whether a product can remain self-sustaining and valuable for five to 10 years; ElevenLabs is “closer to that” today, while still believing much more value remains unbuilt.
Traditional press produced almost no user impact despite extensive preparation. Newsletters, YouTube communities, Discord, Reddit and Hacker News mattered far more; after launch, Mati’s preferred discipline is to serve users, line investors up for a defined future window and reengage when capital is needed instead of living in “continuous fundraising mode.”
6. US investors won by demonstrating partnership before the term sheet
By March 2023, earlier investors were returning, and ElevenLabs ultimately raised roughly $19 million in 2023 from a16z, Brian Kim and NFDG’s Nat Friedman and Daniel Gross. The founders wanted both globally legible credibility—especially in San Francisco—and people they admired who had created something special.
Brian Kim flew to London, where the founders signed a preliminary term sheet together. a16z also showed interest before investing by making introductions, including to celebrities who could work on voice licensing. Mati’s conclusion is blunt: “The speed of execution, speed of investing is the only thing you have.”
Harry dislikes founder roadshows and offered the romantic alternative of an immediate partnership. Mati’s pushback—worth keeping—is that first-time founders do not know their market value or whether an investor behaves as promised; a tight process with a few priority firms, sometimes preceded by one or two lower-priority conversations, provides necessary comparison.
Mati dislikes exploding term sheets but understands why smaller funds fear becoming stalking horses. Harry argued that a very large pricing gap could justify reopening a process, whereas a difference of roughly 20% usually should not drive the decision. Mati’s deeper diligence is to back-reference how a partner acts when things fail; he says checks on some European investors did not yield results as positive as those for the US partners he considered.
7. Small teams and missing titles are designed to preserve ownership
ElevenLabs has roughly 250 people but operates more like 20 teams of five to 10, each responsible for a product area or operating domain. Studio, voice agents, enterprise components, self-serve products and talent functions are sharded into units with substantial independence that can “iterate with reality” quickly.
Formal titles were removed because impact should not depend on tenure, small teams otherwise generate distracting title inflation, and a new employee should be able to become a leader rapidly. Each team still has a decision-maker, but that role can change rather than becoming an honorific that stays forever.
The founders continue interviewing every hire and hope to interview 1,000 people; researchers remain the hardest role. Mati’s biggest hiring regret is waiting after early doubts emerge: if uncertainty persists through the first weeks or months, he now believes separation should happen quickly.
Harry challenged the “small and mighty” label after Mati projected about 400 employees by year-end, with 30-50 offers or new joiners already in view. Mati argues that small outposts in Brazil, Japan, India and Mexico can parallelize expansion; revenue per head matters eventually, but near-term hiring is justified when it improves distribution and retention before competitors arrive.
8. Losing the dubbing launch forced an honest cultural reset
After building text-to-speech, voice recreation and internal speech-to-text, ElevenLabs gave a customer the components needed for dubbing and disclosed that its own combined launch was coming. The customer assigned the combination to an intern as a hack-weekend project and launched about two weeks earlier, reportedly generating tens of millions in revenue over that period.
The blow struck every group differently: research and engineering wondered how an external team shipped first, go-to-market feared losing potential accounts, and the founders watched the company’s original story attract attention elsewhere. “Dubbing was our story,” Mati recalled, and morale visibly collapsed.
His lesson is not to jump directly into reassurance. Leaders should acknowledge, “We are angry at ourselves,” examine what went wrong and only then move quickly into action; relentless execution can repair a one-time mistake, but repetition should carry consequences.
9. Owned compute and product depth are the economic counterweight to commoditization
Mati divides the company’s advantage into research, product and an ecosystem combining distribution with brand. Research creates the six-to-12-month lead; product turns that temporary gap into workflow ownership; distribution and brand strengthen the resulting position.
ElevenLabs calculated that continuous training plus large data transfers would make owned infrastructure break even against rental over roughly two years, assuming GPU improvements were not too extreme. The bet paid off through faster experiments and control, though Mati explicitly allows that future infrastructure innovation “might break that equation.”
He accepts that many AI application companies currently have poor unit economics. Yet a new ElevenLabs model may deliberately ship before cost is optimized so users experience the “magic” first; falling model costs, trusted brands and customer signal are part of the strategy, although at least one competitor in crowded categories will likely lose.
Harry’s criticism was that a horizontal launch lacked a defined ICP. Mati’s rule is conditional: when the technology is genuinely new and its best customer is unknown, horizontal discovery is reasonable; when founders possess domain expertise and know which category they intend to win, “go vertical.”
10. Europe is hard mode only if “from Europe” becomes “for Europe”
Mati agrees that building in Europe is “building on hard mode,” but rejects the claim that Europeans inherently work less. ElevenLabs has found Central and Eastern European “missionaries” who work weekends and care about the company beyond their assigned hours—sometimes more intensely than West Coast hires.
The advantage is underused talent that wants to build an ambitious global company but historically had to join US employers to do so. ElevenLabs describes itself as global: it intends to win in the US, Europe and Asia while retaining most of its team in Europe.
Rather than import every titled executive who has seen scale, the company prefers to grow existing people and pair them with experienced advisers from its US investors’ networks. The bet is on internal growth, with mentorship supplying experience that the local ecosystem may lack.
As a hypothetical “president of Europe,” Mati would largely proxy AI regulation to the US approach, despite acknowledging “a huge set of repercussions.” His fallback would be an opt-in European jurisdiction governed by those rules, giving builders a less restrictive place to operate without imposing the model everywhere.
11. At $200 million in revenue, agents—not narration—become the larger wager
ElevenLabs ended 2023 at roughly $35 million in revenue and now says it has crossed $200 million. Mati estimated about 20 months to $100 million and initially said around 10 months to $200 million, then revised that to “a bit longer, 15 months maybe”; asked whether speed of ARR growth is a [redacted] metric, he said it depends on the time horizon but in general does not matter.
Large enterprises now provide the majority of revenue, which Mati calls relatively sticky, while creators and developers remain a substantial self-serve distribution engine. The largest contract is around $2 million, typically in call centers, customer support or personal assistance; named deployments or relationships include Cisco, Twilio and Epic Games.
The later financing was priced at $3.3 billion. Mati first placed the business around $100-120 million in revenue near the transaction, then clarified that the term sheet arrived when they were probably at roughly $80 million; the October 2024 deal, announced in January 2025, was about 30 times current revenue. Proceeds accelerated multimodal models, international expansion and enterprise-agent integrations.
Voice agents are already a large business but, “if we play it right,” could become a multi-billion-dollar business. ElevenLabs may expand into email and WhatsApp, while its current Decagon partnership could overlap if one party verticalizes; routine appointments and refunds automate first, but high-stakes patient guidance still demands specialized humans.
12. Liquidity and selective risk-taking are preserving independence
ElevenLabs has received acquisition approaches, with the first arriving around its Series A. The founders were briefly curious about the process rather than eager to sell and subsequently became a “flat no”; now they are considering the inverse risk—acquiring a company worth hundreds of millions—while believing they might build the capability better internally.
Nearly every round includes secondary liquidity and a tender offer for vested employee shares. Mati’s reasoning is that childcare, housing and a basic level of a good life should be covered so employees can rationally keep betting on a much larger outcome instead of selling the company early.
Forced to choose among OpenAI at $300 billion, Anthropic at $170 billion and xAI at $120 billion, Mati would buy OpenAI but declined to supply a negative pick. He uses ChatGPT heavily, admires Anthropic’s coding focus, says most people at ElevenLabs use Cursor, and considers Google “definitely in the race”—though explicitly “not super bullish”—with Google’s Gemini 3 models among the innovations he admires.
His clearest recent change of mind is that ElevenLabs can explore products using outside research even when it is also pursuing the underlying research internally. He remains unsure whether founder brand elevates the team or steals attention from it, but is warming to the former; the risk maxim he retains is that the biggest risk can be not taking the risk.