🎥 Watch the full episode here

📆 Published: September 8th 2025 🕒

Estimated Reading Time: 8 mins. Time saved: 112 mins! 🔥

🎯 From £2M at £9M Valuation to £200M ARR: The Untold Story of ElevenLabs

In just four years, ElevenLabs has gone from two Polish founders struggling to raise a pre-seed round to one of Europe's fastest-growing AI companies, crossing £200 million in annual recurring revenue. Co-founder Mati Staniszewski recently joined Harry Stebbings to share the raw, unfiltered story of building a global AI powerhouse from Europe – including the brutal pre-seed fundraise where 30-50 investors said no, the devastating moment a customer launched their core product before they could, and why they believe voice will become the primary interface for technology.

This conversation matters now because ElevenLabs represents a new breed of European tech company: globally ambitious, research-led, and competing head-to-head with Silicon Valley giants. With a team of just 250 people generating £200M ARR, they're proving that European companies can achieve Silicon Valley-scale outcomes whilst building from Europe.

📊 OVERALL SUMMARY

ElevenLabs has achieved what many thought impossible: building a globally dominant AI company from Europe. The company went from £35M ARR at the end of 2023 to £200M ARR in late 2025, taking just 20 months to reach £100M and another 10-15 months to £200M. With only 250 employees, they're generating nearly £1M ARR per employee – exceptional unit economics in the AI space.

Key milestones:

• Pre-seed (2022): Raised £2M at £9M valuation after 30-50 rejections

• Beta launch (January 2023): Immediate product-market fit with creators and audiobook authors

• Series A (June 2023): £19M from Andreessen Horowitz, Nat Friedman, and Daniel Gross

• Series B (2024): Led by Sequoia at £3.3B valuation (30x revenue multiple)

• Current state: £200M ARR, 250 employees, expanding into voice agents for enterprise

The company's success stems from three core pillars: world-class research (maintaining a 6-12 month lead), exceptional product execution (self-serve creator tools plus enterprise voice agents), and strategic distribution (horizontal go-to-market capturing creators, developers, and enterprises simultaneously). Major partnerships include Cisco, Twilio, and Epic Games, with largest contracts around £2M annually.

🏗️ The Origin Story: From Polish Suburbs to Global Ambition

Growing up in Warsaw's suburbs, Mati met co-founder Piotr in a competitive high school that required winning competitions and passing multiple selection steps. "That density of talent was the most motivating factor to explore more, learn more and do more," Mati explains.

After high school, Piotr went to Google, Mati to Palantir. They maintained "hack weekend projects" exploring new technologies. One project in early 2021 focused on audio – analysing speech patterns and giving improvement tips. The catalyst came when returning to Poland and experiencing terrible movie dubbing, where a single monotone voice narrated all characters.

Mati emailed thousands of YouTubers asking if they'd want a dubbing product. The response was lukewarm, but revealed something more valuable: creators wanted to correct mistakes in post-production, preview scripts, and create voiceovers without speaking. Meanwhile, Piotr realised existing text-to-speech models were stuck in the "uncanny valley" – they needed to build completely new models from scratch.

Key Takeaways:

• Density of talent matters more than individual brilliance – surround yourself with people who push you

• Hack weekends with co-founders build shared technical intuition before committing

• Customer discovery can reveal better problems to solve than your original idea

• Sometimes you need to build the research layer yourself rather than relying on existing models

💰 The Brutal Pre-Seed: 30-50 Rejections and a £9M Valuation

The pre-seed fundraise in 2022 was brutal. They spoke with 30-50 investors. The objections: "How will you fix the research?" "The market is very small." "What about defensibility against Amazon and Google?"

The pressure intensified because they'd rejected a US accelerator offer, burning through Google and Palantir savings whilst hiring employees and buying GPUs. Eventually, Credo Ventures, Concept Ventures, and Peter Czaban (Polkadot co-founder) believed. They raised £2M at £9M post-money.

The strategic move: they raised in Q3 2022 but didn't announce until Q1 2023. "We don't want to announce a round just for the sake of announcing a round," Mati explains. They held the announcement until their beta release was ready, then announced both simultaneously.

Key Takeaways:

• Expect 30-50 rejections in pre-seed for a novel category – this is normal, not failure

• Announce funding rounds alongside product launches to maximise impact

• Rejecting accelerators can work if you have savings and conviction – but increases pressure

• Early believers matter more than brand-name investors at pre-seed

🚀 Beta Launch: The Moment Product-Market Fit Hit

January 2023 marked the public beta launch. They sent samples to newsletters, and overnight, 1,000 people joined their waiting list. Then came the powerful signal: an audiobook author copy-pasted his entire 500-page book into their text box, section by section, downloaded each narration, stitched them together, and released it. It got positive reviews. "I was like, okay, we are onto something," Mati says.

On launch strategy, Mati learned traditional press doesn't matter nearly as much as founders think. They prepared extensively for a major publication feature – it had "no impact." What worked? AI newsletters, Discord, Reddit, and Hacker News. "Those people picked it up quicker than anyone else."

The crucial lesson: don't get distracted by post-launch attention. "If the launch goes well, you will get more interest from investors, from events, from media. This is the time you should really focus on getting the product right, and all of that is a distraction."

Key Takeaways:

• Unexpected use cases signal strong product-market fit – the audiobook hack showed desperate demand

• Traditional press is overrated for B2B/developer products – focus on communities where users actually are

• Product-market fit is about 5-10 year sustainability, not initial enthusiasm

• Post-launch attention is a distraction – stay focused on users, not press and investors

🎯 Series A: How Andreessen Horowitz and NFDG Won the Deal

By March 2023, just two months after beta launch, investor interest surged. But Mati and Piotr were deliberate: "Let's optimise for the partners we truly want."

Enter Brian Kim from Andreessen Horowitz. He flew to London and opened with: "I tested your APIs and this thing doesn't work. This works, voice here wasn't good, the stability wasn't very clear parameter." Mati was stunned: "He was the only investor ever to do that – the only person that tested our APIs, decided it was actually valuable, gave me feedback, and then told me he's keen to invest."

A16Z demonstrated value before investing – introducing them to celebrities for voice licensing, helping for two weeks prior. This is the litmus test: "Before you accept any term sheet, that's the best time to test whether they can be helpful."

They raised £19M in June 2023. The combination of A16Z's brand and NFDG's technical credibility gave them global legitimacy and access to the best AI minds.

Key Takeaways:

• Test investor helpfulness before accepting term sheets – ask for specific introductions and see who delivers

• The best investors test your product before investing – shows genuine interest

• Speed matters in fundraising – Brian flew to London twice and closed quickly

• Combine institutional brand (A16Z) with technical credibility (NFDG) for maximum signal

😱 The Darkest Moment: When a Customer Launched Their Product First

Late 2023 brought ElevenLabs' lowest moment. They'd given their API components to enterprise customers. One customer took those components and launched a dubbing product two weeks before ElevenLabs' planned September launch. "That was one of the low moments for me, for my co-founder, for the entire team," Mati admits. "Dubbing was our story. We told that to everyone."

The customer's launch exploded. All the attention ElevenLabs had been building towards for two years went to someone else. The customer generated "tens of millions of revenue" from that product.

How do you recover? Mati's advice is counterintuitive: "I don't think the first reaction should be that everything is fine. I think you should be authentic and tell the team what you are feeling. How has this happened? Actually go through what we've done wrong." Then move quickly from acknowledgment to action. "Relentless execution will prove itself out and show it to the world."

Key Takeaways:

• Don't give away the components of your strategic product before you've launched it yourself

• Acknowledge failure authentically before pivoting to solutions – toxic positivity destroys trust

• Move quickly from acknowledgment to action – don't wallow, but don't skip emotional processing

• Relentless execution is the best revenge

🧠 Research as a Head Start, Not a Moat

Unlike many AI founders who position their models as permanent moats, Mati is refreshingly candid: "Research for us is a head start. All this gives us is advantage over the competition for the next year, two, maybe three." He estimates they're 6-12 months ahead depending on the use case.

The research team is exceptional. Mati estimates there are only 50-100 top voice AI researchers globally, and ElevenLabs has 5-10 of them. But how do you retain this talent when OpenAI, Anthropic, and Meta are offering hundreds of millions?

Mati's answer: the upside is just getting started, speed from research to production is immediate (no corporate red tape), and learning from a small, mighty team beats big company bureaucracy.

The key insight: research buys you time to build product moats. Those 6-12 months let you build integrations, create workflows customers depend on, establish brand, and lock in distribution. By the time competitors catch up on research, you've built product advantages they can't easily replicate.

Key Takeaways:

• Research advantages are temporary – plan for 6-12 months of lead time, not permanent moats

• Top researchers are scarce – there are only 50-100 globally in voice AI

• Speed from research to production is a retention tool for researchers who want impact

• Use research lead time to build product moats through integrations, workflows, and brand

📈 Hypergrowth: From £35M to £200M ARR in 18 Months

The numbers are staggering: from £35M ARR at the end of 2023 to £200M ARR in late 2025. With just 250 employees, that's nearly £1M ARR per employee.

The growth came from two parallel engines:

Engine 1: Self-Serve Creator Platform – Creators, audiobook authors, and developers using the platform for narration and voiceovers. This provides high-volume, low-touch revenue, viral distribution, and product feedback.

Engine 2: Enterprise Voice Agents – Large contracts (biggest is around £2M annually) for companies building conversational AI agents. Primary use cases: call centre automation, customer support, and personal assistance. Major customers include Cisco, Twilio, and Epic Games.

Mati believes the voice agent business is "just scratching the surface" and could become "a multi-billion dollar revenue generating business" on its own.

Key Takeaways:

• Dual revenue engines (self-serve + enterprise) provide diversification and resilience

• £1M ARR per employee is exceptional efficiency for an AI infrastructure company

• Voice agents for enterprise is the biggest future opportunity, potentially multi-billion dollar business

• 30x revenue multiple at Series B reflects market confidence in continued hypergrowth

🏢 Building Small and Mighty: The 20-Team Structure

At 250 people, ElevenLabs is organised into approximately 20 teams of 5-10 people each. "More people frequently doesn't fix the problem," Mati explains. "You don't need that many people to do something special."

But here's where it gets really interesting: ElevenLabs has no job titles. None. At 250 people. Growing to 400 by year-end.

Mati's reasoning: impact matters, not hierarchy. Small teams make titles a distraction. It enables rapid role transitions. "You can have a lower tenure and be a manager of people with much longer tenure if you are the right person."

The plan is to grow to 400 people by end of 2025. And yes, Mati and Piotr still interview every single hire. They plan to continue until 1,000 people [which seems insane!].

Key Takeaways:

• 20 teams of 5-10 people enable speed and ownership at scale

• No titles prevents artificial hierarchy and enables rapid role transitions

• Organise by product area, not function, for end-to-end ownership

• Founder interviews at scale (targeting 1,000 people) signal commitment to culture and talent density

🌍 Building from Europe, Winning Globally

"Building in Europe is building on hard mode," says Anton from Lovable. Does Mati agree? "I think it is. But there is some great advantages too."

The advantages: incredible talent that's underutilised, hunger for global ambition, and Central and Eastern European work ethic. When Harry suggests Americans work harder, Mati pushes back: "We've had a team where we hired some people from the west coast of US and our people from Central and Eastern Europe were like, 'Yeah, they don't work actually that hard as we do.'"

The key distinction: "You want to be able to build from Europe but not build only for Europe. Those two get conflated. You still want the global aspiration."

On European versus American VCs, Mati does crucial reference checks with every investor, asking: "How will they behave if it's not going well?" US investors are "a lot more keen to take the risk." When he checked with some European investors, "that check didn't yield the good results."

Key Takeaways:

• European talent is underutilised, not inferior – they lack ambitious companies to join, not work ethic

• Build from Europe, but for the world – don't conflate geography with market ambition

• US investors take bigger risks and support founders better in downturns (based on reference checks)

• Grow talent internally with external mentorship rather than importing expensive US executives

🔮 The Future: Voice as the Primary Interface

Mati's most contrarian belief: "Voice will be the interface to the technology around us. It will be the primary interface for a lot of technology around us, which I think most people would not agree with."

This isn't just about voice assistants – it's about conversational AI becoming the default way we interact with software, services, and information. ElevenLabs is positioning for this future with their voice agent platform, which includes speech-to-text and text-to-speech models, knowledge base integrations, function calling, and deployment tools.

On whether OpenAI will just do this themselves, Mati is pragmatic but confident. ElevenLabs has focus (owning voice AI specifically), scarce research talent (5-10 of the world's top 50-100 voice researchers), product depth (all the additional steps to make it perfect), and an enterprise platform that OpenAI isn't investing as much time in.

Key Takeaways:

• Voice will become the primary interface for technology – a contrarian bet most people don't believe

• Voice agents for enterprise is a multi-billion dollar opportunity just getting started

• Product depth and enterprise platform create defensibility beyond research advantages

• Focus on one domain (voice) beats generalist AI companies trying to do everything

💡 CONTRARIAN INSIGHTS (Quick Hits)

• Horizontal go-to-market can work – ElevenLabs deliberately launched to creators, developers, and enterprises simultaneously, discovering unexpected use cases whilst building multiple revenue streams

• No titles at 250 people – Zero job titles enables rapid role transitions and prevents artificial hierarchy from slowing decisions

• Research is a head start, not a moat – Their 6-12 month lead is enough to build product moats through integrations and ecosystem lock-in

• Acknowledge failures authentically – When the customer launched their product first, Mati told the team "we are angry at ourselves" before pivoting to solutions

• European talent works as hard as Silicon Valley – It's not a culture problem, it's an ambition and opportunity problem

• Founder interviews at 1,000 people – Mati and Piotr interview every hire and plan to continue, signalling commitment to culture

• Speed is the only differentiator – In AI, where models commoditise in 6-12 months, execution speed is more valuable than research quality

🔗 Links Referenced

• ElevenLabs: https://elevenlabs.io

• 20VC Podcast: https://www.20vc.com

• Full Episode: https://youtu.be/whVdDLtkiKs

• Andreessen Horowitz: https://a16z.com

• Sequoia Capital: https://www.sequoiacap.com

• NFDG (Nat Friedman & Daniel Gross): https://nfdg.com

• Credo Ventures: https://www.credoventures.com

• Concept Ventures: https://concept.vc

This article is part of the Pod Shots series, transforming podcast conversations into actionable insights for busy product leaders, tech founders, and investors. For more summaries, visit our library.

That’s a wrap.

As always, the journey doesn't end here!

Please share and let us know whether you liked this separate Pod Shot, or whether you’d rather stick with your usual programming….. 🚀👋

Alastair 🍽️.