How ElevenLabs took over AI Audio in a whisper

Welcome to the VC Breakfast Club Newsletter - the local frontpage of UK venture.

We help founders share their funding news and stitch together the UK’s venture flywheel.

If you’re announcing a raise, let us know. If you’re fundraising, get in touch.

ElevenLabs launched Eleven v3 (alpha), and it might be the closest AI has ever come to making synthetic speech feel… human.

Multiple agents can speak naturally in the same scene.

With support for 70+ languages and full emotion control. It’s a turning point for AI voice, and the most expressive model ElevenLabs has ever built.

When the company was founded in 2022 by ex-Google and Palantir engineers, the initial mission was to make dubbing actually good, starting with Polish films. The early goal was obsessive in its simplicity: make voice AI good enough that people actually want to listen to it. And it quickly evolved into something much bigger.

Fast-forward to 2025: ElevenLabs powers voiceovers across audiobooks, gaming, education, entertainment, and more. In January, they raised $180M in Series C funding at a $3.3B valuation. So what happened?

📈 Acceleration

Product-led growth: Their original text-to-speech demo (open to anyone, with uncanny realism and emotional nuance) went viral on day one. Creators loved it. So did gamers, indie filmmakers, podcasters.

Viral loops: The product was inherently shareable. People started posting their AI voices. Suddenly, your voice was a remixable asset.

APIs & dev-first mindset: The team rolled out robust APIs and voice cloning tools early on, attracting everyone from casual users to product teams. Today, you will find them embedded into language learning apps, games, and even news readers.

🏰 Moats

v3 Voice Cloning: Eleven v3 (alpha) takes voice realism to the next level with natural rhythm, breath, and tone. It is no longer "AI-sounding", it just sounds real.

Emotional Precision: Fine-tuned performance across joy, urgency, subtlety, and more.

Multi-Agent Dialogue: One prompt, multiple distinct voices and back-and-forth between speakers.

Multilingual, One Voice: One cloned voice speaks across 70+ languages with emotion and identity intact.

🔥 What ElevenLabs got right

Speed to utility: While other players tinkered with lab demos, ElevenLabs shipped a usable tool for everyday creatives.

Emotion-first: ElevenLabs focused on feel. That is what drove word of mouth.

Principled scale: As ethical concerns grew, ElevenLabs didn’t back away; instead they built in safeguards. From watermarking to a dedicated AI safety team, they took trust seriously while developing the tech further.

Simple wedge, big unlock: What started as a better dubbing tool has become infrastructure for a future where everything - apps, interfaces, entertainment - talks back.

🧠 What’s next?

With Eleven v3 (alpha), ElevenLabs has set a new benchmark for lifelike, emotionally rich AI voices. It still takes some prompting, but when it works, the result is truly impressive.

It seems to me they are only getting started.

We’re looking forward to using it more in our workflows.

Sign up over here.

🙋‍♀️ Ines

And that's a wrap! Tune in for Tuesday deep-dives & Sundays breakfast roundups.

Did we miss anything? Or just want to say hello? Hit reply - we'd love to hear from you!

• 👉 Follow the newsletter’s LinkedIn Page

• 💌 Email (Reply to this email)

If you haven’t already, make sure to sign up to get this in your inbox next week.

And if you like this, share it with a few friends!