AI voice startup ElevenLabs has had a hectic start to 2025. Since announcing its $180m Series C at a $3bn valuation in January, the company has launched its own speech-to-text model, opened an office in Japan, secured a deal with Spotify to offer audiobooks on its platform and seen a growing number of competitors entering the market.
Despite the intensity of the past few months, cofounder and CEO Mati Staniszewski appears unfazed when Sifted meets him at the Sana AI Summit in Stockholm earlier this week. Asked when he last took a holiday, he simply shrugs: “Every day is like a holiday for me.”
Founded in 2022 by alumni of Google and Palantir, ElevenLabs develops AI technology for synthetic voice generation. Its core product converts text to speech with emotional nuance and intonation, enabling clients like publishers and content creators to transform written content into audio.
Its technology has been used by Time magazine to narrate articles and by Nvidia to translate corporate videos. UK-based AI avatar startup Synthesia also uses ElevenLabs’ tech to dub videos into other languages.
But the field is becoming increasingly crowded — with both tech giants like OpenAI and Amazon launching their own text-to-voice models, and other startups such as US-based Nara Labs, maker of open-source model Dia, making waves.
“I think over the last year or two, it’s become clear that we are a leader in voice,” says Staniszewski. “We’re at the forefront of the models.”
“And suddenly, we’re in a different position, and you’re seeing a lot of competition emerging — which is exciting. But there’s definitely a growing sense of pressure, if not outright stress.”
Research-led
Like many startup founders, Staniszewski welcomes competition — and believes ElevenLabs has a key advantage: its deep focus on AI research.
“The research side is really hard. My cofounder, Piotr Dabkowski — who is a genius — has repeatedly proven we can lead in research. He’s built what I think is the best team in AI in the world. And that’s something many of these companies can’t easily replicate.”
Dabkowski, who holds degrees in both computer science and engineering from Cambridge and Oxford, was a machine learning engineer at Google before cofounding ElevenLabs three years ago.
“We feel that pressure — we have to be the best at research. We can’t afford to fall behind OpenAI or rely on tools that are 6, 12, or more months behind what we’re doing.”
Aiming for global reach
Since launching three years ago, ElevenLabs has established its international headquarters in London, with a development office in Warsaw and another base in New York. Last month, it opened its first subsidiary in Japan, marking a move into the Asia-Pacific region. According to the company, Japan was specifically picked because of its linguistic diversity, tech innovation, and the opportunity to bridge communication gaps and preserve cultural narratives.
It already has local broadcasters and technology companies as partners that will integrate ElevenLabs’ dubbing with their translation technologies to deliver tailored dubbing services to their clients.
According to Staniszewski, its multilingual focus is what sets it apart from competitors like OpenAI, which largely concentrates on English.
“They serve the English-speaking, often companion-style market. We’re looking at how to make the technology truly global — representing different voices, accents and languages. How do we make it reliable? Scalable? And bring it into real enterprise use?”
Staniszewski not only wants ElevenLabs’ tools to be used globally but also wants the company to have a physical presence in every market. But scaling a startup isn’t easy.
“It’s incredibly difficult. I’m a first-time founder — I’m still learning,” he admits.
ElevenLabs currently employs around 170 people and continues to grow.
“We try to keep our teams small. Everyone’s talking about using AI to amplify work — at ElevenLabs, everyone’s already doing that. But they’re stretched: with the number of clients, the technologies, the products we support.”
According to ElevenLabs, its technology is used by 60% of Fortune 500 companies, as well as tech firms like product demo company Arcade, AI company Perplexity, and Thoughtly’s AI call centers.
“We definitely need to grow, especially in our go-to-market function. We want to take our tech to Japan, India, the US, and not just broadly to Europe — but into individual countries, with people on the ground working directly with clients.”
“And that takes people.”
Innovation in the space
Despite rapid progress in AI, some areas remain too manual for true scale — such as dubbing, the very use case ElevenLabs was founded to solve.
“We started the company with the vision of taking content and making it available in other languages. But this is still an area where AI isn’t ready,” Staniszewski says.
“People can make edits in the cloud — they do this on our platform — but you can’t scale it yet. You can’t have every person seamlessly translate their audio into other languages without needing human intervention. But that will change within the next year; we’ll start to see automatic translations that just sound good out of the box.”
Another area poised for transformation is audiobook narration.
“Right now, most audiobooks are still single-voice narration. Some people experiment with multi-voice productions using AI, but it’s time-consuming and complex,” he says.
“But within the next year, we’ll see audiobooks created with multiple voices, background sounds and music — almost on the fly — for the first time.”
Staniszewski told Sifted last summer about ElevenLabs’ ambitions in conversational AI — including AI tutors, AI healthcare assistants, AI friends and customer support agents. He still believes this is where the most significant breakthroughs will occur.
“The biggest category is what we call agentic AI — or AI agents — and we’re finally bringing voice into that equation,” he says.
“Over the past six months, we’ve seen real use cases that sound and interact like you’d expect, without constant friction.”
At Sana AI Summit, he is about to get on stage to showcase some of ElevenLabs’ features.
“I love doing things live. I think it’s the most real experience you can have. It also means things often aren’t fully ready — they might break,” he says with a smile.
Read the orginal article: https://sifted.eu/articles/elevenlabs-mati-staniszewski-the-pressure-is-increasing/
