Paris AI Startup Gradium Raises $70M in Seed Funding
▼ Summary
– Gradium, a startup from French AI lab Kyutai, launched with a $70 million seed round from prominent investors including FirstMark Capital and Xavier Niel.
– The company develops ultra-low latency audio language AI models, aiming to provide near-instant voice responses for developers.
– It was founded in September 2025 by Neil Zeghidour, a former Google DeepMind researcher with expertise in voice models.
– The startup launched with multilingual support for five languages and plans to add more, positioning itself as a European contender.
– Gradium enters a competitive field against major AI companies and startups, but sees growing demand for realistic, accurate AI voices in expanding applications.
A new Paris-based artificial intelligence company has emerged with significant backing, aiming to transform how developers integrate realistic and responsive synthetic voices into their applications. Gradium, a startup originating from the French AI research laboratory Kyutai, has officially launched with a substantial $70 million seed funding round. The investment was spearheaded by FirstMark Capital and Eurazeo, attracting a notable roster of backers including French telecom magnate Xavier Niel, DST Global Partners, and former Google CEO Eric Schmidt.
Founded just a few months ago in September 2025 by Neil Zeghidour, a former Google DeepMind researcher and founding member of Kyutai, Gradium specializes in audio language models. The core technical focus is on achieving ultra-low latency, which means its AI-generated voices are engineered to respond with near-instantaneous speed. This addresses a critical need for developers building interactive applications where conversational flow is paramount.
The company’s ambition is to provide developers with voice models that are not only faster but also more precise. As a European venture, it has launched with built-in multilingual capabilities, supporting English, French, German, Spanish, and Portuguese from the outset, with plans to add more languages soon. This positions Gradium to serve a global market from its inception.
However, the field of AI voice synthesis is intensely competitive. Established leaders in large language models, including OpenAI, Anthropic, Meta, and Mistral, all incorporate voice and speech recognition into their multimodal offerings. Furthermore, specialized and well-funded startups like ElevenLabs exist, alongside hundreds of open-source models available on platforms such as Hugging Face. Developers currently have no shortage of options for AI voice technology.
Despite this crowded landscape, Gradium is betting on a growing market demand. The shift from simple text-based AI chats to more sophisticated, autonomous AI agents is expected to accelerate. This evolution will likely increase the need for highly realistic and accurate voice expression across diverse sectors, from entertainment and gaming to customer service and workplace productivity tools. The startup’s substantial seed funding suggests strong investor confidence in its technical approach to meeting this future demand.
(Source: TechCrunch)
