Paris-based AI voice startup Gradium has officially emerged from stealth, announcing a substantial $70 million seed funding round. The company, which spun out of the prominent French AI lab Kyutai—itself backed by French telecom billionaire Xavier Niel—aims to revolutionize AI voice technology with its ultra-low latency models.
The impressive funding round was co-led by FirstMark Capital and Eurazeo, with notable participation from Xavier Niel, DST Global Partners, billionaire Eric Schmidt, and other strategic investors. This significant capital injection positions Gradium to accelerate its development in a highly competitive field.
Pioneering Instant AI Voice Capabilities
Founded just a few months ago in September 2025 by Kyutai founding member Neil Zeghidour, Gradium is built on a foundation of deep expertise. Zeghidour previously honed his skills working with voice models as a researcher at Google DeepMind. Gradium has developed advanced audio language AI models specifically engineered to deliver voice at scale with ultra-low latency, essentially creating AI voices that respond almost instantly.
The startup's core mission is to enhance the speed and accuracy of voice models for developers. Catering to a global market from its inception, Gradium launched with robust multilingual support, offering English, French, German, Spanish, and Portuguese, with plans to integrate additional languages in the near future.
Navigating a Crowded AI Voice Landscape
Gradium enters a rapidly expanding and highly competitive market. Frontier large language model (LLM) companies such as OpenAI, Anthropic, Meta Llama, and Mistral all feature their own voice, speech recognition, and multimodal AI models. Beyond these giants, well-funded startups like ElevenLabs, alongside hundreds of other voice and speech models available on platforms like Hugging Face, ensure that developers currently have no shortage of options for AI voice capabilities.
Despite the crowded market, Gradium believes there is a growing demand for what it aims to offer: ultra-realistic voice expression and unparalleled accuracy. As AI continues its evolution from text-based chats to sophisticated AI agents, and expands into diverse applications spanning entertainment, education, and professional work, the need for advanced, natural-sounding AI voices is only expected to intensify.
