Paris-based AI voice startup Gradium has emerged from stealth with a $70 million seed round, marking a strong entry into the competitive AI voice space. The funding was led by FirstMark Capital and Eurazeo, with significant participation from French telecom billionaire Xavier Niel, DST Global Partners, billionaire Eric Schmidt, and other prominent investors. Gradium was founded in September 2025 by Neil Zeghidour, a founding member of Kyutai, a French AI research lab backed by Niel, and a former researcher at Google DeepMind, where he worked extensively on voice and audio models. The company specializes in developing audio language AI models designed to deliver real-time, low-latency voice responses—enabling AI voices that respond almost instantaneously. This focus on speed and responsiveness aims to solve a key bottleneck in interactive AI systems, particularly as applications shift from text-based chat to dynamic, voice-driven AI agents. Gradium launched with multilingual capabilities, supporting English, French, German, Spanish, and Portuguese, with plans to expand to additional languages. This multilingual foundation gives the startup a strategic edge, especially as AI adoption grows across Europe and global markets. Despite a crowded field—ranging from tech giants like OpenAI, Anthropic, Meta, and Mistral, which all offer advanced speech and multimodal models, to well-funded startups like ElevenLabs and a vast ecosystem of open-source models on Hugging Face—Gradium is targeting a niche: ultra-realistic, expressive, and highly accurate voice generation with minimal delay. The demand for such capabilities is expected to surge as AI agents become more integrated into daily life, from customer service and entertainment to productivity tools and virtual assistants. Gradium’s vision is to empower developers with tools that not only deliver lifelike speech but also maintain high performance and scalability. With strong backing and a clear focus on speed and quality, Gradium is positioning itself as a key player in the next phase of AI voice evolution—where responsiveness and naturalness are just as critical as intelligence.