AI Voice Startup Gradium Secures $70M Seed Funding

Published
December 02, 2025
Category
Emerging Technologies
Word Count
257 words
Voice
aria
Listen to Original Audio
0:00 / 0:00

Full Transcript

Paris-based AI voice startup Gradium has secured seventy million dollars in seed funding, launching out of stealth on Tuesday. The funding round was led by FirstMark Capital and Eurazeo, with significant participation from French telecom billionaire Xavier Niel, DST Global Partners, and billionaire Eric Schmidt, among others.

Gradium, spun out of the French AI lab Kyutai, was founded just a few months ago in September by Neil Zeghidour, who has extensive experience in voice models from his time at Google DeepMind. The startup aims to develop audio language AI models that provide ultra-low latency voice responses, essentially creating AI voices that react almost instantly.

Gradium's technology is designed to enhance the speed and accuracy of voice models available to developers. Notably, the startup has launched with multilingual support, initially offering capabilities in English, French, German, Spanish, and Portuguese, with plans to expand to additional languages.

Gradium enters a competitive landscape, contending with established companies such as OpenAI, Anthropic, Meta Llama, and Mistral, all of which have developed voice and speech recognition technologies. Additionally, startups like ElevenLabs and numerous models available on platforms like Hugging Face present further competition.

However, the demand for ultra-realistic voice expression and accuracy is expected to grow as AI transitions from text-based interactions to more sophisticated AI agents, applying its capabilities across various sectors including entertainment and professional environments.

The confidence shown by investors in this round highlights the belief that AI will significantly transform communication technologies within the telecommunications industry, marking a pivotal moment for the integration of AI in voice applications.

← Back to All Transcripts