Gradium Secures $70M Seed Funding for AI Voice Technology

Gradium Emerges with $70 Million Seed Funding

Gradium, a new company originating from the French AI research facility Kyutai, publicly launched on Tuesday after securing $70 million in seed funding.

This substantial investment was spearheaded by FirstMark Capital and Eurazeo, and included contributions from prominent figures such as Xavier Niel, Eric Schmidt, and DST Global Partners.

Focus on Low-Latency AI Voice Models

Gradium specializes in the creation of audio language AI models. These models are engineered to produce voice output on a large scale, characterized by exceptionally low latency – meaning near-instantaneous responses.

The company was established recently, in September, and is led by Neil Zeghidour, a former researcher at Google DeepMind and a founding member of Kyutai.

Multilingual Capabilities from the Start

Gradium aims to enhance the speed and precision of voice models for developers.

Notably, as a European-based startup, Gradium immediately offers support for multiple languages, including English, French, German, Spanish, and Portuguese, with plans for further expansion.

A Competitive Landscape

Gradium enters a highly competitive market.

Leading large language model (LLM) companies – OpenAI, Anthropic, Meta Llama, and Mistral – already possess voice, speech recognition, and multimodal technologies.

Furthermore, well-funded startups like ElevenLabs and a vast collection of voice/speech models available on Hugging Face present significant competition.

Growing Demand for Realistic AI Voice

Despite the crowded field, the demand for Gradium’s proposed capabilities – highly realistic voice expression and accuracy – is expected to increase.

This growth will coincide with the evolution of AI from text-based interactions to AI agents and its broadening application across diverse sectors, including entertainment and professional workflows.

As AI transitions beyond typed conversations and expands into areas like AI agents, the need for sophisticated voice technology will become increasingly critical.