Analysis
Gradium, a Paris-based voice-AI startup, reopened its seed round to bring in $100 million in total funding, adding Nvidia as a new investor alongside existing backers FirstMark Capital, Eurazeo, DST Global Partners, and individual investors Eric Schmidt and Xavier Niel. The round extends financing the company first raised in December 2024, when it launched with $70 million as a spinout of Kyutai, the French AI research lab.
The company, founded by Neil Zeghidour -- a researcher with stints at Google Brain, DeepMind and Facebook -- builds voice-AI models designed around ultra-low latency, aiming to close the gap between how a human conversation flows and how AI voice interactions typically feel: laggy, turn-based and obviously synthetic. Gradium's pitch is that near-instantaneous response time is the unlock that makes AI voice products usable for real-time applications rather than novelty demos.
“Gradium's pitch is that near-instantaneous response time is the unlock that makes AI voice products usable for real-time applications rather than novelty demos.”
Renault is a notable named customer, giving Gradium a commercial reference point outside the consumer chatbot category that dominates most voice-AI headlines. Nvidia's participation as a new investor is also a signal in itself: the chipmaker has increasingly used strategic investments in AI application-layer startups to deepen ties across the ecosystem it supplies compute to, rather than purely for financial return.
Gradium is opening a Bay Area office as part of the raise, a move the company describes as "strengthening its position at the heart of the world's leading AI ecosystem." For a Paris-founded startup, establishing a genuine U.S. presence is increasingly treated as a competitive necessity rather than an option -- both to recruit engineers who might otherwise land at OpenAI, Anthropic or Google, and to stay close to the U.S. customers and investors who still set the pace for AI infrastructure spending.
The round adds to a broader pattern of European AI startups raising substantial capital while explicitly building U.S. footholds, rather than trying to compete purely from Europe. For investors, Gradium's traction with an industrial customer like Renault -- rather than purely consumer usage numbers -- is one of the more concrete signals of real enterprise demand for low-latency voice AI outside the major U.S. labs' own product lines.