Funding · 8 Jul 2026 · 2 min read

Gradium raises $100M seed to chase ElevenLabs

The Paris-based voice AI startup lands Nvidia backing, but for builders, the real story is its promise of ultra-low latency.

Pen-and-ink illustration: a small, powerful rocket ign. For the story "Gradium raises $100M seed to chase ElevenLabs".
— Pen-and-ink illustration: a small, powerful rocket ign. For the story "Gradium raises $100M seed to chase ElevenLabs". —

What happened

Gradium, a Paris-based AI voice startup, re-opened its seed round to new investors. According to TechCrunch, the round now totals $100 million. The company announced the new funding on Thursday.

Nvidia joined the round, which previously included FirstMark Capital and Eurazeo. Gradium says it will use the funds to open a Bay Area office and compete for talent. The company spun out of French AI lab Kyutai and aims to deliver ultra-low latency voice models for responsive AI agents.

How the room's reading it

The immediate chatter on X frames this as a direct challenge to ElevenLabs. With a $100 million seed, Gradium is now seen as a heavily-backed contender in the high-quality voice market. The Nvidia investment is read as a major validation — a signal that Gradium’s low-latency models have serious technical merit and a potential hardware-level optimisation path.

Some developers are pointing out that while the funding is huge, the real test is displacing an incumbent like ElevenLabs, which already has significant developer adoption. The consensus among investors seems to be that the market for voice AI is large enough for multiple winners, especially for specialised use cases where latency is everything.

Sailfish's take

We see this less as a funding story and more as a latency story. The $100 million is table stakes in this market. The real prize is building a voice model that feels truly conversational — and that comes down to cutting the awkward pause after a user stops speaking. We've built enough voice agents to know that sub-200ms response times are where the magic happens.

Nvidia's backing suggests Gradium might have a genuine architectural advantage. We're not switching any production workloads from ElevenLabs yet. But we are putting Gradium's API on our list to test for one thing only — its performance in a real-time, back-and-forth conversation. That's the benchmark that actually matters.

Our take — your read?

Be the first to weigh in.

Sources
— END OF DISPATCH — Funding