Gradium tops $100M seed funding as voice AI heats up
Gradium’s extended seed round and new strategic backing show real-time voice infrastructure becoming a major developer category.
Voice AI is moving beyond transcription into low-latency conversational systems that can power real-time agents.
What happened
Gradium raised roughly $30M in fresh funding, taking total seed financing to more than $100M.
Nvidia joined as a new investor.
The company builds real-time speech-to-text, text-to-speech, translation and edge-device voice models, alongside an open-source framework for voice-agent development.
Why it matters
The scale of the seed financing is the signal.
Voice systems need low latency, natural interaction and efficient inference if they are to become a practical interface for agents. That creates a specialised infrastructure market beyond general text models.
Strategic backing from a major AI-compute company adds another layer to that competition.
The bigger picture
Voice is becoming a foundational interface for AI applications.
The next wave of agent products may depend on dedicated speech infrastructure rather than bolting a generic voice layer onto text-first systems. Gradium’s financing reflects growing conviction in that stack.