Nvidia Makes the Chips Behind AI Music. Now It's Being Sued for Allegedly Stealing the Training Data Too.

What does Jamendo's lawsuit against Nvidia actually say?

Jamendo, the independent and royalty-free music platform owned by Winamp Group, filed a new complaint against Nvidia on October 13, 2026, in the U.S. District Court for the Northern District of California. The suit alleges Nvidia copied hundreds of thousands of audio files, along with their associated metadata, directly from Jamendo's catalog without permission. Jamendo says the material was used to train two of Nvidia's own AI audio systems: Fugatto, which generates audio, and Audio Flamingo, a model built to understand and describe sound. Jamendo is represented by Ropers Majeski.

This isn't Jamendo's first swing at Nvidia. The platform had already brought a similar claim against the chipmaker earlier this year, then voluntarily dismissed it, only to come back with this new filing. Jamendo hasn't said publicly why the first suit was dropped, but refiling signals the dispute over how Nvidia sourced its training data is far from settled.

Why does it matter that the defendant is a chipmaker, not a music AI startup?

Every AI-music training lawsuit that has made headlines so far has gone after companies built specifically to generate music: Suno, Udio, the startups that turned prompts into finished tracks and got sued by the major labels for it. Nvidia is different. It's the infrastructure company, the one that sells the GPUs everyone else trains their models on. Fugatto and Audio Flamingo are Nvidia's own research systems, not products it sells directly to consumers the way Suno does, but the fact that Nvidia built and trained them at all puts a hardware giant squarely in the same legal fight as the AI-music startups it supplies chips to.

That's the real shift here. If scraping claims can reach a company whose core business is silicon, not songs, it suggests independent rights holders see no meaningful line between "we built the AI music app" and "we built the tools the AI music app runs on." For a platform like Jamendo, built on licensing music from independent artists rather than majors, this is also a case about an indie-first platform asserting the same kind of leverage the major labels have used against Suno and Udio.

What happens next?

Nvidia hasn't publicly detailed how it sourced the training data for Fugatto or Audio Flamingo, and the complaint is freshly filed, so there's no ruling or settlement yet. What's clear is that Jamendo chose to come back after dismissing its earlier suit rather than let the dispute drop, which suggests it believes it has stronger grounds, or more leverage, the second time around.

Why it matters

The target isn't a music-AI startup, it's a hardware company training its own audio models, which widens the pool of who can be sued over scraped music and pulls independent platforms, not just the majors, into the fight over AI training data.

What we think

Nvidia selling the GPUs that power AI music generation is one thing. Nvidia training its own audio AI on an independent artist platform's catalog without a license is a different problem entirely, and it's the kind of case that could reshape how every infrastructure company treats the data behind the models it builds in-house.