Jamendo, a Luxembourg-based online music platform, has filed a federal lawsuit in California against Nvidia, seeking at least $20.3 million over the alleged unauthorized use of its music database to train AI models. The complaint accuses Nvidia of using a 56,000-track dataset, created by Jamendo and the Music Technology Group at Barcelona’s Universitat Pompeu Fabra, to train its Fugatto and Audio Flamingo products despite restrictions limiting the data to non-commercial academic use. The U.S. lawsuit follows a related case in Belgium, where a judge in Ghent recently rejected Nvidia Belgium’s jurisdictional objections and allowed the matter to proceed.
The litigation highlights the growing legal risks facing technology companies that use publicly available datasets for commercial AI development. Jamendo claims it attempted to negotiate a commercial license with Nvidia from March 2024 through June 2025 before issuing an $18.3 million invoice, calculated at $329 per track for an international license. According to the complaint, Nvidia failed to pay the invoiced amount and continued to retain the benefits of the alleged unauthorized use.
The lawsuit brings claims for copyright infringement, breach of contract, unjust enrichment, and unfair competition. Jamendo alleges that Nvidia avoided standard licensing costs to accelerate its AI development, stating in the suit that “by avoiding the costs of licensing or independently developing a comparable dataset, Nvidia has reduced its research and development expenses.” The plaintiff is seeking the full invoice amount plus interest.
