Spotify and Kobalt today announced a licensing agreement for Spotify's forthcoming fan-made covers and remixes tool, extending the arrangement to songwriters represented by the world's largest independent music publisher and marking the first time the publishing side of the copyright has been named in the sequence.

The announcement was published on August 13, 2026, nine days after a comparable agreement with Merlin, the digital licensing body for independent labels and distributors. According to Spotify, the tool will allow fans to create covers and remixes of songs from participating songwriters, and it will arrive as a paid add-on for Spotify Premium subscribers rather than as a feature of the free, ad-supported tier.

That distinction between songwriters and recording artists is the substance of the news. Covers and remixes touch two separate copyrights: the sound recording, controlled by labels and distributors, and the underlying musical composition, controlled by publishers and the songwriters they represent. Spotify's earlier agreements in this sequence sat on the recording side. This one does not.

Two copyrights, two sets of counterparties

A remix of a commercial track reproduces a master recording. A cover does not - it reproduces the composition, performed anew. A tool that permits both therefore needs clearance from both categories of rights holder, and a compensation route back to each.

Charlie Hellman, SVP and Global Head of Music at Spotify, framed the point directly in the announcement. "Covers and remixes begin with the song, which makes music publishers and songwriters essential partners in building this tool the right way," he said. "This agreement with Kobalt is grounded in consent, credit, and compensation for the songwriters who take part, and every creation drives listeners back to the original song. We look forward to welcoming more partners as we continue to build."

The wording tracks closely against what Hellman said on August 4, when Spotify extended the same model to Merlin's membership, an organisation whose independent labels and distributors account for 15% of the global recorded music market. On that occasion the conditions were attached to "participating artists." Today the same conditions attach to "the songwriters who take part."

Consent, credit and compensation appears in both statements as a named precondition rather than as a downstream effect, and both statements carry the requirement that derivative creations route listeners back toward the source. What changes is who holds the consent. On the Merlin agreement, participation was elective at the artist level, made available through labels operating under Merlin's existing Spotify agreement. Here, the qualifying phrase throughout is "participating songwriters," which places the election with the writer rather than with the performer or the label.

Neither announcement explains how a single recording with several credited writers and several credited performers is treated when those parties disagree.

Kobalt frames the deal around training

Laurent Hubert, CEO of Kobalt, supplied the second quoted statement, and it introduces a term absent from the Merlin announcement.

"Fans have always sought deeper engagement with the songs they love, whether through listening or creating their own covers and remixes," Hubert said. "Spotify's new AI-powered creative tools are built on a foundation of ethical training, fair compensation, and intentional guardrails. This agreement is designed to ensure our songwriters can connect with their audience in unprecedented ways, while directly benefiting from the creative opportunities these new features provide."

Two elements of that statement are worth separating from the marketing register in which they arrive. The first is the phrase "AI-powered creative tools," which is the most explicit characterisation of the product yet offered by either party to any of these agreements. The August 4 announcement described a covers and remixing tool without attaching the technology label. The second is "ethical training." Training is a separate act from generation, and it is the act that has attracted litigation.

That timing is not incidental. On July 31, 2026, the 42nd Civil Chamber of the Munich I Regional Court ruled against Suno in a case brought by the German collecting society GEMA, setting a penalty of up to 250,000 euros for each future breach. The chamber found that model weights reproduce the works used to train them, applied United States copyright law to the training question and rejected fair use, and declined to refer the matter to the Court of Justice of the European Union. GEMA represents composers, lyricists and music publishers. The rights at issue in Munich were publishing rights.

Thirteen days later, a publisher is quoted describing a training process as ethical. The two events are not connected in either document, and no such connection is claimed here. The proximity does, however, mark where the commercial argument now sits.

What Kobalt brings

Spotify described Kobalt in the announcement as the world's largest independent music publisher. No catalogue figures, songwriter counts or territorial detail appear in the release, and the company did not state which portion of Kobalt's represented repertoire will be available inside the tool.

The announcement also gives no indication of whether participation runs through Kobalt's administration agreements automatically or requires a separate election by each writer. The release refers throughout to participating songwriters, which implies an election exists, without describing the mechanism by which it is exercised.

Spotify's newsroom lists a separate, earlier item recording a direct licensing agreement between Kobalt and Spotify. No date accompanies that listing in the material reviewed. The practical reading is that today's agreement layers a specific derivative-works permission on top of an existing commercial relationship rather than establishing one.

The sequence, and what it now covers

Spotify has been assembling this framework in public for roughly eighteen months, and the counterparties have been added in a discernible order.

In January 2025, Spotify and Universal Music Group set out a licensing arrangement covering both recorded music and publishing. That deal introduced a direct licence with Universal Music Publishing Group across multiple territories and represented Spotify's first direct publishing agreement since the Music Modernization Act of 2018. Research cited at the time suggested roughly 20% of existing subscribers had expressed willingness to pay extra for enhanced features, which is the thesis a paid add-on now tests.

At its Investor Day on May 21, 2026, Spotify reached an agreement with Universal Music Group specifically on AI covers and remixes. The Merlin agreement followed on August 4, disclosed alongside second quarter results. Spotify's own newsroom additionally lists a partnership item naming Sony Music Group, Universal Music Group, Warner Music Group, Merlin and Believe as collaborators on artist-first AI music products.

Kobalt is the first publisher named on its own terms in the covers and remixes sequence.

Co-CEO Gustav Söderström told analysts on the August 4 earnings call that the model rests on consent, credit and compensation, and that a research preview will precede any launch. Nothing in today's announcement moves that position. There is still no launch date, no price, no stated territorial scope and no disclosed split between Spotify, publishers, songwriters and the listeners who make the derivative works.

Where the money is supposed to come from

The commercial location of this product is subscription revenue. Spotify states that the tool will be a paid add-on for Premium users and that it will create an additional source of income for participating songwriters. That places it alongside pricing rather than inside the advertising business.

The scale of the two sides is not comparable. In the second quarter of 2026, Spotify reported 4,331 million euros in Premium revenue against 446 million euros from the ad-supported segment, with Premium subscribers crossing 300 million for the first time and monthly active users reaching 777 million. Advertising accounted for roughly 9% of total revenue in that quarter. An add-on priced against 300 million subscribers has a larger addressable base than the entire ad-supported business.

Spotify has also demonstrated that add-on pricing works at scale on the same base. Audiobooks+ passed 100 million euros in annual recurring revenue and one and a half million subscribers by the second quarter, according to the company's own disclosure. A covers and remixes add-on is a structurally similar product: an incremental charge layered onto an existing subscription, sold to a subset of the paying base.

Why the marketing community has a stake

Three mechanisms connect a subscription product to media buying, and none of them is stated in the announcement.

Attribution infrastructure is the first. A system that credits participating songwriters on every derivative creation requires per-creation metadata linking output to source composition, which is a harder problem than linking output to a master recording. Publishing splits are frequently fractional, disputed and administered across multiple societies. Building royalty accounting for user-generated derivatives at platform scale means building composition-level provenance data, and provenance data built for royalties is reusable for classification. Spotify adopted the DDEX standard for AI disclosure in music credits in September 2025 and, in June 2026, launched a Verified by Spotify badge for podcast shows while reaffirming its ban on AI voice-cloning impersonation.

Rights-cleared derivative audio is the second. Brands and agencies commissioning music-adjacent creative have worked with limited access to legally clean derivative versions of commercial recordings, and the composition side has been the harder half of that clearance. A licensed catalogue of covers built with songwriter-level consent recorded at creation time would constitute a supply of derivative works that does not currently exist in structured form. Nothing in the announcement addresses commercial or synchronisation use, and no such rights are claimed. The point is architectural, not immediate.

Time spent is the third. Creation surfaces generate longer sessions than passive listening, and session length feeds the inventory available to the free tier. Whether remix outputs surface anywhere on the ad-supported side has not been stated, and the add-on sits behind a Premium paywall, so any inventory effect would be indirect at best.

The wider context is a Spotify advertising business that has been rebuilt on a separate track. At its Investor Day, the company described a platform organised around two engines, high-impact sponsorships and scaled biddable channels. By the second quarter of 2026, automated channels had reached nearly 40% of ad-supported revenue, active advertisers had grown 60% year over year to 33,000, and 7,000 of those advertisers were using the company's AI audio asset creation tool.

That last figure is the quieter parallel. Spotify is licensing generative creation on the consumer side while shipping generative creation on the buy side, and the compensation architecture announced today applies only to the former.

The unresolved questions attached to this class of agreement have already surfaced elsewhere in the industry.

Universal Music Group settled its copyright infringement litigation with Udio on October 29, 2025, pairing a compensatory payment with new recorded music and publishing licences and a walled-garden platform scheduled for 2026. Udio disabled downloads and implemented fingerprinting and filtering during the transition.

Within days, a coalition of artists demanded transparency on the terms of that arrangement. The coalition raised two points that bear directly on today's announcement. The first concerns multiple credited contributors on a single work and whether existing agreements authorise the rights holder to license that work into an AI system at all. The second concerns payout mechanics: whether money from such deals reaches creators as new revenue or is applied first against unrecouped advances.

Kobalt operates as an administrative publisher, a model in which the writer retains copyright and the publisher collects. That structure sits differently against the recoupment question than a traditional publishing deal does. The announcement does not address it, and no payout mechanism is described.

What the announcement does not say

The gaps are the same gaps that were present nine days ago, with one addition specific to the publishing side.

No launch date is given. No price is given, and no indication of how the add-on will be priced against the existing Premium tier. No revenue split appears between Spotify, Kobalt, participating songwriters and the users who create the derivative works. Territorial scope is undefined. Whether outputs can leave the platform, and whether other listeners can discover them, is not stated, which matters because listener routing back to the original song is the mechanism Hellman named as the justification for the model.

The addition is this: a cover or remix implicates both the composition and, in most cases, a master recording. An agreement covering Kobalt's songwriters does not by itself clear the recordings on which those compositions appear, and a Merlin or Universal agreement covering recordings does not by itself clear the compositions. Which songs actually become available inside the tool will be determined by the intersection of those permissions, song by song, and no party has published a map of that intersection.

Hellman's reference to welcoming further partners indicates the list is incomplete.

Timeline

Summary

Who: Spotify and Kobalt, described in the announcement as the world's largest independent music publisher. Charlie Hellman, SVP and Global Head of Music at Spotify, and Laurent Hubert, CEO of Kobalt, are the named speakers. The agreement affects songwriters represented by Kobalt who elect to participate.

What: A licensing agreement covering Spotify's upcoming tool for fan-made covers and remixes. The tool will be offered as a paid add-on for Spotify Premium subscribers and is stated to create an additional source of income for participating songwriters, with consent, credit, compensation and routing back to the original song named as conditions. It is the first agreement in the sequence to name the publishing side of the copyright.

When: The announcement was published on August 13, 2026, nine days after the equivalent agreement with Merlin on August 4. No launch date for the tool has been disclosed, and Spotify has said a research preview will precede any launch.

Where: Neither party disclosed territorial scope for the add-on. Spotify reported 777 million monthly active users and 300 million Premium subscribers in the second quarter of 2026.

Why: Covers and remixes reproduce compositions as well as recordings, so a tool permitting both requires clearance from publishers in addition to labels. The agreement arrives thirteen days after a Munich court quantified per-breach liability for training a music generator on works represented by a publishing collecting society, sharpening the commercial distinction between negotiated licences and litigation exposure, and it extends a compensation architecture to a class of rights holder that user-generated covers have historically bypassed.