Type “remix music tool” into a search bar right now and most of what comes back is Spotify. That is because of a Financial Times report, picked up across the trade press this week, that Spotify is targeting an early 2027 launch for a licensed AI remix and cover feature, possibly timed to Grammy week. The deals behind that tool have been covered as they landed, one press release at a time. What has not been explained is what those deals actually license, because it is not one license. It is a stack of separate, bilateral agreements, each covering a different right, each struck with a different company, and none of them disclosing what a stream of a remix actually pays.
That matters more than the launch date. A remix music tool cannot legally exist on a platform the size of Spotify without clearing two separate bundles of rights for every track it touches: the master recording, and the underlying composition. Spotify has been clearing them separately, with separate counterparties, on a track-by-track, artist-by-artist opt-in basis. Here is what is actually in each piece.
Two rights, four deals, no shared terms#
On the master side, the recording itself, Spotify has signed with Universal Music Group, with Merlin (the licensing body representing independent labels and distributors, who collectively account for roughly 15% of the global recorded-music market, per Merlin’s own figures), and with Believe, whose deal also folds in its distribution arm TuneCore. On the publishing side, the composition and lyrics, the deal so far is with Kobalt, covering songwriters independently of any of the master deals above. Believe’s own announcement describes the tool as built for remixes and covers, “not entirely new works,” meaning it functions as a licensed sampling and re-arrangement layer over existing catalogue rather than a generative composition engine.
None of the four announcements disclose a royalty rate or a revenue split. Each confirms that streams of a remix “generate royalties for the participating artists and rights holders,” in Believe’s language, and that the tool is being built as a paid add-on to Spotify Premium, reported at $4 to $5 a month on top of a standard subscription. But how that add-on fee divides between master owner, publisher, songwriter, and Spotify itself has not been stated by any of the four companies. On a claim this specific, the honest answer is that the split has not been disclosed, and nothing in the public record should be read as implying one.
What a remix music tool actually asks an artist to sign#
The consent structure is the part of this that most coverage has skipped, and it is the part that determines whether any given song can be remixed at all. Believe’s announcement lays out a two-step process: an artist first opts in generally to being considered for AI-related programs, and only after that is the Spotify tool specifically presented to them for a second, separate yes or no. Universal Music Group’s framing of its own separate deal with the remix app Hook makes the same point from a different angle: Hook “spent its first years earning licenses and building the rights framework with the industry first,” and the resulting toggle lets a rightsholder switch remixing and sound effects on or off per track, after the fact, at the catalogue owner’s discretion.
That is a materially different relationship than the one implied by “tool.” A remix music tool here is not a piece of software a fan controls. It is a permission a rights holder grants, revisable, and layered: label or distributor permission for the master, publisher permission for the composition, and platform terms from Spotify on top. An artist whose label has signed with Merlin or Believe does not automatically have their own catalogue included; the opt-in sits with the company that holds the rights, and UMG has said it is hoping around 1,000 of its artists choose to participate, a figure that is itself a target, not a count of who has agreed.
Who still is not in it, and why that is a licensing fact, not a delay#
Sony Music and Warner Music Group are both absent from Spotify’s list of remix-tool partners as of this reporting. That is not simply slower paperwork. Warner has built its own AI licensing relationship with Suno, and Spotify executives have said on an earnings call that the remix tool does not require a full catalogue to launch, which reads as acknowledgement that Warner and Sony staying out is a live possibility, not a transitional gap. A song whose master sits with Sony or Warner cannot appear in Spotify’s remix tool regardless of what the songwriter’s publisher has agreed to, because the master right and the publishing right are cleared on entirely separate tracks that do not substitute for each other.
This is also where the useful comparison lives: Hook, the social remix app UMG licensed separately from the Spotify deal, shows that “AI remix tool” is not one legal shape. Hook’s backers include Waverley Capital, led by former Warner Music chief Edgar Bronfman Jr., and Khosla Ventures; it closed a $10 million Series A in February. Its deal with UMG covers sharing remixes on social platforms and building fan communities inside the app, a different use case from Spotify’s in-platform, stream-monetised version, licensed under different terms, with its own on/off switch per track. Two companies, two products, two separate sets of rights, both using the words “AI” and “remix.” Reporting on either one as if it answers for the other is the mistake worth avoiding.
What this means for artists#
For an artist or a smaller label watching this from the outside, three things are worth holding onto. First, inclusion is not automatic: a remix music tool built this way only touches a given song if the master owner, the publisher, and the artist (where consent is passed down to them) have each separately said yes. A song can clear on the publishing side and never clear on the masters side, or the reverse. Second, the money terms are not public yet, by any of the four companies involved, so an artist should not assume a specific payout rate exists simply because a deal has been announced; “not disclosed” is the accurate status, not an oversight on anyone’s part. Third, consent here is revisable: the UMG-Hook model of a per-track toggle suggests that participation is not a one-time decision baked permanently into a contract, and artists represented by a Merlin member label or a Believe-distributed catalogue should ask their label or distributor directly whether their specific catalogue has been opted in, rather than assuming a company-wide deal covers every release automatically.
The deals that exist, UMG, Merlin, Kobalt, and Believe with TuneCore, are genuine licences, each resolving a specific bundle of rights, each sourced from the companies’ own statements. What is not yet a fact is a launch date confirmed by Spotify itself (the early 2027 timeline comes from reporting, not from Spotify), a disclosed royalty split, or a complete roster of participating catalogues. Those gaps are not filled here with a guess, because filling them with a guess is exactly the error this kind of story tends to make.
