Spotify ended the second quarter with 300 million Premium subscribers, €4.78 billion in revenue, €655 million in operating income, and a record 33.4% gross margin. On the same day, the company added Merlin to its upcoming AI covers and remixes product, bringing a major chunk of the independent music business into the project alongside Universal Music Group.
The product will launch as a paid add-on for Premium subscribers. Spotify says artists and songwriters will participate voluntarily, receive credit, and get paid when fans use their work to create licensed covers and remixes.
Spotify hasn’t disclosed the price, royalty formula, ownership structure, payment triggers, deductions, reporting requirements, or what happens when a fan-made version competes with the original.
That leaves a much bigger question than whether Spotify can sell an AI remix tool.
Spotify licenses most of the music on its service. It generally doesn’t own the songs or recordings. So when a fan turns an existing track into something new, which rights get used, who gets paid, and does the AI version earn under the same rules as the original?
Artists can’t judge the deal until Spotify answers those questions.
Spotify’s margin target explains the incentive
Spotify’s core business is already working. Subscription growth, price increases, and tighter cost control have pushed the company into sustained profitability. Management now wants gross margin to reach between 35% and 40% by 2030.
AI spending has to help get it there. Spotify expects marketing and AI-related investments to add approximately €200 million in operating expenses this year. A paid remix product gives the company a direct way to turn that spending into additional revenue from customers who already subscribe.
Spotify says it pays roughly two-thirds of its revenue to recording and publishing rights holders under its current business. Those rights holders then pay artists and songwriters according to their individual label, distribution, publishing, and collection agreements. Spotify doesn’t pay a fixed amount for every stream. Payments are calculated through streamshare.
The AI add-on doesn’t have to follow that structure.
Spotify could charge separately for generations, credits, exports, revisions, sharing, or access to certain songs. The company could also negotiate a completely different rights-holder pool for AI activity.
That’s where the margin opportunity sits.
Spotify can pay artists incremental money while retaining more of each AI dollar than it keeps from an ordinary Premium subscription. Those outcomes don’t conflict. An artist can receive a new check and still get a weak percentage of the new business.
Spotify licenses the music. It doesn’t get every right in one bundle
A commercially released song usually contains at least two separate copyrighted works.
The composition covers the music and lyrics. The sound recording covers the specific recorded performance, commonly called the master. Those rights are frequently owned, controlled, and licensed by different parties.
Spotify’s UMG announcement confirms that the AI product required separate recorded-music and music-publishing licensing agreements. Spotify needed access to both sides because a fan-made creation may use the underlying song, the existing recording, or both.
Spotify is licensing permission to build the product. It isn’t buying ownership of every participating song.
Labels, artists, publishers, songwriters, producers, estates, distributors, and collection organizations can all sit somewhere in the rights stack. The exact parties depend on the song and the contracts behind it.
That matters because saying an “artist gets paid” tells us almost nothing.
Which artist? Paid through which right? Paid by Spotify directly, or after a label, publisher, distributor, or collection society takes its share? Paid when the remix is created, or only if somebody listens to it?
The product announcement skips all of that.
“Many Men” shows how quickly the payment question gets messy
Take 50 Cent’s “Many Men (Wish Death).”
A normal stream of the original recording generates revenue on both sides of the song. Recording royalties flow to the rights holder controlling the master. Publishing royalties flow to the owners of the underlying composition.
Spotify pays those rights holders. What eventually reaches 50 Cent depends on his ownership interests and private agreements. Spotify’s ordinary royalty system doesn’t guarantee that the performing artist owns the master, controls the publishing, or receives every category of payment attached to the track.
Now let a fan create an AI country version of “Many Men.”
If the tool produces a completely new recording without using 50 Cent’s original audio, it could function more like a cover. The composition rights still matter because the fan is using the song’s melody and lyrics. The original master may not be used.
That could mean the songwriting and publishing side gets paid while the owner of the original recording receives little or nothing from the use, depending on Spotify’s negotiated terms.
The economics change again if the fan-made version uses 50 Cent’s original vocal, instrumental, beat, stems, or other parts of the master.
Now the product is using the original sound recording along with the underlying composition. The payment waterfall could include the master owner, publishers, songwriters, performers, producers, distributors, and other parties with contractual interests.
Does the remix trigger the same rights, enter the same royalty pool, and receive the same contractual treatment as the original recording?
Nobody outside the deal knows.
Spotify hasn’t said the AI version will use its normal streamshare system. It hasn’t said a generated track will carry the same royalty treatment as the original. It hasn’t even said whether the primary payment happens when the fan creates the remix or when somebody listens to it.
“Artists will be compensated” doesn’t answer the question.
A cover and a remix could produce completely different checks
Spotify keeps discussing covers and remixes together. The terms aren’t interchangeable.
A cover generally creates a new performance of an existing composition. A remix generally transforms or incorporates an existing sound recording.
That difference can determine which rights holders participate.
Artists need Spotify to publish a clear payment map:
- Does generating a cover trigger a publishing payment?
- Does using the original vocal or master trigger a recording payment?
- Does Spotify pay again when the generated track is streamed?
- Does the AI version enter the ordinary streamshare pool?
- Does the original recording receive attribution without receiving the same royalty?
- Does Spotify deduct computing, product, processing, or distribution costs first?
- Does the fan-made recording receive its own master royalty?
- Who receives that master royalty?
The final question could become the biggest one.
If the fan creates a new recording, Spotify still needs rules governing who can use, distribute, and monetize it. Copyright ownership may depend on how much human creative expression the fan contributed, while Spotify’s product terms could separately control how the output is used. Spotify hasn’t said whether that party will be the fan, the original artist, the label, the publishers, Spotify, or some combination.
It also hasn’t said whether the fan can download the track, release it elsewhere, earn money from it, use it in a video, or claim any ownership over the result.
A fully licensed product can still pay the artist badly. It can also create a valuable new recording that the artist doesn’t control.
Nobody has explained who owns the generated track
AI output creates a separate ownership problem.
The U.S. Copyright Office says AI-assisted work may receive copyright protection when a human contributes enough original creative expression. Merely entering prompts isn’t enough by itself. Human selection, arrangement, editing, performance, or modification may create protectable elements, depending on the circumstances.
That means ownership of a fan-made Spotify remix may depend on what the user actually does, how much the AI system determines, and what the product’s terms require.
Spotify could demand a broad license to host and distribute the result. A label could negotiate control over remixes derived from its masters. A publisher could retain authority over the underlying composition. The fan could potentially own limited human-created elements.
Those are possibilities, not disclosed product terms.
Spotify hasn’t published the terms.
Artists also need to know whether opting in grants Spotify permission beyond the immediate fan remix.
Can Spotify use the original recording, isolated vocals, stems, prompts, user activity, or generated outputs to train and improve future models?
Does consent to remix one song include consent to imitate the artist’s voice?
Does the agreement cover only copyright, or does it also cover the artist’s name, image, voice, and likeness?
Can Spotify use thousands of fan generations to build a better model of an artist’s sound without paying separately for the training value?
The UMG and Merlin announcements don’t answer any of those questions.
An artist might approve a fan-made remix product and still reject having their work used to improve Spotify’s AI system. Those permissions shouldn’t be bundled together inside one vague opt-in.
The artist may not control the opt-in button
Spotify and Merlin say participation will be optional. The Merlin announcement says artists on labels covered by its Spotify agreement will have the option to participate.
The announcement doesn’t explain who exercises that option.
Does Spotify ask the artist directly? Can the label approve participation under an existing recording agreement? Can a distributor enroll an independent artist? Does the publisher separately approve the composition? Can one songwriter block a remix when several writers own the song?
An artist can technically have a choice while somebody else controls the paperwork.
Withdrawal raises another set of questions.
Can an artist leave immediately? Does Spotify stop future generations while keeping existing remixes active? Do already-created tracks continue earning money? Does the fan retain access? Can Spotify keep any data gathered while the artist participated?
Artists need control that works after the launch announcement, not only when the contract gets signed.
They also need audit rights.
Spotify should show the gross amount generated by each song, every deduction, the payment allocated to each rights category, and the final amount sent to the artist’s label, distributor, publisher, or collection organization.
Without that reporting, an artist will see only the last number in a chain they can’t inspect.
Discovery needs receipts
Spotify music chief Charlie Hellman says every creation will drive listeners back to the original work. That’s a claim about how the product should function. It isn’t proof of how listeners will behave.
A fan-made remix could increase interest in the original. It could also replace time the listener would’ve spent streaming it.
Spotify earns the add-on fee either way.
Artists need data that separates those outcomes:
- How many paid generations used the artist’s work?
- How much gross revenue did those generations produce?
- How many users streamed the original afterward?
- Did listening to the original increase or decline?
- Did the generated version receive playlist or recommendation exposure?
- Can the remix compete with the original on charts?
- Does listening to the remix pull money from the same royalty pool?
- How much did Spotify keep?
Discovery is not a royalty statement.
Spotify owns the customer relationship and the behavioral data needed to answer these questions. Artists shouldn’t have to take the company’s word for it.
Merlin adds scale without adding transparency
Merlin represents members across more than 70 countries and approximately 15% of the global recorded music market. TechCrunch reports that its network brings more than 30,000 labels into the product’s potential licensing pool.
That gives Spotify enough independent catalog to test the product without waiting for the entire music industry. Spotify says an initial research preview will reach a subset of users and that the company doesn’t need a complete catalog to begin.
The strategy makes sense. Spotify can start small, measure demand, adjust pricing, and use the results to bring more rights holders into the product.
Artists will be participating in the same experiment.
Spotify will learn what fans will pay, how often they generate tracks, which artists drive the most demand, and how creation affects listening. That information will shape later licensing negotiations.
Artists need access to the same data. Otherwise, Spotify enters those negotiations knowing exactly what each catalog is worth while the artist sees only a royalty statement.
That’s a major shift in bargaining power.
The Streaming Wars Take
Spotify has already built a profitable subscription machine. The AI remix add-on gives it a second transaction around music it already distributes.
The fan pays to hear the song. Then the fan pays again to transform it.
Spotify controls the product, price, billing, interface, data, recommendations, and distribution. Rights holders supply the songs, recordings, performances, voices, and names that make the product worth buying.
That puts Spotify at the cash register while everybody else argues over the receipt.
Artists should demand the full rights map and payment waterfall before opting in. Which copyrights get used? Who owns the output? What triggers payment? What does Spotify deduct? Does the AI stream count like the original? Can the company train on the results? Can the artist leave? Who keeps the remixes after they do?
Spotify may create meaningful new artist income. It may also build a higher-margin revenue category that gives artists a smaller percentage of each transaction than they receive from conventional streaming.
Both can happen at the same time.
Until Spotify publishes the ownership rules and royalty math, “additional revenue” remains a promise without a percentage.
The Streaming Wars is intentionally ad-free
We don’t run display ads. Not because we can’t, but because we don’t believe in them.
They interrupt the reading experience. They cheapen the work. And they burn advertisers’ money on impressions nobody actually wants.
So we chose a different model.
We say the things people in this industry are already thinking but don’t say out loud. We connect the dots beyond the headline and focus on explaining why things matter to the people working in this business.
If you believe industry coverage can exist without clutter and interruption, you can support it here → SUPPORT TSW.
Support is optional. But it directly funds research and continued coverage — and helps prove this model can work.
Support TSW →
