The money trail
A distribution subscription is about thirty-five dollars a year. Delivering a song to the streaming services costs the company delivering it close to nothing. So the question worth asking is not what the price is. It is what the price is for, and who receives it.
Everything on this page comes from company filings, regulator decisions, court records and company announcements. Where a figure is press-reported rather than confirmed by the parties, it says so.
Why the price is what it is
The data format that releases travel in is free to implement; the standards body charges nothing for the licence to use it. Storing an audio file and keeping a link alive costs cents a year at commodity prices. Several distributors leave catalogues online indefinitely for people who have stopped paying, and advertise that as a feature, which tells you what that costs them.
For a few years one of these companies was listed on a stock exchange and had to publish audited accounts, broken out by division. Believe, which owns TuneCore, reported its self-serve arm separately, and those numbers answer the question directly.
Read the first two together and you have the answer to what the price is for. Less than half the money that division earned came from the yearly fee people think they are paying. A quarter came from percentages taken on social platforms and video, and most of the rest from added services like publishing administration and marketing tools. The subscription buys the customer. The customer is then sold the things that carry a commission.
One more line from those accounts is worth sitting with. The self-serve division handled 36.5% of all the music sales the group processed in 2024, while producing 6.5% of its revenue, because on the paid tiers the streaming royalties pass straight through to the artist and the company books only the fee. Enormous volume, small revenue, and a division-level margin around twenty per cent. That is a very good business, and none of it depends on charging more for the pipe.
That window has now closed. Believe was taken private by the investment firms EQT and TCV with its founder and left the Paris exchange in July 2025, so no further accounts will be published. It is the last look anyone gets. Worth noting who did not do well out of it: the shares floated at €19.50 in June 2021, fell about 18% on the first day, bottomed near €8.14 in September 2022, and were bought out at €17.20. No dividend was ever paid. Someone who backed it at the float and held to the end lost about 12%.
Who owns the biggest one
Each step below is from a company announcement or a filing with the United States Securities and Exchange Commission.
CVC is itself a listed company on the Amsterdam exchange, managing about €212 billion, and the fund making the purchase closed at €26 billion. So the chain from an artist's card to the end of the line now runs: the artist, the distributor, a buyout fund, and that fund's own public shareholders.
DistroKid publishes nothing about its finances; its own regulatory filing records revenue as "decline to disclose". But the private credit funds that lent the money for the 2021 purchase must report their holdings every quarter. Eight of them disclosed positions totalling at least US$89.6 million at 30 June 2026, and the loan was repriced during 2026 from a rate of about 9.2% down to about 7.9%. The full size of the facility is not disclosed anywhere, so that figure is a floor rather than a total.
Three of those lenders also bought equity alongside the loan. They mark it in their filings at 1.24 times cost in December 2025 and 1.35 times by June 2026, which is the only independent, dated valuation of the company anywhere in public.
None of that is unusual for a business of this kind, and none of it is hidden: it is in filings anyone can read. It is simply not what most people picture when they buy a thirty-five dollar subscription to put a song on Spotify.
What is actually being bought
When Universal Music Group moved to buy Downtown Music Holdings, the parent of CD Baby, the European Commission opened a full investigation. Its objection was not that Universal would control too much distribution. It was that Universal would gain access to competitors' royalty accounting data through a platform called Curve.
The remedy was the complete divestment of Curve: its customers, its data, its people, its source code and its hardware, supervised by a monitoring trustee. European Commission conditional clearance, 13 February 2026. The acquisition completed a week later; Curve was sold to the independent licensing body Merlin and a co-investor, completing 17 August 2026.
Then look at what the major companies did next to that. Warner signed in April 2026 to buy Revelator, another royalty accounting and rights management platform. Meanwhile both Warner and Universal closed their own self-serve services for everyday artists: Warner's shut at the end of July 2025, Universal's went invite-only and removed its DIY artists back in 2022.
They stopped competing for subscriptions and started buying the accounting layer underneath. The pipe is not the prize. The record of who is owed what is the prize.
One smaller mechanism points the same way. DistroKid runs a free, opt-in service called Upstream that puts artist profiles, streaming statistics and contact details in front of record labels. The named launch partner is Republic Records, which belongs to Universal.
Ownership, plainly
CD Baby, FUGA, Songtrust, AdRev and Soundrop, all acquired with Downtown in February 2026. Plus Ingrooves, Virgin and [PIAS]. Its own self-serve service, Spinnup, is closed.
The Orchard, bought outright in stages by 2015. AWAL, bought in 2021 for about US$430 million after a competition review. A minority stake in LANDR.
ADA, which absorbed the African and Middle Eastern services in 2026. A minority stake in LANDR. Signed to buy the royalty platform Revelator in April 2026. Bid for Believe in 2024 and walked away.
DistroKid (CVC, with Insight retaining a stake). TuneCore through Believe (EQT and TCV). Too Lost, which bought Australia's G.Y.R.O. in July 2026 (GoldState and TA Associates hold minorities). Amuse (Bridford Group).
ONErpm, which has taken no outside capital in sixteen years. Ditto Music, never sold. EMPIRE. RouteNote, bootstrapped. Symphonic remains founder-led with minority investors.
MGM Distribution (Sydney, since 1998), Rocket (Melbourne), Community Music through UNIFIED, Xelon, Musible and Ordior. G.Y.R.O. was Australian-owned until July 2026.
Ownership is not a verdict on service. A founder-owned company can treat artists poorly and a fund-owned one can treat them well. It matters because it tells you who the company answers to when those two things pull in different directions.
On the record
In a New York case decided in March 2026, an artist's claims against DistroKid for breach of contract and for a proper accounting survived an attempt to have them dismissed, with the court treating the artist as an intended beneficiary of the promise to pass through all royalties. A negligence claim was dismissed. The case had not been decided on its merits at the time of writing.
In a separate case in February 2025, DistroKid successfully defended a copyright claim by arguing that it is an automated system carrying out user commands, without the volition needed to infringe. That defence is reasonable, and it is also worth reading twice: the service is a set of pipes when that helps, and a curator when that helps.
Universal, ABKCO and Concord sued Believe and TuneCore in November 2024, alleging large-scale infringement through fraudulent artist accounts using deliberately misspelled versions of famous names, and sought at least US$500 million. The parties settled and the case was dismissed in April 2026, on undisclosed terms.
Where more than 90% of a track's monthly plays are judged artificial, Spotify does not pay for them and charges the distributor about ten dollars per track. Distributors commonly pass that charge on, with a handling fee added. An artist whose song is targeted by a bot farm they did not hire can end up paying for it.
What Australian law already says
Since 9 November 2023, using an unfair term in a standard form contract has been against the law in Australia, not merely unenforceable. Standard form means a pre-written contract the customer cannot negotiate, which describes every distribution sign-up. The protection covers consumers and small businesses under $10 million turnover or a hundred staff, which describes almost every musician.
A term is unfair when all three of these are true, as the competition regulator sets them out:
The rights of the two sides are lopsided. A term letting one side, and only one side, end the arrangement is on the regulator's own list of examples.
The business has to show the term protects something real, and the burden of showing that sits with the business, not the customer.
Financial harm, or harm of another kind, to the person on the other side of it.
Nobody has tested a music distribution contract against that. No regulator anywhere has taken action against a distributor over any of this, and none of what follows is a finding that any company has broken the law. It is simply the test, and the facts that would matter under it.
The middle limb asks what legitimate interest is protected by removing music when someone stops paying. The obvious answer would be the cost of keeping it there. Storing a ten-track album and delivering it to a hundred and fifty stores costs, at published commodity cloud prices, well under a dollar a year. The data format that releases travel in is free to implement. No streaming service publishes a fee for keeping a release live.
And there is a second fact that speaks to the same limb: the same company that removes the music also sells a product that switches the removal off, for a one-time charge per release. If the removal can be turned off by paying, it was never a cost problem.
One distributor's own help pages state that takedowns cannot be reversed, that stores treat them as permanent, and that putting the music back requires a brand new release with a new barcode. A new barcode means a new release page: new release date, new link, and the loss of whatever position the old one held. Play counts can survive, but only where the audio and details are identical, under a specific service feature designed for something else. And a catalogue that restarts from zero faces the thousand-play threshold again before it earns anything.
New unfair trading rules take effect that day, and three of them land squarely here: businesses must set out before you sign up how to cancel; they must send renewal reminders at least every six months; and if you signed up online, you must be able to cancel online, through something "easy to find, simple" and requiring only necessary steps.
Weigh that against what is on the record now: one service states plainly that auto-renewal cannot be switched off on its unlimited plans, and that during the grace period the renewal is added to your cart and "is not removable". Another says only that it gives "ample notice" before removing music, without saying how much.
The regulator has also said where it is looking. Its stated priorities for 2026 and 2027 lead with manipulative practices in digital markets, and the Chair named the target: "subscription traps and other dark patterns that manipulate consumer behaviour and unfairly impact consumer choice."
The context nobody puts in the pitch
Put those together. The independent share of the market went backwards in 2025, and the reason given is the rule that pays nothing for tracks under a thousand plays a year. Meanwhile the share of the money earned by music made with AI is one tenth of one per cent, and that is where a great deal of the industry's attention has gone.
An artist losing income to a payout threshold is not losing it to AI. That is worth saying plainly, because the two get argued as though they were the same fight.
So what do you do with this
Not "leave your distributor". Distribution is a real service and for an artist with an audience it is worth the fee several times over. The point of gathering this is narrower and more useful: know what you are buying, know what happens when you stop, and keep a door of your own that nobody can close.
Compare the terms on the Terms Table. Get the answers you are owed from the Rights Engine. And when enough makers want the same four or five changes, there is a lawful way to ask for them together, which is the Bargaining Table.
Corrections to anything on this page are welcome and get published. Every figure here has a source behind it, and any company named is welcome to put its own numbers on the record.