Wall Street turned hit songs into a steady income asset.
NurPhoto via Getty Images
In March, a 54-year-old man from Cornelius, North Carolina, pleaded guilty in a Manhattan federal court to a scheme that had drained money from one of finance’s newer income assets. Michael Smith, prosecutors said, used artificial intelligence to churn out hundreds of thousands of songs, then deployed an army of bots to stream them billions of times, siphoning more than $8 million in royalties over seven years. He agreed to forfeit $8,091,843.64 and faces sentencing on July 29. The case is the first of its kind in the United States, and for anyone who has bought into the booming business of music royalties, it is a warning shot.
How A Song Became An Income Stream
For most of music’s history, a song was a song and only that. Over the past decade it became a bond. Investment firms worked out that the royalties a hit throws off, every time it is streamed, synced to an advert or played on the radio, behave rather like the coupon on a fixed-income security: small, steady and uncorrelated with the stock market. They packaged them accordingly. Blackstone-backed Hipgnosis sold a $1.47 billion bond secured on a catalogue of songs in 2024. Primary Wave closed a music fund at $2.225 billion in April. More than $8 billion of music-backed bonds have been issued since 2020, and the funds buying catalogues run to many billions more.
The buying spree was not a subtle one. The Hipgnosis catalogue that now backs the bond holds the rights to songs by the Red Hot Chili Peppers, Neil Young, Journey, Shakira and Bon Jovi, among roughly 45,000 titles. Blackstone took the listed Hipgnosis Songs Fund private in 2024 for about $1.6 billion, and a ratings agency later valued the catalogue at $2.36 billion. Sony, Universal and a parade of private-equity firms have chased the same kind of rights, on the same logic: a proven hit earns a stream of cash for decades, and a stream of cash can be turned into a security and sold. The whole edifice assumes the cash keeps flowing at a predictable clip.
The pitch suits an uncertain year: an income that keeps paying whether shares rise or fall, because people stream Fleetwood Mac in a boom and a bust alike. Once the preserve of pension funds and specialist managers, this income is now within reach of ordinary investors through funds and listed vehicles. It rests on a single assumption that the past few months have called into question, that the royalty cheque behind the bond will keep arriving at roughly the size everyone expects.
Few people set out to buy a slice of someone’s back catalogue, so it is worth saying how a saver ends up exposed. Pension funds and insurers, hunting for income that does not lurch with equities, have parked client money in royalty strategies for years. Some of the bonds built on song catalogues carry investment-grade ratings, which is what lets conservative institutions hold them; the Hipgnosis-backed Lyra bond was rated A-, a notch comfortably inside the investment-grade band. A growing roster of funds now markets royalty exposure to retail buyers directly. The typical investor has not knowingly bet on a Shakira catalogue; the bet may simply sit, unremarked, inside a pension pot or an income fund chosen for its supposed steadiness.
The One Number That Holds It Together
To see the crack, you have to understand how streaming actually pays, and it is not the way most people assume. Spotify, Apple Music and the rest do not hand over a fixed sum every time your song is played. Instead they take the month’s subscription and advertising money, drop it into a single pot, and divide that pot across all the streams that happened. Your slice depends on your share of the total. The Department of Justice spelled it out in the Smith case: royalties are “made proportionately to musicians and songwriters from a pool of funds.”
The pool is broadly fixed in any given month, and that is the fact the whole investment case hangs on. If the number of streams chasing the pool goes up, each stream is worth a little less. A catalogue can play exactly as well as it did last year, attract the same listeners, and still earn less, because more songs turned up to split the same money. What these funds own is, at bottom, a share of a pie, and the pie is being cut into more pieces.
The Flood
Most of the extra pieces now come from a machine. Deezer, a streaming service that has been unusually open about the problem, reported in April that AI-generated tracks make up 44% of all the music uploaded to it each day. That is roughly 75,000 synthetic songs a day, up from about 10,000 a day a year earlier, a sevenfold jump in twelve months with no sign of slowing.
The honest version of this story is more interesting than the alarmist one. Those 75,000 daily uploads are not yet heard at anything like the same scale. Deezer puts actual consumption of AI music at between 1% and 3% of total streams, and says 85% of the streams that AI tracks do get are flagged as fraudulent and stripped out of royalty payments. The pool is not collapsing today. What is building is supply pressure on a fixed pot, alongside a platform’s own admission that most of this material exists not to be enjoyed but to skim. Deezer released a free tool on June 11 letting listeners scan playlists across 20 services, including Spotify, for synthetic tracks.
When The Skimming Turns Criminal
Smith’s case shows what the deliberate end of this looks like. He did not chase one viral hit, knowing that would trip the platforms’ fraud alarms. He spread his fake streams thinly across a vast catalogue of AI songs, at one point running as many as 10,000 bot accounts and generating, by his own estimate, 661,440 streams a day. A billion fake plays on a single track screams fraud. A billion fake plays dribbled across tens of thousands of forgettable tracks looks, to a detection system, almost like ordinary listening.
The Mechanical Licensing Collective, the body that distributes streaming royalties to American songwriters, flagged the anomaly early and challenged Smith’s representatives, work it says prevented the diversion of those particular royalties. The system worked. Its own message afterwards was wary rather than triumphant: it pledged to keep investing in fraud detection because the threat is not going away. Every dollar a bot pulls from the pool is a dollar that does not reach a genuine catalogue, including the genuine catalogues sitting inside investors’ funds.
Why This Lands On Your Statement
A music fund’s value is the present worth of the royalties its catalogue will earn in years to come, and those royalties are a share of a pool. If AI floods the pool with supply, lawful or fraudulent, the share attached to any one human catalogue is squeezed. The quality these assets were sold on, dependable and predictable cash flow, is the quality a supply shock to the royalty pool erodes. The risk is not bolted on from outside; it sits in how the asset earns. It also turns the funds’ fortunes on a contest most buyers cannot see: whether the platforms’ fraud detection can keep removing fake streams faster than the machines can manufacture them.
A further question hangs over the whole arrangement, and the regulators have not answered it. American royalty rates are set through a government process called the Copyright Royalty Board, and it is not yet settled whether AI-generated tracks are entitled to the compulsory licence that human-made songs draw on. If the machines diluting the pool turn out to have a contested claim on it, the rules governing who gets paid, and how much, could shift under the funds that have already bought in. For an asset marketed on its stability, that is a lot of unsettled ground.
What A Sensible Income-Seeker Checks
Music royalties are not a bad investment for any of this. The income is genuine, and its independence from the stock market is genuinely useful. The headline yield is simply the beginning of the homework rather than the end. Ask whether the catalogue a fund holds is human and verifiable, or padded with synthetic filler. Ask how the manager models future royalties, and whether it has built any allowance for pool dilution into those forecasts, or simply assumed last year’s per-stream rate will hold. Ask whether the platforms’ defences, Deezer’s detector, Spotify’s fraud measures, are keeping pace with the flood, because the value of the holding depends on machines on the other side winning a race most investors never knew was being run.
These royalties are a share of a pool that anyone can now pour into at almost no cost. Wall Street turned songs into bonds on the belief that a great catalogue’s slice of that pool was safe, and the slice was never fixed nor the pool guarded. Before buying the income, it is worth counting the other hands now reaching into the same pot.
