Taylor Swift started her career in country music. She later moved heavily into pop with 1989, then shifted again toward folk and alternative sounds with folklore, a progression that the Recording Academy has documented across her career.
There is no reason to think Swift made these changes as an investment strategy. But look at her catalogue as a collection of assets and something interesting happens. She now owns music that reaches different listeners, playlists and genres instead of relying on one sound forever.
That is the basic idea behind diversification. Instead of depending on one thing to perform well, you spread your exposure across several things. Music makes this idea surprisingly interesting.
Michael Jackson took diversification much further
Taylor Swift still has one major weakness as an example. Whether the song is country, pop or folk, the success of every recording remains closely linked to Taylor Swift.
Michael Jackson eventually went much further. According to Sony, Jackson bought ATV Music Publishing for $47.5 million in 1985, giving him an interest in a catalogue containing thousands of compositions, including a large collection of Lennon-McCartney songs.
Jackson was no longer financially exposed only to Michael Jackson music. He had an interest in songs written by other people and performed by other artists. If interest in his own music declined, royalties from unrelated songs could still generate income.
Jackson later combined his publishing interests with Sony's publishing business. In 2016, Sony reported that it would pay $750 million for the Jackson Estate's remaining 50 percent interest in Sony/ATV. The comparison is not as simple as saying Jackson turned $47.5 million into $750 million, since the business changed dramatically over those decades, but it shows how valuable a large catalogue of music rights can become.
For a long time, this kind of investment was mostly available to major publishers, wealthy artists and specialist investors. That is starting to change.
You can now build a music portfolio yourself
Platforms such as ANote Music, Royalty Exchange and SongVest now let smaller investors gain exposure to music royalties. ANote lists parts of music catalogues that investors can buy into, Royalty Exchange operates a marketplace for royalty-producing rights, and SongVest offers SongShares tied to portions of royalty streams from individual songs.
This makes the idea much easier to picture. Instead of spending all your money on royalties from one song, you could spread it across several catalogues. One might contain older rock music, another modern pop, another Latin music and another songs that earn heavily from television licensing.
The question is whether that actually makes the investment safer. Before answering that, we first need to understand what a return from music even looks like.
How do you make money from a song?
Music rights can produce money in two main ways. The first is royalties. When music is streamed, played on radio, performed publicly or licensed for something like a film or advertisement, the relevant rights owners can receive payments.
The second is the value of the rights themselves. If you buy a royalty interest for €1,000 and later someone is willing to pay €1,200 for it, the asset itself has increased in value. This is similar to owning a stock, where an investor can receive dividends while also gaining or losing money as the share price changes.
ANote Music already publishes an index that tries to measure this distinction. Between August 2020 and May 2026, its Royalty Index showed an annualised royalty return of 10.72 percent, while its broader Total Return Index, which also includes changes in catalogue prices, returned 9.15 percent annually over the same period.
Those numbers do not mean every music investor earned around 10 percent. They describe ANote's own catalogue index, while an individual investor's result would depend on what they bought, the price they paid, fees and other factors.
Still, the index gives us something useful. It shows that we can start thinking about music in terms of income, price changes and overall return.
What if music had its own S&P 500?
Now imagine taking this much further. What if there were an index containing 500 of the world's biggest songs?
Call it the Music 500.
Instead of looking at whether one song had a good year, we could follow the income generated by hundreds of major songs together. Some would be new hits, some decades old, some pop, some rock, some country, some Latin and some from completely different markets.
The simplest version could track how much royalty income those 500 songs generate each year. If the portfolio earned $100 million one year and $106 million the next, its royalty income grew 6 percent. A more advanced version could also track what the rights to those songs are worth, giving us something closer to the return an investor might actually have earned.
That could answer questions that are surprisingly difficult today. Do old hits produce more stable income than new ones? Do Christmas songs behave differently because most of their revenue arrives during one part of the year? Do songs from different countries rise and fall together?
For context, IFPI reported that global recorded music revenue reached $31.7 billion in 2025, up 6.4 percent from the year before. But that does not mean a music investor earned 6.4 percent. Industry revenue and investment returns are different things, and a Music 500 could help show the difference.
Ten genres do not automatically make a diversified portfolio
Imagine owning royalties from ten different genres. It sounds diversified because the music itself is different, but perhaps every one of those songs earns most of its money through Spotify and Apple Music. A major change in streaming economics could then affect all ten at once.
Diversification works best when the things you own do not all depend on the same source of success. A stronger music portfolio might therefore spread risk across more than genre, including different countries, languages, ages of music and types of royalties.
A Christmas catalogue may behave differently from workout music. An old song that has earned steadily for 30 years may behave differently from a viral hit released six months ago. A song earning heavily from film licensing may also behave differently from one almost entirely dependent on streaming.
The important question is not simply how different the songs sound. It is how differently they make money.
AI makes building the catalogue much easier
This is where AI changes the discussion.
In our earlier article, AI Music Is Turning Songs Into a Portfolio Game, we looked at how generative tools change the economics of releasing music. If producing another track becomes faster and cheaper, creators can test more ideas instead of putting most of their resources behind a small number of releases.
The idea was to view an AI catalogue as a collection of experiments. Some songs may earn almost nothing, while a smaller number may find an audience and carry much more of the catalogue's value. The point was not simply to release as much music as possible, but to lower the cost of testing ideas and learn from what works.
Diversification adds another layer to that idea.
Imagine an AI music creator who normally makes electronic music. Creating a second catalogue of country music no longer necessarily means finding country musicians, booking studios and building an entirely new production setup. The same creator could potentially produce country, ambient, jazz, rock and many other styles at relatively low cost.
That makes it much easier to diversify what you produce. But it does not automatically diversify how you make money.
You could generate 1,000 songs across 50 genres, but if they are all distributed through the same platforms, promoted through the same channels and ignored by the same listeners, the income may not be diversified at all.
What are the returns on an AI music catalogue?
AI also changes what a good return might look like.
A traditional recording can require musicians, studio time, mixing, mastering and other production costs. If one song costs $5,000 to make and eventually earns $6,000, the financial result looks very different from an AI-assisted song that costs $50 to produce and earns $1,000.
The second song generated much less revenue, but it produced a much larger return relative to what was spent creating it. That means an AI catalogue should not be judged only by streams or total revenue. Production cost matters too.
This builds directly on the argument in AI Music Is Turning Songs Into a Portfolio Game. If AI lowers the cost of each experiment, a creator can afford more failed releases before finding one that works.
There is an obvious catch, though. The same technology that lowers your costs also lowers everyone else's costs. More music can be created while listener attention remains limited.
We explored another side of this in What Makes an AI-Generated Song Valuable?. As technically competent music becomes easier to produce, simply having another polished track becomes less unusual. Value may instead depend more on the song itself, the audience around it, artist identity, clear rights and whether anyone actually wants to hear the music again.
An AI catalogue containing thousands of cheap songs could therefore still be a terrible investment. Low production cost only helps if some of the music earns something back.
What would an AI Music 500 look like?
Now take the Music 500 idea and apply it specifically to AI. Imagine an AI Music 500 tracking 500 of the highest-earning AI-generated or heavily AI-assisted songs.
It could start answering questions we currently have very little data on. Do AI songs earn strongly for a short period and then fade as newer music replaces them? Does background, sleep or focus music produce steadier income than AI pop? Does most of the money end up concentrated in a small number of recognizable AI artists?
We could also compare the AI Music 500 with a wider Music 500 of conventionally produced music. Perhaps AI music would earn less revenue per song but cost much less to make. Perhaps catalogues spread across many genres would produce steadier income, or perhaps most AI songs would earn almost nothing while a tiny number of successful tracks carry the returns.
That last possibility fits closely with the portfolio argument from our earlier article. Cheap production makes it easier to try more ideas, but it does not guarantee that those ideas become valuable.
We simply do not have enough public data yet to know what those returns would look like. That is exactly what makes the question interesting.
A large catalogue is not automatically a good portfolio
The finance comparison gives AI music creators an important warning. More songs do not automatically mean less risk, more genres do not automatically mean more diversification, and cheaper production does not automatically mean better returns.
A useful catalogue would need songs that earn money for different reasons. Perhaps one performs well on streaming, another through licensing, another in a particular country and another as evergreen background music.
If every track depends on the same audience, platform and discovery system, the catalogue may be much less diversified than it appears.
What a real music index could teach us
Music is already an investable asset. Rights are bought and sold, catalogues generate income, and services such as ANote Music, Royalty Exchange and SongVest increasingly give smaller investors access to parts of that market.
AI now adds something different. Instead of only making it easier to invest in existing catalogues, it can make it much cheaper for creators to build large catalogues themselves.
The next question is whether those catalogues actually behave like diversified portfolios. A Music 500 could tell us how major songs perform together, while an AI Music 500 could show whether lower production costs create attractive returns, how quickly AI songs lose value and whether spreading releases across genres really spreads risk.
That would help formalize the ideas discussed in AI Music Is Turning Songs Into a Portfolio Game. AI makes it easier to treat songs as a collection of different bets rather than putting everything behind one release.
Diversification adds one important condition. The goal is not simply to make more bets, but to make bets that do not all depend on the same thing going right.
