Suno and BMG Are Rebuilding AI Music Around Licensed Rights

By Brian k. Neal

Suno did something unusual before the music industry had decided what to do with it.

It built the market first.

More than 100 million people have used Suno. The company has said it reached more than two million paying subscribers and roughly $300 million in annual recurring revenue. In June, it raised more than $400 million at a $5.4 billion valuation. What began as a startling demonstration that artificial intelligence could generate a convincing song from a prompt has already become a substantial consumer business.

That makes Suno’s new partnership with BMG more consequential than another AI licensing announcement.

On August 12, Suno and BMG announced a global strategic alliance covering BMG’s recorded-music and music-publishing businesses. The agreement creates a framework for BMG artists and songwriters who choose to participate in new AI opportunities, with compensation and rights protections attached. It also settles Suno’s prior use of BMG-controlled recordings and publishing works.

But the settlement isn’t the real play.

BMG is joining Suno as the company prepares what it calls its first music model developed in partnership with the music industry.

Suno isn’t simply licensing music for the product it already built.

It is rebuilding the permission structure underneath it.

That distinction matters because Suno developed in almost the reverse order of the traditional streaming business.

A streaming service needs licensed music before it has much of a consumer proposition. Without recordings, there is nothing meaningful to stream.

Suno’s proposition was different. The product wasn’t access to a library of finished songs. It was access to the ability to make one.

Type an idea. Describe a sound. Wait a few moments. Music appears.

That simplicity allowed the consumer experience to spread while a much more complicated question remained almost invisible beneath the prompt:

What gave the machine the ability to understand music well enough to make it?

That question eventually reached the courts.

In 2024, major record companies sued Suno and rival Udio, alleging copyrighted recordings had been copied without authorization to train their models. Suno defended its practices, and the confrontation became part of the much larger argument over whether generative AI companies can train on copyrighted material without permission.

For a while, that made the battle look relatively simple: AI companies on one side, the traditional music industry on the other.

That picture is beginning to break apart.

Warner Music Group settled its litigation with Suno in November 2025 and entered a partnership with the company. Warner said Suno would introduce new licensed models in 2026 and eventually retire its existing models. Participating artists and songwriters would have control over certain uses involving their names, images, likenesses, voices and compositions.

Now BMG is joining that transition.

Its agreement simultaneously deals with yesterday and begins negotiating tomorrow.

The previous use gets settled.

Future participation gets structured.

And the technology keeps moving.

That is why this shouldn’t simply be read as another music company embracing artificial intelligence. Something more fundamental is happening.

The copyright fight is beginning to become a business model.

The first argument was whether permission was required.

The emerging argument is about what permission means.

Who can grant it? What does it cover? What does it cost? Who gets paid? Can a songwriter participate while an artist does not? Does permission involving a composition extend to a recording? What happens with an artist’s voice or identity? Can participation eventually be withdrawn?

Suno and BMG have not publicly answered all of those questions.

And that matters.

The existence of a licensing framework shouldn’t be confused with the resolution of the economics underneath it. BMG describes participation as opt-in, and neither company has publicly disclosed the complete financial mechanics of the arrangement.

So this isn’t a story about Suno suddenly gaining unrestricted access to everything BMG controls.

It is about something more structural.

Rights are becoming part of the product design.

For most of the recorded-music business, licensing determined where a finished song could go.

A recording could be sold, played on radio, streamed, synchronized into a movie, placed in an advertisement or licensed into a video game. Different rights and payments surrounded each use, but the basic relationship remained familiar: someone created music and someone else consumed it.

Generative AI changes that relationship because the citizen is no longer limited to choosing what to play.

The citizen can participate in making something.

That means an AI music company’s competitive advantage may eventually depend not only on what its model knows how to do, but on what it has permission to do.

We can already see that competition beginning.

Udio has reached agreements with major music companies around licensed AI creation. Klay has secured agreements involving Universal, Sony and Warner while positioning its technology around licensed music. Warner has partnered with both Udio and Suno.

The music industry isn’t choosing one AI winner.

It is beginning to establish relationships across competing systems.

That moves AI music into a different race.

The first race was technological:

Can the machine make a convincing song?

That barrier fell remarkably quickly.

The next race is institutional:

Can you build a powerful music model with enough legitimate rights underneath it to operate at enormous consumer scale?

Model quality matters. Computing power matters. Price and interface matter.

But rights are different.

They have owners.

They have contracts.

They have histories.

They have competing interests.

And they have to be negotiated.

That makes rights themselves a competitive asset.

For Suno, this transition is particularly consequential because the company isn’t trying to establish those relationships before anyone has used the product.

It is doing it after millions of people already have.

That changes the leverage on both sides.

The music industry controls something Suno increasingly needs: legitimate rights.

Suno controls something the music industry increasingly has reason to understand: a large population of people who have demonstrated that they want to participate in making music through a prompt.

BMG isn’t negotiating with a laboratory experiment.

It is negotiating with an existing consumer behavior.

And Suno isn’t negotiating with an interchangeable supplier. BMG operates across recordings and publishing, placing it within multiple layers of the rights system Suno now needs to navigate.

Neither side has everything it needs on its own.

That helps explain why litigation and partnership no longer look like opposites.

A lawsuit can challenge how a technology was built while a licensing agreement determines the conditions under which the next version can operate.

Warner moved from plaintiff to partner.

BMG’s agreement resolves previous-use issues while creating a framework for future participation.

The conflict doesn’t simply disappear.

It becomes negotiation.

And that negotiation is now reaching inside the product itself.

Suno has said its existing models will eventually be retired as the next generation arrives. The company has also introduced measures around watermarking and fingerprinting and is changing its download policies beginning September 3.

Those moves matter because Suno isn’t only trying to answer what can legally enter an AI music system.

It is beginning to confront what happens when enormous amounts of music can come out of one.

Generative AI radically reduces the friction involved in making music. A person who couldn’t realistically produce one finished recording can generate many. Someone capable of making dozens can potentially make thousands.

That creates a problem beyond the original training debate.

It creates abundance.

A world in which almost anyone can generate music is also a world in which streaming services, discovery systems, royalty pools and human attention could face quantities of music unlike anything the traditional recording business was designed to absorb.

Suno’s attempts to identify generated works and place more friction around mass exporting therefore belong alongside the licensing conversation.

The company is increasingly dealing with both sides of the machine:

What is allowed to go in.

And:

What happens to everything that comes out.

BMG doesn’t solve that problem.

Neither does Warner.

And today’s partnership shouldn’t be presented as though the music industry has finally figured out generative AI.

It hasn’t.

The economics remain largely private. Creator compensation remains an enormous question. Universal and Sony remain part of Suno’s unresolved legal landscape even as the major music companies build relationships with other AI platforms.

There is no settled model yet.

But there is now enough movement to see what is changing.

Suno built consumer demand before it finished building the rights architecture capable of supporting that demand.

The technology arrived.

People used it.

Investors funded it.

Copyright owners challenged it.

And now the authorized version is being negotiated while the earlier version of the product is still standing.

For the citizen, that transition may initially be almost invisible.

A college student sitting in a dorm room can still encounter Suno as something remarkably simple: describe the music you want and receive a song.

But underneath that simple interaction, the system is becoming considerably more complicated.

Licenses.

Publishing rights.

Recording rights.

Creator opt-ins.

Identity protections.

Compensation.

Model training.

Generated-output identification.

Distribution rules.

The citizen may never see most of that architecture.

But those invisible relationships may ultimately determine which AI music companies survive long after generating a convincing song becomes ordinary technology.

That is what makes BMG’s arrival significant.

BMG isn’t simply another catalog Suno can place behind its technology. It is another rights holder willing to participate in the model intended to replace the generation of Suno technology that helped create the copyright fight in the first place.

The first race in AI music was to make the machine capable of creating a song.

Suno proved there was an audience for that.

The race now is to build a rights system capable of supporting what the machine can already do.

And Suno and BMG are building directly inside that race.