May 2026
It is very rare for people in the music industry to have a consensus on anything. And yet, within the traditional music industry, the walled gardens concept for Gen AI platforms like Suno and Udio is increasingly treated as the safest solution for AI music (however it is defined now).
In addition to that, there has been a lot of critique towards DSPs (as always represented first of all by Spotify) for not making efforts to properly attribute AI music. As if users really cared (especially when some reports claim they can’t even distinguish it).
So, what if these two things are connected, and Spotify is not rushing with AI designations and separation for a reason? And if we listen to some of the most extreme views, Spotify should not allow AI music to be distributed or available at all. Deezer is known for a very proactive approach to AI distribution, and potentially they can become pioneers in restricting AI uploads. What if others see more downside than upside in doing this?
Imagine a case where the walled gardens approach is universally applied and the music generated on Suno and Udio never leaves their walls. I guess there are many people and independent artists arguing for this approach. As a result, you have millions of people creating music there, and not all of them are royalty scammers. If we assume for a moment that there are true talents using AI to create songs, or maybe creating white noise with functional value, we get to a situation where something worth listening to is not available on Spotify. This creates an alternative environment for music consumption.
AI platforms argue that they create additional value and potentially increase the music revenue pie rather than eating existing shares from human-first music on streaming platforms. I agree with the part related to creating new ways of producing music, like superfans making covers and remixes. But does it also mean people will dedicate more of their lives to actually listening to this music? Where exactly is this additional time going to come from if people already have everything they need to fill their lives with music available on streaming? I struggle to understand this part of the story. Music has a certain place in everyone’s life, very different across listener groups, but is it expected that the fan-made version of Man I Need by Olivia Dean will win the attention competition against Netflix, podcasts, or TikTok videos?
If not, we get to a point where the industry pushes part of its listeners (and revenue with them) from Spotify to pure AI platforms. Artists achieve the goal of not letting AI music flood DSPs. But those same artists are likely to get paid less now since their share of the pie is smaller and they are not participating in the new “AI” part — unless they license their music to those AI platforms for recreation.
We don’t know for sure, but it means the net-net impact for right holders will depend on pricing for users (size of the new pie) and the terms and revenue share models of AI platforms (split of the pie). And while we know quite well how Spotify shares revenue and how people react to price hikes, it remains a mystery whether licensing deals between labels and these platforms are any less attractive than DSP deals right now. There is also uncertainty around public performance rights, mechanical rights, and the connected royalties.
Keeping all of this in mind, is it really beneficial for right holders to fight for the walled gardens framework?
While I was writing these thoughts down, Universal Music Group and Spotify came out with an extremely relevant announcement providing their perspective on the place of AI-generated music within existing DSP infrastructure. Spotify is going to allow users to create remixes and covers of songs for which artists opted in. Making it available for an additional fee ensures that the pie is not going to get smaller, but actually grows. It directly attacks the ethically responsible value proposition that Suno and Udio use as one of the main use cases justifying their existence.
What is left other than scams and AI slop now? Personalised background music (don’t we have enough playlists already? OpenAI’s ChatGPT is also integrating with Spotify). Sounds for social media and visual content (more likely to become part of video editing apps like Edits and CapCut). Low-cost commercial utility music (probably the best use case for it).
Probably the only valuable application remaining is true music creation, just like smartphones enabled photography. It requires skill and practice, and most likely integration with DAWs. A great creative tool, but not as groundbreaking or destructive to true musicianship as some people anticipate.
Unless this music stays within walled gardens and those platforms turn into streaming alternatives with exclusive content. DSPs used exclusive releases from big names before with mixed success. What if this time with millions of creators instead of a few superstars the result will be different and not in favour of the “old” industry?
Get new notes by email