As 2025 comes to an end, we can see signs that the music industry may be ready to move on from a model that hasn’t really changed in over a decade. Every mode of distributing music to the public has its era, and then something else takes over. 2026 could be the year when music services like Spotify, Apple Music, YouTube Music, and Deezer level off in their industry impact. We have a few clues about what’s coming next that 2026 should bring more into focus.
The basic model of today’s streaming services – unlimited listening from a huge catalog of music as long as you pay a fixed monthly subscription fee – dates back to 2001. The current marketplace, with mobile apps, publishable playlists, and free listening with ads, goes back to 2011. Apart from a shift in consumption from on-demand (users pick the specific music they want to hear) to playlists and increasingly sophisticated algorithms, not much has changed since then.
As the chart below shows, streaming has been the dominant source of revenue in the music industry for ten years and the industry’s growth engine for even longer. But that growth has stalled in Western markets. Using inflation-adjusted RIAA numbers, U.S. recorded music revenue from streaming was flat from 2023 through the first half of 2025, and total revenue was actually down 1.7% from the first half of 2024 to the first half of 2025.
(Outside of the West is a different story: IFPI figures show overall revenue growth still above 20% in Africa and South America, though subscription prices in those territories are much lower than in North America and Europe, which collectively account for 70% of total global revenue.)
Annual U.S. recorded music revenues, 1973-2024, $millions, 2024 dollars. Source: RIAA.
Bill Rosenblatt
Music Formats Come and Go
The chart also shows the various other eras of music distribution that the RIAA has tracked since 1973: vinyl, tapes (8-tracks and cassettes), CDs, digital downloads, and streaming. As it shows, eras come and go. Streaming should be no different.
Yet there is one important difference between streaming and prior formats: when one format passed its peak, we knew what was coming next; but with streaming, we don’t. When vinyl hit its peak in 1978, cassettes were on the rise because of their portability and home recording potential. CDs were already surging in popularity when cassettes dominated the market in the late 1980s. The launch of Napster in 1999 led to the downfall of CDs; we all knew that digital downloads would be next in some form. And we knew that streaming would succeed downloads when Spotify launched in the U.S. in 2011; downloads started their decline three years after that.
In contrast, there is no “format” for distributing music to the public that is the obvious successor to streaming. The major record label groups and major streaming platforms are currently negotiating what amount to tweaks to the existing models, such as higher-priced subscription tiers with features such as high-resolution audio and superfan perks. But these will, at best, extend the growth of subscription streaming revenue for a couple of years.
We do have some idea that AI will power the next phase of the music industry. We already know that AI can be powerful for creating music, but we aren’t sure how it will affect distribution and consumption. Yet we’re starting to get a few clues – clues that should become more apparent in the coming year. These clues tell us how the next sources of activity for the music industry will move away from the existing streaming models.
AI-Powered Fraud
One such clue lies within the huge numbers of new tracks that users are producing with generative AI and uploading to streaming services. Deezer has been providing highly illuminating data about this, which suggests that far more of these tracks are being generated for fraud than because fans actually want to hear them.
Deezer has technology for detecting both streaming fraud and pure AI-generated tracks. The service has reported that 70% of the streams that it detects as fraudulent are streams of pure AI tracks, while only 0.5% of legitimate streams are of pure AI tracks. Meanwhile the volume of pure AI tracks that Deezer takes in daily rose sharply from 10,000 at the beginning of 2025 to 50,000 in November. That’s a third of the roughly 150,000 tracks that it accepts per day on average – even as that total daily number has leveled off over the past two years.
This may suggest that the volume of tracks that legitimate artists and labels submit to music services is decreasing. But the much more likely explanation is that streaming services and the indie distributors that feed them (such as TuneCore, Symphonic, and DistroKid) are getting better at detecting and rejecting fraud and spam tracks – even as the number of pure AI tracks that they don’t block skyrockets.
Thus, fraudsters are turning to AI for the tracks they stream to generate bogus royalties; AI enables them to generate such tracks in unprecedented volumes and in ways that are more likely to evade detection. And it’s reasonable to assume that other major music services besides Deezer have the same issues.
This is increasingly problematic for everyone in the value chain: it diverts royalties to fraudsters that should be going to labels and legitimate artists; it mars discovery and other aspects of the user experience; and it imposes burdens of storage, administration, and fraud detection on the service providers themselves.
The problem has become an AI-powered arms race between fraudsters and fraud detection schemes (which themselves use AI). It could easily grow to the point at which it moves from a peripheral issue to a defining factor in music services’ user experiences and profitability. Consider that at the current pace of acceleration, without any structural changes, pure AI tracks submitted to music services will outnumber those created by humans (with or without AI assistance) by the middle of 2026.
Music services will need to become more selective in the tracks they accept and decide how much effort they are willing to expend on detecting bad actors. In addition to the potential for criminal liability, services that ignore this could be doomed to the same fate as MySpace, the social network of the 2000s that was once a vibrant community for indie musicians and their fans but deteriorated into a hive of spam, fraud, and predation.
Music services are already taking tentative steps in these directions. Deezer has a policy of not paying royalties on pure AI tracks and not including them in algorithmic playlists. Spotify has been culling large numbers of tracks from its catalog, though it has used the vague term “spammy” to describe them. Amazon and SoundCloud (as well as Spotify) are members of the nonprofit Music Fights Fraud Alliance along with several digital distributors. It’s also quite possible that other major music services are taking similar steps but not publicizing them.
Meanwhile, the major labels’ responses to this – which arise out of settlements in their lawsuits against Suno and Udio – have been to get those generative AI platforms to restrict downloads of generated tracks from their walled gardens, to limit the tracks that end up on Spotify or other music services. It remains to be seen how successful this will be in stemming the deluge of AI tracks submitted to service providers, but it’s safe to assume that it will only slow the problem down as users turn to other AI platforms.
Specialized Music Services for Specialized Users
These factors will in turn create opportunities for more heavily curated and specialized services, which will be more attractive to certain audiences and less attractive to fraudsters.
Some of these will focus on specific genres of music or on independent artists. Indie-focused services follow in the footsteps of pioneers like Bandcamp and Audiomack; the latter is thriving as a hub for indie hip-hop, Afrobeat, and Latin music. Existing genre- or geography-specific services like Deedo (Francophone-focused African music), JioSaavn (South Asian), and IDAGIO (classical) should also flourish.
A new entrant in this area is Vocana, a streaming platform (currently in beta) focused exclusively on indie artists. Vocana gives artists access to users’ email addresses and other fan data that the major music services don’t share, emphasizes user-driven discovery through social features (a la MySpace) over algorithms, and employs a user-centric royalty model that rewards artists with more devoted fans.
Other services will use AI as a tool for fan engagement. One example is Hook, a music-focused social network app that enables users to create and share AI-powered remixes and mashups of tracks in its fully licensed library of music from over 1200 artists. (Audiomack introduced a simpler, non-AI version of this idea called Audiomod last year.)
The industry has been talking for many years about tools and services that enable fans to interact directly with music from their favorite artists; these apps are finally making that a reality. They will appeal to certain types of fans, and their users’ remixes and mashups will be far more meaningful in their own environment than they would in the vast oceans of content on Spotify or Apple Music or YouTube. Genre-focused services also focus on communities of fans; they are powered by genre-specific metadata that is disproportionately resource-intensive to maintain but enhances discovery.
One can also imagine future services that use AI – perhaps from platforms like Suno or Udio – to generate “new” music from scratch in near-real time to meet users’ occasions and moods. But the major labels would surely push back on existing music services that attempt to add such features, so they are more likely to be relegated to isolated services for the foreseeable future.
2026 should see more development and proliferation of these specialized types of music services. Labels and artists should encourage them because they can be attractive sources of revenue and fan engagement for recorded music after general music service subscriberships stop growing. Whether the major music services end up developing or owning these niche platforms, and whether they coalesce into a movement that we can say is the successor to streaming, are questions that should take longer than 2026 to answer.
