Site icon News Today

How Many Streams Are AI Tracks Actually Generating on DSPs?


Photo Credit: Ilias Chebbi

We know AI audio is heavy on volume and comparatively light on actual listeners. But are machine-generated tracks racking up far fewer streams and fans than most believe?

DMN Pro explored that interesting question yesterday, shortly after Tencent Music described its services’ AI music consumption as “negligible” and representing “a declining share of listening.”

The disclosure didn’t necessarily come as a revelation: Deezer had previously pointed to AI tracks’ (meaning those pumped out via generative models with no human contributions) accounting for between 1% and 3% of on-platform streams – with 85% thereof flagged as fraudulent and demonetized.

But it did provide fresh evidence supporting the important idea that AI slop, staggering peak release volume aside, is failing to resonate with fans and divert support from proper artists.

And it also set the stage for further discussions about AI audio’s precise prevalence across on-demand platforms.

Soon after DMN Pro published its report, a top exec at a major distributor reached out with a telling stat: Of the 300 million or so actively available tracks this distributor has released on DSPs across the U.S. and Europe, only 40 million have generated at least one stream.

Put differently, 260 million of the uploads, about 87%, haven’t received even a single play on platforms such as Spotify, Apple Music, and YouTube Music.

One can be forgiven for questioning the astonishing stat off the bat – though the source is credible, and the figure aligns with other evidence. Amid debate over Spotify’s controversial 1,000-stream royalties minimum, Luminate indicated that two-thirds of on-platform tracks had received between zero and 100 plays way back in 2023.

Of course, much – referring in large part to the AI boom’s effects – has changed in the interim. And last year, Deezer signaled that 78% of new uploads had failed to register a stream during the prior 12 months.

Naturally, the data points raise additional questions yet. Chief among them: If the overwhelming majority of AI listening is attributable to fraud on leading DSPs – and if fans are simply ignoring most of the uploads – has the machine-generated audio avalanche (and the commercial fallout thereof for real music) been overblown?

Perhaps to some extent, and given Suno’s aggressive push to subpoena stream-volume data from the Mechanical Licensing Collective and Apple, one can rest assured that we’ve not heard the last of the subject.

At the same time, it’s important to distinguish between AI audio’s stealing away fans from actual music (which appears increasingly unlikely) and diverting streams as well as listeners (which was and seemingly still is happening).

That Spotify pulled down 75 million (and probably more; higher-ups aren’t eager to divulge an updated total) “spammy” tracks is widely known. So is the media coverage of viral AI “artists” such as The Velvet Sundown and Breaking Rust.

As analyzed by DMN Pro, these virtual acts have receded into the background just as quickly as they burst onto the scene. And while they’ve also experienced material monthly listener decreases on Spotify – evidently, true fans accounted for a minority of the totals – the consumption falloffs aren’t solely attributable to reduced media mentions and waning virality.

To be sure, an “AI Persona” called Agtverse received far less attention than The Velvet Sundown despite boasting more Spotify monthly listeners (257,000 versus 208,000) as of December 2025. At present, Agtverse’s listener count is hovering beneath 50,000 – down 82% in roughly one year.

Similarly, SireWolf suffered an 88% slip, to 22,000 monthly listeners, during the same window. Algorithmic changes and adjacent market realities come to mind as possible culprits here.

If certain fans are indifferent or open to AI audio, many others are ardent critics; that AI slop has infiltrated several of their favorite artists’ Spotify pages isn’t helping. (A quick social media scan will, in fact, turn up firmly worded posts penned by disappointed supporters who found that “new releases” were unauthorized machine-generated garbage.)

At the intersection of both points, vehement fan and artist opposition to AI isn’t exactly a positive for a platform that’s preparing to roll out a paid remix and cover tool.

Especially because only one of the three major labels has licensed this tool, maintaining relatively positive professional relationships with indies (i.e., ensuring that their comparatively small catalogs don’t drown in a sea of lower-cost audio) is presumably a priority for Spotify.

Last month, the DSP revealed it’d booted 50 million AI-generated tracks from recommendations since 2026’s beginning.





–

Exit mobile version