MBW Views is a series of op-eds from eminent music industry people… with something to say. The following MBW op-ed comes from MusicInfra CEO and Co-founder Björn Lindvall (pictured inset).
Here, Lindvall argues that generative AI’s biggest threat to the music business isn’t to creativity but to infrastructure — and that the flood of data from AI-generated and AI-assisted music will overwhelm the industry’s aging royalty and attribution systems unless they’re rebuilt now.
We are seeing early signs of a music industry problem related to generative AI. It’s a challenge not to creativity, but to infrastructure. The problem: AI and the ways it can create and transform music will generate far more data than the current system can handle.
This is already visible. Recent reporting shows that Deezer is now seeing roughly [90,000] AI-generated tracks uploaded every day. This accounts for a meaningful share of new music being added to the platform. Many of these tracks are demonetized, but they still move through the same infrastructure. They are uploaded, processed, categorized, and in some cases analyzed for ownership and attribution. The volume alone is the key issue. The system is already under strain, and this level of output is still in the early stages.
The next phase of this problem may be even harder to manage. Fully AI-generated tracks are often easier to identify and isolate, but the landscape becomes far more complicated once smaller amounts of AI derived material start appearing inside otherwise human made songs. At that point, platforms and rightsholders may need to determine which parts of a track were AI-assisted, how those contributions should be tracked, and whether they should be compensated differently or excluded from royalty structures altogether.
This introduces a far more granular level of attribution and reporting than the industry is currently designed to handle.
AI music companies are seeking and securing licenses from labels and publishers. These agreements are a positive development for the industry because they recognize the value of music and establish a path for compensation. At the same time, they introduce a new level of complexity into how music usage is tracked and paid. Every token used to remix or generate any part of a song must be tracked and attributed back to rightsholders. Rightsholders must be paid, and the number of tokens involved in AI systems can reach into the billions.
This is not a simple scaling issue. It is a change in how usage itself is defined. The industry has been built around clear units of consumption such as streams, downloads, and performances. AI changes this structure. Music is no longer only played. It is modified, recombined, and used as input in generation systems that produce new outputs. Each step in that process can create new data that has licensing implications. Each step requires attribution. Each step requires payment logic that can operate at scale.
The challenge is not limited to generative AI companies. A wider set of products is emerging that also increases data volume. These include applications, social platforms, games, and other music driven services that use AI to create new experiences. Some generate personalized music in real time. Some allow users to remix or transform existing tracks. Others integrate music into interactive systems where usage is continuous rather than discrete. These systems also produce data. They also create usage events that must be tracked. They also require compensation structures that can handle complexity.
The combined effect of these developments is a large increase in the amount of music related data that needs to be processed and reconciled. What was once a relatively structured flow of usage data is becoming a constant stream of fragmented and overlapping events.
The infrastructure that supports the music industry today was not designed for this environment. The systems that handle usage tracking, royalty attribution, and payout processing are already under pressure from the streaming era. Hundreds of thousands of tracks are uploaded every day. Global listening happens across many platforms and formats. Even in this environment, there are long-standing issues with delayed reporting, inconsistent metadata, and incomplete attribution. Payments are often delayed. Ownership information is frequently unclear or disputed. These problems exist even before AI is added to the system.
AI increases both the volume and the granularity of data. It moves the industry from tracking full plays to tracking smaller and more complex forms of interaction. It introduces new forms of usage that are not always easy to categorize. It also increases the number of systems that must communicate with each other in real time. The result is additional strain on infrastructure that is already fragmented and outdated in many areas.
At the same time, there is broad agreement that the industry should grow. There is also broad agreement that music has been undervalued relative to its cultural and commercial importance. AI presents an opportunity to expand how music is used and how it generates revenue. It could create new licensing markets and new forms of engagement between artists and audiences.
However, none of this can happen without addressing the systems that support it. Growth without infrastructure creates risk. The ability to track usage accurately is directly tied to the ability to pay rightsholders fairly. If the systems cannot keep up with the level of activity, then trust in the system begins to weaken.
The core issue is that much of the music industry infrastructure is old. It has been built and modified over decades through a combination of internal systems and third-party solutions. In many cases it reflects layers of technical compromise rather than a unified design. This creates inefficiencies in how data is processed and how payments are calculated. It also limits how quickly the system can adapt to new forms of usage.
Other industries have dealt with similar challenges. They have rebuilt systems to handle large-scale, real-time data processing across global networks. They have created infrastructure that can support complex transactions with transparency and speed. Music can follow a similar path, but it requires coordinated effort and investment.
The choice is whether to address these limitations now or to continue building new layers of complexity on top of systems that are already under strain. AI will continue to increase the volume and complexity of music-related data. The question is whether the industry adapts in time to manage it.Music Business Worldwide
