The music industry’s tipping point is Right Here, Right NowStreaming is buckling under its own weight. The economics and structure that served it well in its first decade are not the ones that will get it through the next ten years. You might say that streaming is going through its ‘start up to scale up’ phase. AI is the disruption lightning rod of the moment, but transformational as it may prove to be, it is simply catalysing pre-existing disruptions. ‘Fixing’ the problems thrown up by AI would be dealing with symptoms rather than causes. The music industry is at a tipping point. There is still time for the creators and businesses within it to help shape what comes next, but that window of opportunity is both small, and closing.

Is anyone earning what they want from streaming?

When streaming first emerged, artists were worried it would not pay them enough; then the debate moved on to whether too much value lay with the biggest artists and labels; now with the superstar artist production line stuttering, the majors want a new royalty system to protect their income. Meanwhile, Spotify still struggles to generate a consistent profit. So the long tail, the majors, creators, and streaming services all think that streaming isn’t paying them enough. Which begs the question: just who or what is streaming paying enough?  Whatever the answer may be, the clear takeaway is that a royalty and remuneration system designed when albums, charts, downloads, and radio still ruled the roost, is failing to adapt to today’s much changed music world.

Remuneration pains are a symptom of consumption

A host of potential innovations are vying to be the solution to streaming’s remuneration woes (fan powered / user centric, two-tier licensing, etc.) but royalty challenges are the output, not the input. Streaming has shifted the majority of music behaviour from active listening to lean-back consumption, using algorithms to push consumers towards niches. The result is a consumption landscape shaped by fragmentation and passivity. There is a lot more consumption than before, with more consumers monetised, but the previous, finite artist economy has been replaced by an in-effect infinite song economy. Consumption needs ‘fixing’ before remuneration. 

While there are encouraging shifts towards monetising fandom, those tools will never have full effect if audiences are simply spending their time listening passively. There will, quite simply, be no fandom to monetise.

Machines on all sides

These are the two key sets of market dynamics that AI, and some other emerging technologies, will make worse, not better. Lean-back consumption is where AI will have the biggest, near-term impact. Context based playlists deliver music that is good enough. It is all about the overall soundscape rather than individual tracks, and even less about the artists. Production music libraries, like Epidemic Sound, have already shown that their music is plenty good enough for such playlists. Generative AI is waiting to pick up the baton, and may be able to do it even better if the music is specifically designed for the hyper-specific music that algorithms have taught consumers to expect. What is more, generative AI can get even more specific by evolving to the listener’s use case (i.e., like Endel). And if DSPs were to generate AI music themselves, then they could a) improve margins; b) stuff playlists; c) push users to the music. They who control the algorithm, control the listener.

And if that wasn’t bad enough for traditional labels and artists, a rising wave of virtual artists is hitting the market, such as K-pop acts Mave, Plave and Eternity, building on the foundations laid by the (now almost heritage) trailblazers like K/DA and Aespa. And even if these virtual artists have humans behind them, they are still a machine-centred challenge to wholly human artists (slightly crazy we even have to think in those terms these days!)

So, machines are opening a two-pronged attack on traditional labels and artists: 1) AI is competing for lean back, while 2) virtual artists compete for lean in (fandom).

Choose your poison

The industry’s strategy is to compel DSPs to take down problematic AI music and to keep the long tail in check with lower royalty rates. But that is unlikely to be enough. For example, why wouldn’t superstar virtual artists be eligible for the same royalty rate as superstar human artists? Regardless of whether the superstars are virtual or human, arguments that superstars deserve higher rates for pulling people to DSPs in the first place becomes less convincing every day, as consumption becomes ever more fragmented and ever less reliant on superstars.

But the scale of this problem is about to erupt like a volcano. Because the existential threat will come from AI in the hands of humans. AI will accelerate the consumerisation of creation trend that has been harnessed by artists and fans alike on TikTok, Snapchat, BandLab and a host of other places. Throw simpler-than-simple generative AI into social platforms and suddenly you have the potential for consumers creating ‘music’ at the same rate they create photos and videos. 

Millions of new ‘songs’ every day would break streaming royalties. So, labels would just get DSPs to keep those tracks off streaming, right? Not necessarily. These would be tracks made by people, so they would bring with them ready-made audiences of friends, family, colleagues and connections. Everyone becomes a fan of everyone else. It is the zenith of the network effect. And AI creations do not need to have millions of streams to disrupt streaming economics; millions of them only need to have at least one stream each.

And if friends can’t listen on DSPs, then they’ll listen on the social apps. Which means less time spent on streaming and further cultural dilution for DSPs. As one investment analyst put it to me: labels are faced with a ‘choose your poison’ choice, i.e., lower royalties now (due to dilution) or lower royalties later (due to smaller user bases).

Build a better train?

The entirely understandable temptation is to make what we have, work better. But sustaining innovation is unlikely to be enough. Just in the same way that it wasn’t enough for train companies to build better trains when Henry Ford’s new-fangled Model T car came to market. 

To be clear, building a better train is a not a bad option. Today, nearly a century on from when the last Model T rolled off the production line, trains still play a pivotal role. But for music, everything points to making streaming work better AND building something new.

Streaming fixed the problems of piracy and tumbling music sales. In doing so, it had the unintended consequence of commodifying music consumption. Without a new fork in the road, generative AI will simply hasten the utter domination of convenience. Pop will eat itself. AI will bring huge amount of value right across the music business, but portions of it will also hasten a reductive race to the bottom for convenient consumption.

Which is why, the time is now to start building plan B. To elevate a music world centred around fandom, identity, creativity, and exceptionalism. These are the fundamentally human elements of music that can (at least for now) clearly demarcate what is inevitably going to become a two-track music world. 

Five years ago, it would have been crazy to be thinking about how machines will shape the near future of both the business of music and of music itself. Just imagine what we might be discussing five years in the future?…..