Sustaining vs. Disruptive Technology

On Spotify’s latest earnings call, there was a few moments that felt more important than the raw numbers (which were impressive in and of themselves- revenue grew 13%, gross margin 33.1% and over $11bn paid out to rights holders in 2025 alone).

Michael Nagle/Bloomberg via Getty Images

As a tangent on this tangent- Sachin Saggar also provided an excellent recent analysis on how Discovery Mode favours Spotify’s gross margin objectives in his “Flipside” monthly newsletter.

Back to the point, and the Spotify earnings call. There were some revealing lines from Co-CEO Gustav Söderström:

“Technology is seldom disruptive on its own. Significant disruption happens when new technologies enable new asymmetric business models.”

That framing, echoed and expanded on by Ben Thompson at Stratechery and unpacked by Dan Fowler in a recent post cuts to the heart of the current AI debate in music and, more broadly, in technology.

The real question isn’t whether AI is powerful, but whether it enables new business models entirely.

The Innovator’s Dilemma, Revisited

In The Innovator's Dilemma, Clayton Christensen draws a distinction between sustaining and disruptive innovations. Sustaining technologies improve performance within an existing business model, whereas disruptive technologies enable a fundamentally different model- often one that starts with inferior metrics but unlocks new markets and growth.

Streaming was disruptive not because it was technically better than downloads on day one, but because it enabled a new model: subscription access over ownership. Spotify wasn’t just a better iTunes, but changed the entire economic structure of the industry. Söderström used this and another example in the call- Uber, who didn’t invent GPS or mobile payments but indeed utilised them to unlock a new market-clearing mechanism.

Does AI also fit this same bill, as an enabler? And if so, does its impact have a sustaining or disruptive effect?

AI in Music: Sustaining or Disruptive

The dominant narrative suggests AI is inherently disruptive. AI-generated music, voice cloning, remixing, stem separation. Vast volumes of AI generated content. AI agents all around (insert Matrix GIF here).

The assumption is that because the technology feels radical, the market impact must be too- but technology alone doesn’t dictate structure.

Spotify’s argument is that in consumer media, the dominant model (subscription plus advertising) is likely to endure. In that world, AI becomes sustaining: a force multiplier enhancing personalisation, engagement, retention, and monetisation within an existing flywheel.

Consider what Spotify emphasised:

• 90 million users have tried AI DJ

• Prompt-based playlist creation is building a new “language-to-music” dataset

• Engineers are shipping code faster using AI-assisted tools

• AI increases engagement → increases retention → increases LTV

That is classic sustaining innovation- AI is strengthening incumbents who already have scale, data, and distribution along the tenets that made them acquire that scale in the first place.

In fact, AI may disproportionately benefit platforms that already sit at the centre of user attention. If personalisation improves, churn falls. If engagement rises, pricing power improves. If more catalog is uploaded, discovery becomes even more valuable.

From this perspective AI isn’t breaking the model, but deepening it.

Where Disruption Could Still Emerge

That said, it would be naïve to assume AI won't be disruptive. Christensen’s insight wasn’t that incumbents always win, but that incumbents often miss core shifts that redefine value. The challenge is for existing (particularly successfully) businesses to entertain and engage with potential seismic tremors- being willing to disrupt their own, currently successful, models so as not to be outstripped by newer upstarts without legacy baggage.

The open question then, is this- does AI enable a new business model in music?

Some possibilities:

1. Outcome-based personalisation. If AI agents curate not just content but emotional states, productivity, social coordination- does value shift from access to outcomes?

2. User-generated infinite supply. If listeners can generate perfectly personalised tracks instantly, does the scarcity that underpins catalog economics erode?

3. Disintermediated distribution. Could AI-native platforms combine creation and distribution in ways that collapse the distinction between DSP and studio?

So far these models haven't proven durable at scale. Most AI music tools still rely on the same downstream distribution infrastructure to reach audiences. As Söderström noted, cultural moments still “happen” on scaled platforms (which of course, he’d like to maintain).

But history teaches that disruption often begins in overlooked niches. The first streaming services were inferior to CDs on quality and selection, and the first Ubers delivered a worse service than taxis in many cities. Disruption isn’t always obvious at inception.

Aggregation in Low-Friction Environments

Another natural conclusion from Spotify's positioning- when friction drops markets often aggregate rather than fragment. The internet was supposed to lead to infinite independent websites. It did in a way, but a larger trend was the enablement of platforms to build gigantic scale.

AI dramatically lowers the cost of software production (trending to zero). Spotify engineers can now generate and test code on a commute. That increases velocity (which could, and probably will, be a newsletter in its own right). But lower production friction doesn’t automatically mean market fragmentation- it can just as easily reinforce aggregators who can ship faster, test more, and integrate AI at scale.

In a world where content supply explodes, curation will become more valuable. Spotify’s thesis appears to be that AI only increases the importance of personalisation and data moats. And it makes sense- if supply grows from tens of millions of tracks to hundreds of millions, the platforms that best interpret taste win. This is also paralleled in data aggregation and analytics- as we've explored previously, interpretation (wisdom) at scale is the true differentiator.

Taste (and wisdom) is not canonical. It’s contextual, cultural, and dynamic. Spotify hints at where defensibility may lie: proprietary behavioural datasets that can filter noise. Again, sustaining- but strengthening.

A Strategic Choice Then

For music and technology companies alike, the key strategic question is not “Is AI disruptive?” but rather “Does AI enable a business model that makes our current one obsolete?”

If the answer is no, then the winning strategy is integration. Move fast, invest early, align incentives, strengthen margins. If the answer is yes, then the battle is structural- and subsequently existential.

Spotify is keeping its eye on both, but doubling down on the former. AI enhances engagement, increases retention, improves margins, accelerates development. It fits the existing flywheel.

But markets are rarely static- the companies that survive long-term are those that can both exploit sustaining innovation and remain alert to genuine disruption. Christensen warned that incumbents often over-invest in sustaining improvements because these serve their best customers- the danger is a new model redefines what “good” means, and unlocks different, new, customers.

The music industry has already lived through one such redefinition- ownership giving way to access, with revenue shifting from transactional to recurring. Whether AI marks the next structural shift remains to be seen.

But the most important thing to remember is this: Technology is not destiny- business models are.