Spotify is starting to sell the technology it built for itself

Look for a line called “technology sold to enterprises” in Spotify's accounts and you will not find one.

In the second quarter of 2026, Spotify reported €4.777 billion in revenue. Premium produced €4.331 billion, or 91%. Advertising produced the remaining €446 million, or 9%.1

That is the published split.

No Backstage line. No Portal line. No Confidence line. No convenient “developer tools” percentage waiting to become a chart.

So the first answer to “how much of Spotify's revenue comes from technology rather than music?” is annoyingly simple: public data does not let us calculate it.

There is another trap in the question. Premium is not the same thing as “music revenue.” Spotify's subscription economics include music royalties, audiobooks and different partner costs. The company reports its business as Premium versus Ad-Supported, not by the exact kind of content being consumed.12

Still, underneath the very visible consumer company, Spotify is building another business.

It sells developer portals, engineering plugins, enterprise support and an experimentation platform. Several of those products came directly from systems Spotify first built to solve its own internal problems.

That is where the Amazon Web Services comparison becomes useful.

Not because Spotify is about to fill the planet with data centers. Almost the opposite: Spotify says Google Cloud Platform provides the vast majority of its primary data storage and computing.1

The similarity is more general: a large company builds infrastructure for itself, discovers that other companies have the same problem, then turns part of the solution into a product.

Amazon did it with AWS. Spotify is now trying to commercialize parts of the way Spotify itself gets built.

The developer business is still invisible in the financial statements

The most useful table starts with what we do not know.

The periods and currencies differ, so this is not a size comparison. It shows a reporting difference. Spotify does not currently present its developer-software activity as a separate financial segment. AWS, meanwhile, has become large enough to change how Amazon's entire income statement is read.

For full-year 2025, Spotify generated €17.186 billion in revenue: €15.350 billion from Premium and €1.836 billion from advertising.2

The B2B products are real. Their individual revenue, ARR, customer counts and margins are simply not disclosed in the financial documents we reviewed.

It would be tempting to translate that silence into “basically nothing.” Maybe. But that would be a guess wearing a tie.

The narrower statement is the useful one: Spotify does not disclose their weight separately. Public accounts do not tell us whether they are too small to be material, included elsewhere, or both.

This is exactly where AWS history becomes interesting.

AWS shows why revenue share can be the wrong number to stare at

Amazon says AWS grew out of its own difficulty provisioning and managing infrastructure for internal teams. It had built reusable components for itself, then realized those components could form a platform for other companies. S3 and EC2 arrived in 2006.13

By 2015, AWS was much easier to see in Amazon's accounts. It generated $7.880 billion of Amazon's $107.006 billion in sales, about 7.4% of the group.14

That was not a huge revenue share.

But AWS reported a segment operating margin of roughly 23.6% that year. Amazon's consolidated operating margin was around 2.1%.14

There is an accounting caveat. Amazon did not allocate every cost, including certain stock-based compensation expenses, to segments in the same way. So the two margins are not a perfect apples-to-apples comparison.

The economic contrast was still hard to miss. Selling compute on demand had a very different cost structure from moving physical goods, subsidizing delivery and operating a vast retail network.

Ten years later, the difference is much larger.

In 2025, AWS generated $128.725 billion, roughly 18% of Amazon's total revenue. Its operating income was $45.606 billion, while Amazon as a whole reported $79.975 billion.15

That means AWS produced about 57% of Amazon's consolidated operating income while contributing 18% of sales.

None of this predicts Spotify's future. Backstage does not become AWS because both stories contain the phrase “internal tool.”

It does show why asking only for revenue share can miss the point.

A business can be relatively small in revenue while becoming disproportionately important in margin, growth, bargaining power or strategic value.

Backstage existed because Spotify was getting difficult to build

As Spotify grew, developers had to navigate more services, tools, pipelines and documentation. Writing code was no longer the whole problem. People also needed to know where things lived, who owned them and how to use them without memorizing the company.

Spotify built Backstage as an internal developer portal to put a usable interface over that complexity.3

It brings service catalogs, documentation, ownership, templates and infrastructure tooling into one place.

In March 2020, Spotify released Backstage as open source.3

Backstage is now a project hosted by the Cloud Native Computing Foundation.5 That makes the path different from simply launching a proprietary SaaS product. Spotify first allowed a shared layer to become infrastructure for a wider ecosystem.

The problem is not glamorous. That is part of what makes it credible.

Spotify currently describes an internal engineering environment with 800 teams, 2,900 production microservices, 37,200 active data pipelines and more than 2,900 production deployments per day.4 Those are Spotify's own figures, not independent proof that Backstage makes engineering better. They do show the amount of organizational surface the portal has to make navigable.

In a small team, somebody can still ask across the room: “where is the service that does this?”

At thousands of components, architecture by shouting becomes less scalable.

Backstage turns part of an organization into a product surface: catalog, conventions, creation paths, ownership and shared interfaces.

Then Spotify started selling the industrialized version of that knowledge.

Open source becomes the ground floor, paid products sit above it

In October 2025, Spotify Portal reached general availability.6

Portal is Spotify's commercial managed developer portal built on Backstage. Spotify handles hosting, upgrades and parts of the operational setup. The company also sells premium plugins and enterprise support.7

The sequence is familiar:

1. solve an internal problem;
2. extract a generic layer;
3. open it so an ecosystem can form;
4. charge for the pieces enterprises would rather buy than maintain themselves.

The slightly ridiculous detail is that Spotify sells some of this through AWS Marketplace.8

The public Spotify Plugins for Backstage listing currently shows a $100,000 twelve-month user-contract dimension, while noting that purchases are generally made through private offers and pricing can vary.9

That number tells us almost nothing about Spotify's total revenue. We do not know contract count, discounts or deployment sizes.

It does prove something more modest: this is no longer just a nice open-source side project. Spotify now has enterprise software SKUs with enterprise-sized price tags.

Backstage is not the only example.

Confidence sells another internal Spotify habit

Spotify also commercializes Confidence, its experimentation and feature-management platform.10

The underlying practice is familiar inside large software products: release a feature to part of the audience, measure what changes, compare variants and decide from evidence rather than from whoever sounds most certain in the meeting.

Confidence currently has a free plan, a Growth plan starting at $449 per month, and custom Enterprise contracts.10

That looks much more like conventional SaaS than a streaming service.

Spotify now has several visible technical monetization layers:

None of these products is separately visible against Spotify's €17 billion annual revenue today.

Together, however, they describe a strategy.

Spotify is trying to turn the cost of learning how to operate Spotify into a commercial asset.

For years, the company has spent money, engineering time and mistakes learning how a large software organization works. If some of that knowledge can be packaged and sold, the same investment does two jobs: it helps Spotify run better, then becomes software another company can buy.

That is probably the most useful part of the AWS analogy.

But Spotify's “AWS” does not sell cloud infrastructure

Now the analogy needs to be broken before it becomes too comfortable.

AWS directly sells compute, storage, databases, networking and hundreds of services built on its own infrastructure. Its economics involve data centers, massive capital expenditure and physical scale.

Spotify does not do that.

In its Q2 2026 filing, Spotify says Google Cloud Platform handles the vast majority of its primary data storage and computing.1

Spotify is not turning its servers into AWS.

It is turning the layer above the servers into products.

Backstage organizes services. Portal packages the managed experience. Plugins encode platform practices. Confidence organizes experimentation.

One way to describe the arrangement is that Google sells Spotify invisible machines, while Spotify tries to sell other companies a way to avoid losing their minds while using all those machines.

That is a different market. The barriers are different. The potential ceiling is different too.

So the AWS comparison should not be a prediction of size. It is a comparison of organizational transformation: the point when a capability built for yourself becomes something other companies can buy.

Spotify already has another platform layer inside music

There is a second path that could already matter more financially than Portal or Confidence. Public accounts do not let us compare the two, but this one acts directly on streaming economics.

Spotify does not make money only from listeners and advertisers. It also has mechanisms where artists and their partners give up value, directly or indirectly, in exchange for better access to Spotify's audience.

Discovery Mode is a clear example.

Eligible artists do not pay an upfront campaign budget. Instead, when a selected track generates streams in certain Discovery Mode contexts, Spotify applies a 30% commission to the recording royalties generated by those streams.11

Other streams are not subject to that commission.11

More broadly, Spotify says in its annual report that certain “marketplace programs” contribute to Premium gross-margin improvement by reducing net music royalty costs.2

Those two facts should not be collapsed into one neat equation. Spotify does not publish the financial contribution of each program, and not every marketplace benefit is Discovery Mode.

The direction is still interesting.

The basic streaming model looks like this: Spotify collects money from users, then pays a large part of the economics back into the music-rights ecosystem.

Spotify says it paid more than $11 billion to the music industry in 2025.12 That is Spotify's own figure, but it illustrates why streaming has a constraint that pure SaaS does not: much of the core catalog is licensed from other parties who must be paid.

Marketplace tools partially reverse the flow. Spotify controls the audience, recommendation surfaces and interface. Parts of the music ecosystem are therefore willing to surrender more value for additional exposure inside that infrastructure.

That is a very platform-like move.

It may matter more to understanding Spotify than the narrower question, “how much does Backstage make?”

Spotify's valuable asset may be the infrastructure between different groups

Spotify depends on licensing agreements with rights holders to provide much of the catalog that makes the service useful.2

What Spotify increasingly controls are the layers between participants:

• the interface between listeners and catalogs;
• recommendation systems;
• marketing tools for artists and labels;
• listening and experimentation data;
• the methods its own engineering organization uses to operate at scale;
• now, products that export some of those methods to other companies.

This is where Amazon becomes relevant again.

Amazon started as a retailer. By building infrastructure for itself, it eventually discovered that part of the machinery could become a market of its own.

Spotify is nowhere near that point with developer tools today. Nothing in its accounts supports the claim that Portal, premium Backstage products or Confidence already form a financially large business.

But those products show that Spotify no longer thinks only like a streaming application.

It knows some of the systems built behind the application may be useful enough for other organizations to buy.

What would make the thesis real

The case becomes much stronger if future reports start showing any of the following:

• Spotify publishes revenue or ARR for developer products;
• a third segment or new revenue category appears in the accounts;
• Portal, Confidence or premium plugins disclose meaningful customer volumes;
• enterprise software becomes material enough for Spotify to discuss its effect on margins;
• artist-side “marketplace programs” become more financially transparent.

We are not there yet.

So the conclusion should not be “Spotify found its AWS.”

That is too early, and probably too literal.

The more interesting conclusion is that Spotify has started doing something Amazon did before it: taking infrastructure created to solve its own problems and trying to turn it into an independent product.

Amazon eventually sold the servers and services around them.

Spotify is starting by selling the way Spotify gets built.