Spotify Co-CEOs outline AI ambitions and more on Q1 2026 earnings call
AI was a recurring theme on Spotify's Q1 2026 earnings call on Tuesday (April 28). The company's leadership team was grilled by analysts on everything from whether Spotify would consider launching an AI music creation tier, to how it justifies rising compute costs, to whether standalone AI music platforms could one day compete with Spotify itself.
Co-CEO Gustav Söderström, who took the reins alongside Alex Norström at the start of 2026 following Daniel Ek's move to Executive Chairman, used the call to frame the current moment as one of the most significant in the company's history – comparing the AI opportunity to the launch of the iPhone and App Store.
Söderström also reiterated his push for AI-generated "derivatives" of existing music – a subject he first addressed in detail on Spotify's Q4 2025 earnings call in February, when he described derivatives as "an untapped opportunity for artists to make money off of their existing IP."
That call came a few months after Spotify announced partnerships with all three major music companies, Merlin and Believe to develop "artist-first" AI music products.
The AI discussion came alongside continued financial momentum and subscriber growth. As reported by MBW earlier today, Spotify posted total revenue of €4.5 billion (USD $5.30bn) in Q1, up 14% YoY at constant currency, with operating income of €715 million ($837M) and an operating margin of 15.8%. Monthly active users grew to 761 million, while premium subscribers reached 293 million
Spotify's shareholder deck revealed that cloud and AI spend was one of the key drivers of a 17% underlying year-over-year increase in operating expenses in Q1 (alongside higher marketing costs). The reported figure showed operating expenses declining 5% YoY to €780 million, but that was flattered by currency movements and a €49 million social charges benefit.
Analysts pressed Söderström on the increased spend. He said: "The way to think about it is we have not increased our headcount, actually we slightly decreased our headcount, but we are spending more compute per employee. And that is because we're seeing tremendous return in terms of productivity." (The company's headcount stood at 7,258 at quarter end.)
On what the compute spend covers, Söderström said Spotify is both using AI tools to accelerate its development pace and "training rather large models in-house, because we have lots and lots of unique data that no one else has."
He added: "Some of it is training cost, and that's upfront. And we'll capture that value when those products roll out."
To justify the investment, Söderström compared the current moment to the launch of the iPhone and App Store: "I usually make the analogy to 2009 when the iPhone came up, and the App Store came [out], and believe it or not, I was actually here back in 2009, so I lived through that. It was a time of tremendous opportunity. Some people sat around and waited. Spotify did not, we took the opportunity."
He continued: "We think this opportunity is as big or possibly bigger. So we're taking that opportunity. But we are very diligent and very disciplined about those investments," he said.
Söderström pointed to early results from AI-powered features. Spotify's AI DJ is now used by 94 million subscribers, closing in on 100 million and "driving billions of hours of engagement." Song DNA, released just four weeks before the call – had already reached 52 million users.
Asked whether Spotify believes in an AI music creation tier, Söderström used the question to reiterate his argument, first made on Spotify's Q4 2025 earnings call in February, that Spotify sees an opportunity that "no one is addressing right now" for existing artists and AI.
He said: "What we do believe in is that there is a lot of opportunity out there for creators who want to use AI tools, but there is an opportunity that no one is addressing right now for existing artists. And we really want to address that part. We don't think existing artists should be left out of AI. We think that may actually be the most interesting part of music."
"We have the capabilities and technologies we need. We are the right company to solve this problem."
Gustav Söderström
He drew a comparison with other industries: "If you look at other industries, existing IP is actually the most valuable IP, not the least valuable. But because of [how] AI music works right now, that is not addressable. That's the problem we want to solve."
In a separate answer on derivatives, Söderström elaborated: "Right now, existing creators are largely left out of the AI opportunity altogether… That's because the copyright problem is much more complicated to solve well, and the attribution problem of who should get paid what is much harder. But we love hard problems. So that's the problem we want to go after."
"We have the capabilities and technologies we need. We are the right company to solve this problem," he added.
In September 2025, Spotify relaxed its longstanding shuffle-only restriction for mobile free users, allowing them to manually select and play tracks from playlists and albums for the first time. The overhaul also introduced search-and-play and share-and-play features, in what Spotify's VP of Markets and Subscriptions, Gustav Gyllenhammar, told MBW at the time was driven in part by Gen Z retention challenges with the old mobile experience.
On the Q1 call, Norström said the impact has exceeded expectations. He said Spotify has seen a "step change" in active days per month among free-tier users since the launch. "It's just blown up my expectations fully since we launched this last summer," Norström said.
However, the gap between free tier engagement and ad revenue remains a challenge.
Ad-supported revenue grew approximately 3% YoY in Q1, well behind premium revenue growth of approximately 15%. CFO Christian Luiga noted that Q1 was "one of the few times we've had a negative development on the year-over-year gross margin on the ads business," but framed it as a short-term dynamic driven by engagement outpacing monetization.
Norström characterized the broader ad business transition as a necessary rebuild: "About 1.5 years or 2 years [ago] almost, we observed that there was a gap… essentially, we saw us missing out on a TAM where people were putting a lot of money. And this [area] was programmatic, it was automated sales and it was biddable exchanges."
"The decision we made back then was a pretty tough one because we had to essentially rebuild the entire stack… Now that transition is done. So now it's about execution," he said. Spotify's biddable channel now represents more than a third of ad revenue. Luiga reiterated that the company expects ad revenue growth to pick up in the second half of 2026.
In November 2025, Spotify launched a revamped subscription structure in India, Indonesia, Saudi Arabia, South Africa and the UAE – introducing three tiers (Premium Lite, Premium Standard and Premium Platinum) in place of its single Premium offering. In India, the top-tier Premium Platinum was priced at more than double the cost of the previous standard subscription.
Asked about these tests on the Q1 call, Söderström said: "This is one of my personal favorites… the early indications is that when we deploy these types of value proposition frameworks, we do see a structural increase in ARPU."
Söderström laid out a broader monetization framework: "Spotify always had the business model to capture the long tail, which requires a free tier, and to capture a bunch of the averagely engaged users in premium. But until we launched [the] audiobooks add-on, we didn't really have a tool to capture the people who want to read for hundreds of hours a month. We had a clear theory that we could capture the entire power law, but we haven't proven it to ourselves until recently. Now we have those three tools."
The firm's Premium monthly ARPU stood at €4.76 ($5.57) — up 5.7% YoY at constant currency.
The final question of the call came from analyst William Packer, who noted that "investor concerns over AI disruption have increased" and asked Söderström to outline Spotify's key "moats" ie defenses against three specific threats: standalone low-cost AI music alternatives, large platform peers offering free AI music services, and competition from AI-first alternatives that integrate label content.
Söderström pushed back on the premise: "I don't really like the word moats. I think there are fair advantages, possibly that you've earned because you worked really hard."
His central argument was that Spotify's nearly 20 years of listening data gives it an advantage that cannot easily be copied, because understanding user taste is fundamentally different from the kind of factual knowledge that AI can commoditize.
"Some of the things in the world are facts that can be easily commoditized by LLMs, such as the capital of Texas or something," Söderström said.
"Other things are not so easily commoditized. And it turns out luckily for us, taste is not easily commoditized because it's not a fact, it's an opinion. It differs between people, it differs between regions, it differs between people in those regions, and use cases. And on top of that, it changes weekly."
Söderström said Spotify is investing in a proprietary "large personalization model," which the company internally calls a "taste model" – trained on data from its 700 million-plus users.
He expressed confidence in the durability of that advantage: "If you theoretically said that someone could somehow snapshot all our user data, all 700 million-plus users, that, they could train a model on that and then that model is pretty useless after about maybe 2, 3 weeks as culture moved on. So we actually think it's very sustainable, and you need to be at scale to keep these models valuable."
Key Insights
- This topic is currently trending
- Experts are closely monitoring developments
- It may impact future decisions
