Spotify closes Q3 at €4.97B and pitches AI agents as a tailwind
Spotify reported Q3 2026 on Oct 22: 306M subscribers, 788M MAU and €4.97B revenue; it calls itself one of few categories that thrive in the agentic age.
Spotify, the music, podcast and audiobook streaming service, reported third-quarter 2026 results on October 22: 306 million paying subscribers, 788 million monthly active users, €4.97 billion in revenue and €672 million of operating income. No new agent feature shipped that day — the three assistant integrations landed in October 2025, April 2026 and September 2026. The news was the framing: co-CEO Gustav Söderström used the release to argue that Spotify sits in “one of the few categories that actually stands to thrive in the agentic age,” and on the call, per TechCrunch, called agent growth “a tailwind rather than a headwind.”
Key points
- Spotify beat three of its five quarterly guidance lines and matched the other two: subscribers at 306 million against a 305 million guide, gross margin at 33.0% against 32.9%, operating income at €672 million against €670 million, with 788 million MAU and €5.0 billion revenue in line. Revenue grew 16% year over year, 14% at constant currency.
- Söderström’s argument rests on what the task is. Agents are good at absorbing chores; entertainment sits on the opposite side. His line: “Nobody asks an agent to ‘summarize a song’ for them.”
- The outside doorways were already built: ChatGPT on October 6, 2025, Claude on April 23, 2026, and Meta’s Muse on September 23, 2026. The Q3 shareholder deck lists the Muse integration under product and platform, described as listening to audio “through an AI agent conversation.”
Background
The three integrations do not carry the same terms. Spotify’s ChatGPT post says connecting is opt-in and “you can connect or disconnect at any time,” and that Spotify will not share music, podcasts or other audio or video content with OpenAI for training. The Claude post says the same about control and training. The Muse post of September 23 describes only capabilities (queue and play audio, resume podcasts, save songs to the library, build playlists, schedule playback) and says nothing about what data moves between Spotify and Meta, or how a listener disconnects.
Two earlier entries cover adjacent ground. Spotify pulling 50 million suspected AI spam tracks was a content-moderation story; Meta’s Muse passing 3.4 million downloads in under three weeks tracked that app’s own growth. This quarter is about the distribution layer: when a listener opens their mouth inside someone else’s assistant, what part of the chain does Spotify still own?
Commerce already ran this experiment. Shopify opening its catalog and Shop Pay to any agent handed over checkout to win agent-driven orders. Spotify is trying to keep the decision about what plays while still accepting the request.
The numbers
The shareholder deck is the reference; the €4.97 billion that Music Ally and Music Business Worldwide report is the same figure as the €5.0 billion in the release, rounded differently. Operating expenses rose 26% year over year (24% at constant currency). Strip out currency and the €22 million of social charges tied to share price, and the increase was 19%, which management attributes to marketing plus cloud and AI spend.
| Metric | Q3 2026 | Y/Y | vs guidance |
|---|---|---|---|
| Premium subscribers | 306M | +9% | above 305M |
| Monthly active users | 788M | +11% | in line |
| Revenue | €4.969B | +16% (+14% cc) | in line at €5.0B |
| Gross margin | 33.0% | +136 bps | above 32.9% |
| Operating income | €672M | +15% | above €670M |
| Operating expenses | €967M | +26% (+24% cc) | — |
| Free cash flow | €735M | -9% | — |
| Diluted EPS | €2.48 | €3.28 a year earlier | — |
By segment, premium revenue reached €4.487 billion (+17%) and ad-supported revenue €482 million (+10%), with ad-supported MAU at 499 million (+12%). Monthly revenue per premium subscriber was €4.95, up 9.3%. Q4 guidance: 818 million MAU, 316 million subscribers, €5.2 billion revenue, 33.8% gross margin and €820 million operating income. Music Business Worldwide adds the nine months reported plus the Q4 guide to project €2.862 billion ($3.32 billion) of operating profit for 2026, up 30% from €2.198 billion in 2025, which would be a record and the third straight profitable year.
What others say
TechCrunch’s Lauren Forristal carried the call quotes the release left out. Söderström said the goal is “to be the media service that you use from any agent” and said Spotify is “seeing really good adoption of Spotify inside these agents.” Forristal also reports that Spotify was the first music service to integrate with Meta’s Muse. Spotify’s own September 23 post confirms the integration but not the “first” framing; that word comes from TechCrunch and from Music Ally’s September 24 headline. Music Ally adds that Meta had brought on Ticketmaster earlier as a music-adjacent connector, which keeps the claim narrow: first streaming service, not first music company.
The mechanism TechCrunch describes is the part to flag: Spotify isn’t sending listener data to partners; the agent asks Spotify for something, and Spotify’s own systems decide what to play. That sentence has one source. Spotify’s three announcements never phrase it that way, and the Muse post is the weakest of the three on data and exit terms.
The financial read differs by outlet. The Hollywood Reporter’s Georg Szalai put operating income €2 million over guidance, tied the 26% expense increase to marketing, cloud and AI, and noted shares up more than 3% premarket. Deadline’s Jill Goldsmith had the stock up 10% on the day to a ten-month high, with management guiding another 10 million net subscriber additions in Q4. On earnings per share, 24/7 Wall St. and Zacks both call it a miss but disagree on the numbers: $2.79 against a $2.85 consensus in the first, $2.85 against $3.10 in the second. Spotify’s own guidance has no EPS line, which is why “met or beat guidance” and “missed analyst estimates” can both be true in the same quarter.
Our take
For anyone running a content or subscription product, this quarter offers a comparison worth copying into a strategy doc. When the front door moves to ChatGPT, Claude or Muse, a platform can either become a set of APIs and let the caller own selection, or keep the request-taking thin and hold recommendation, ordering and listening history itself. Spotify can pick the second option because it is betting that entertainment consumption won’t be outsourced: listening is the thing the user wants to do, and an agent can’t do it for them. Tools don’t get that protection; users will take whatever answer the assistant hands them.
Cost is the other half. Of the three expense drivers management named, two are agent-era infrastructure: cloud and AI. A company with 33.0% gross margin is carrying that build-out on the income statement, and the full-year €2.862 billion operating profit depends on delivering €820 million in Q4, 22% above the €672 million just reported.
Three things stay unresolved in the public material. The “no listener data to partners” architecture rests on a single outlet. The Muse announcement describes neither the data boundary nor an exit path, while the ChatGPT and Claude posts both promise disconnection at will. And Spotify hasn’t said whether agent-driven plays land in the same royalty and ARPU accounting as in-app plays. Those three answers decide whether keeping the decision layer is shipped architecture or a talking point for one earnings cycle.
What to watch
Q4 guidance pins year-end at 818 million MAU, 316 million subscribers and €820 million operating income, against a 33.8% gross margin. In the same window Spotify plans to take audiobooks from 22 markets to more than 180, which changes the content-cost mix and tests whether the agent doorway works as well for books and podcasts as for music. The specific signals to look for next quarter: any segmented disclosure of agent-channel listening, and whether the Muse integration gains the disconnect language its siblings already have.