Spotify's earnings call pitch: be the music service AI agents choose

Spotify's third quarter brought 788 million monthly active users and nearly 5 billion euros, or roughly 5.6 billion dollars, in revenue. The more interesting part of the earnings call, according to TechCrunch's account, was how management described the future of listening and why the spread of AI agents could work in the company's favor.

Co-CEO Gustav Söderström told investors that Spotify sits in one of the few categories that stands to thrive in the agentic age. Agents, he argued, excel at taking over things people would rather not do, and entertainment is the opposite case: nobody asks an agent to summarize a song.

What changes is the entry point. Instead of opening the app, searching for an artist and assembling a playlist, a listener can ask an assistant for music for a run, a podcast on a topic they were just discussing with an AI, or an audiobook that resumes exactly where they stopped.

Spotify wants to be the service those agents call. It was the first music service to integrate with Meta's Muse, and it also plugs into ChatGPT and Claude. Söderström said the goal is to be the media service people use from any agent, turning the growth of agents into a tailwind rather than a headwind, and that adoption of Spotify inside those agents already looks good.

He was careful about data. Spotify does not hand listening history to partners, he said; the agent asks Spotify for something, and Spotify reasons about the request with its own systems before deciding what to play.

The arrangement, with the service keeping the reasoning step, sits against visible friction elsewhere. Amazon has blocked Meta's Muse from shopping on its platform, a reminder that large players can restrict outside agents when access, security, privacy or commercial control come into question.

Söderström also described Spotify as an intelligent media service encapsulated as an agent, usable by clicking in the app or by talking to it, asking questions and giving instructions. His example: ask Spotify to walk back through listening history to 2008, pick out the true favorites, stretch their tempo, sort them by key and BPM, and mix a running playlist at an exact cadence. Years of that history, including which artists get replayed, which tracks get skipped and what was discovered, is the differentiator against a general assistant that knows little about someone's taste.

Two questions are open. One is which assistants keep expanding their use of Spotify and whether platform-level blocks like Amazon's spread. The other is how Spotify measures and monetizes playback that increasingly happens inside someone else's agent, since no public numbers yet show whether that traffic converts into subscriptions and advertising value.