Spotify Wants to Be the Music Service AI Agents Choose

Spotify reported 788 million monthly active users in its third quarter of 2026 and nearly €5 billion (around $5.6 billion) in revenue for the same period, according to TechCrunch. The same report framed the company's AI strategy around one goal: to be the music service that agents select when a user asks for audio.

Co-CEO Gustav Söderström said Spotify operates in one of the few categories that stands to thrive in the agentic age. An AI agent is an assistant that carries out tasks for a user, such as playing a song on request. His stated objective is to be the media service a user can reach from any agent. He described this as turning the growth of agents from a headwind into a tailwind. He also said Spotify is seeing strong adoption inside AI agents.

The integrations named so far are concrete. Spotify was the first music service to integrate with Meta's Muse assistant, and it works with ChatGPT and Claude. In each case, the agent is the front end that the user sees, and Spotify is the back end that supplies the music.

The architecture matters more to engineers than the headline numbers. Spotify does not send listener data to its partners. Instead, the agent sends a request, and Spotify's own systems reason about it and decide what to play. The recommendation logic, and the behavioral data behind it, stay inside Spotify's stack. Partners receive playback decisions, not the profile that informed them.

The agent push is the latest step in a series of AI deployments the company has described publicly over the past year. In February, Spotify said its best developers had not written a line of code since December, and it credited AI. That same month, it rolled out AI-powered Prompted Playlists to the UK. In March, it announced a beta feature that lets listeners review and edit their Taste Profile, the model of listening habits that drives recommendations. In August, it added Merlin to its AI music remix and covers effort.

At its Investor Day 2026, which the company's newsroom documented in remarks from Co-CEOs Alex Norström and Gustav Söderström, Söderström said Spotify invested early in machine learning as a driver of retention. That position explains the current emphasis. Recommendation quality has long been the mechanism that keeps listeners engaged, and the company is now trying to extend that mechanism to requests that start outside its own app.

The financial picture has a cost side. In August, Reuters reported that Spotify had forecast third-quarter profit below estimates and said higher marketing and development costs would weigh on profit in that quarter. The revenue and user figures reported this week therefore sit alongside an explicit warning that spending was rising.

In my view, the central question for the agent strategy is where the recommendation decision lives. If agents route music requests to Spotify and Spotify reasons over its own data, Spotify gains distribution without giving up the profile that makes its recommendations valuable. That differs from an arrangement in which a partner receives listener data and builds its own experience on top. Whether other services adopt the same model, and whether agent platforms accept it as the default, are open questions.

The broader context here is that the same architecture could make a listener's taste profile more portable across assistants. That would help users who currently rebuild their preferences in each new tool. For developers, the more immediate interest is the request-and-reason pattern itself. It keeps the model and the data in one place while exposing a narrow interface to third parties. That design is worth watching as more consumer services try to become callable components inside agent workflows.