METAM

Below is Dolphin Research's recap of the $Spotify(SPOT.US) 26Q3 earnings call Trans

I. Key takeaways

1. Shareholder returns

a. From Jan to Oct 21, the company repurchased 2.2 mn shares (~$1.1 bn), offsetting full-year SBC dilution and starting to return excess cash. b. Since buybacks restarted in Aug 2025, total repurchases reached ~3.0 mn shares, about 1.5% of float. c. Capital allocation remains unchanged: prioritize organic and inorganic growth first, then opportunistic shareholder returns. Cash and equivalents stood at €9.9 bn at quarter-end with no debt excluding lease liabilities.

2. Q4 guidance

a. Users: MEU of 818 mn, net adds of 30 mn QoQ, below last year's record pace driven by the new Free tier. Subscribers of 316 mn, net adds of 10 mn. b. Revenue of €5.2 bn. ARPU up 4.5%–5% YoY on recent price hikes and add-ons like Audiobooks+, partly offset by the high base from price increases in 150+ markets last year. Ads continue to trend toward double-digit growth. c. GPM of 33.8%, up ~70 bps YoY, factoring upfront investments to expand audiobooks to 180+ markets and the Spotify Partner Program to 50+ markets. This implies full-year GPM is ~130 bps higher vs. 2025. d. OP of €820 mn.

3. Expense and margin cadence

a. Opex growth peaked in Q3 and is set to decline meaningfully from Q4, with the trend extending through 2027. b. No guidance for 2027. Margin will fluctuate by quarter due to seasonality and investment cadence, but management expects steady progress toward the 2030 target in 2027.

4. This quarter's key metrics

a. Total: revenue ~€5.0 bn, +14% YoY at constant FX. OP of €672 mn, slightly above guidance, with OPM of 13.5%. GPM of 33%, 10 bps above guidance and up ~140 bps YoY. b. By segment: Premium revenue +15%, driven by +9% subs and ARPU uplift. Ad revenue +9%, led by biddable channels. c. One-offs: social security expense of €22 mn, €13 mn above forecast due to stock volatility. Excluding this, OP beat guidance by €15 mn, aided by marketing and personnel savings and stronger GPM. d. Cash flow: FCF of €735 mn this quarter, weighed by working capital and cash tax timing. LTM FCF of €3.2 bn.

II. Call details

2.1 Management highlights

1. Sustained subscriber growth

a. At IPO in 2018, subs were 71 mn and have added 25 mn+ per year since, expected to reach 316 mn in 2026. All regions grew this quarter, with beats led by RoW and LatAm. b. Paid penetration: Sweden near 50%, global under 4%. UK and Germany at 20%–25%, LatAm ~11%, and under 3% in EMs. Mature markets will keep deepening a sticky paid base, while EMs still have hundreds of millions of sub opportunities. c. Content: signed a new multi-year deal with Joe Rogan. On covers and remixes, brought in Believe and TuneCore as partners and will advance remix features under a 'licensed, credited, revenue-sharing' framework.

2. AI agents and 'everywhere' strategy

a. In Feb, the company set out to build the world's first truly intelligent agented media platform: Spotify itself acts as a conversational, commandable agent, usable in-app and callable by other agents. b. It has integrated early with ChatGPT, Meta Ray‑Ban glasses and OpenAI devices, and was the only media service on Meta Muse at launch. The logic parallels the eSDK a decade ago that made smart speakers and other connected devices tailwinds. c. Business logic: Spotify can monetize background listening via Premium or ad-supported audio. Running in the background on Waze and Google Maps for years is the precedent. Agents handle chores well, while entertainment is the opposite—no one asks an agent to 'summarize a song'.

3. Interactivity and the Large Taste Model

a. About one year after launch, nearly 40 mn paid users listened to mixes. Consumption rose ~90% QoQ, and millions of users build mixes monthly. b. Mix data down to the millisecond shows how tracks should connect, feeding the Large Taste Model, recommendations and Spotify agent capabilities. This improves retention. c. Early deployment of the Large Taste Model has driven double-digit gains in content discovery, with longer listening time in some algorithmic playlists.

4. Audiobooks and value-add subs

a. In Sweden, audiobooks are nearly as large as music, which management views as 'early behavior' rather than a local anomaly. Audiobooks will expand to 180+ markets by year-end, making Spotify the first truly global audiobooks service. The goal is market expansion, not share grabs. b. The Spotify Partner Program added 35 markets this fall. Creator memberships launched two days ago, letting superfans pay creators directly, echoing the 'power law' theme from Investor Day. c. In markets with the longest runway for audiobooks and SPP, these have contributed to acquisition, retention, DAUs and margin improvement.

5. Returns on spend

a. Marketing has raised consumer perception of product value, supporting future sub growth and monetization. b. AI investment, with headcount flat, has accelerated feature velocity.

2.2 Q&A

Q: After adding new frictions to the Free tier in markets like India and Indonesia, how are early results? How should we think about paid conversion?

A: These changes are positive for Free, Premium and revenue, with no quantified conversion disclosed. Changes include improving data quality in sign-up flows, adjusting ad load and migrating low-end device clients to better-retaining clients to lift LTV. Management attributes long-term sustained growth to such optimizations. India has 1.4 bn people and GDP growth of 7%–8% for years. Management said there is no better market.

Q: Buybacks have accelerated. With FCF rising, how are you thinking about capital allocation and could buybacks speed up further?

A: Priorities unchanged—organic first, M&A second, excess cash to buybacks, with no commitment to further acceleration. M&A cases include WhoSampled (powering song info features now reaching nearly 100 mn users), Findaway (accelerating audiobooks) and Sonantic (supporting AI DJ). On buybacks, the plan to offset ~1.0 mn shares of SBC dilution is complete. Total repurchases now stand at 2.2 mn shares at just over $1.0 bn. The CFO added that the firm will keep a strong balance sheet and cash to preserve strategic flexibility, while returning excess cash as compounding FCF accumulates, which began this quarter.

Q: When users listen via third-party AI agents like Muse, who owns the customer relationship and data? How do you avoid becoming a back-end utility?

A: Listening does not shift away. Third-party agents merely 'remote control' Spotify in the background, and data loops back positively. The company does not send music or data to partners. Requests come from partners to Spotify, and Spotify infers and decides what to play based on its own user knowledge. For example, if a user asks an agent in Muse to generate a podcast, it is published on Spotify and visible only to the user, giving Spotify otherwise-unavailable interest data. The approach mirrors the eSDK era when any speaker OEM could easily integrate: rebuild Spotify as an agent so other agents can call it 'agent-to-agent'. This underpins renewed investment in the Large Taste Model, time‑stitching, mixes and covers. Management said adoption of Spotify inside agents is strong, and while many pull back, Spotify is leaning into agent growth.

Q: With the global rollout of audiobooks, fitness, expanded SPP and US Taste Profile in H2, will 2027 monetization rely on new tiers, add-ons or price?

A: Too early to specify 2027. The framework is 'add value, then monetize'. First, raise Free tier value; the new Free tier launched a year ago has worked for 12–18 months. Second, enrich Premium; management said the value added to Premium this year is unprecedented for the company, with 'power law' products for niche audiences on top. Only then monetize via ads, price, add-ons, more subs and higher active days.

Q: How did ads trend in Q3 vs. the expected H2 inflection, and what drives Q4 revenue guidance for ads?

A: Ad trends are tracking expectations, driven by the rebuilt ad tech stack, with no separate Q4 ad growth disclosed. The three constants for advertisers are brand affinity, high-quality licensed content and high-quality usage contexts. This includes not just time spent but dayparts, context and active days. Internally, the mandate is to make inventory easier to buy: simpler automated access, better measurement, broader audio ad demand and lower creative production friction. These efforts are underway and support Q4 guidance.

Q: How will AI and cloud investment evolve, and how does it align with scaling user-facing AI?

A: Spend is set to decline from Q4 and through 2027 as planned, with returns already visible in discovery and listening time. The Large Taste Model is early in rollout, already driving double-digit discovery gains and poised to add billions of listening hours annually, with further iteration to come. On AI features, tens of millions already converse with Spotify in natural language, and mixes are scaling fast. On costs, token prices at model vendors have dropped sharply, and the company's routing platform orchestrates across multiple models, including open source, keeping costs in check. On R&D efficiency, beyond doubling lines of code and PRs, over two six‑month cycles the pace of end-user delivery has accelerated markedly. This lifts interaction, retention, active days, discovery, saves and listening time. Costs will remain disciplined, with investment when opportunities arise.

Q: With ads recovering, could ads drive overall revenue growth by 2027?

A: Once ads are in double-digit growth, they will lift overall revenue, though not explicitly tied to 2027. Drivers include the biddable channel, now at 40% share, and podcasts. Ads are about 10% of revenue, and double-digit growth vs. low single digits provides a lift to total revenue.

Q: You started selling internal dev tools externally. What's the logic and how big is the opportunity?

A: Treated as an option value for now, with no further guidance. The company invested to stay at the forefront of agented development, and others asked to use these tools. Since costs will be incurred anyway, spreading them over more developers, not just the company's thousands, improves cost amortization.

Q: Family plan anti-sharing measures for non-cohabiting users—where are they enforced, when will it be done, and what's the financial impact?

A: Not a 'crackdown' but routine verification improvements, with no timeline and no financial impact disclosed. The Family plan has been around for ~12–13 years and is a big share of subs. As loopholes are found, verification keeps improving, and this is a long-term routine effort.

Q: After nearly a year using tools like Claude Code, how do you view AI ROI, and what does it imply for investment levels?

A: AI returns are seen in faster delivery and monetization, and investment will roughly hold at current levels. Early indicators like lines of code and PRs are visible, but the company has completed two six‑month cycles, with significant acceleration in end-user features in the latest cycle. These features boosted adoption, interaction and retention, and audiobooks directly lifted ARPU. The company will keep evaluating ROI, invest when opportunities arise and communicate as before.

Q: With no debt and nearly €10 bn cash, how will you create shareholder value with this cash?

A: Consistent with prior stance: prioritize organic and inorganic, and return excess cash—no change. A strong balance sheet provides strategic options and flexibility. Given the cash balance and compounding FCF, the company will return excess cash while staying opportunistic on investments.

Q: Will AI agents cannibalize opportunities for Studio by Spotify or assistant-like bespoke podcast products?

A: Management sees this as a tailwind. Studio was built in anticipation of the agent era and the opportunity is 'very large'. The company first ships agented features on desktop for frontier users, like generating podcasts, then builds skills for third-party agents so users can generate Spotify content with personal context outside the app. This increases Spotify's value. In-app creation remains available. Management likens this to 'Connect for the agent era': users can access Spotify from any third-party interface, but playback remains controlled and understood by Spotify. The shift toward personalized content is clear.

Q: Expanding audiobooks from 22 to 180+ markets—what is the margin impact, and can early-market positives be replicated?

A: Positive effects mirror early markets. Margin is said to progress toward Investor Day targets, without quantification. The journey began about three years ago, when heavy listening to uploaded audiobooks in Germany was discovered. Since then, audiobook listeners rose 40% pre-global rollout, with time spent up in sync. The core impact is higher active days, improving retention and LTV. The audiobooks add-on shows strong early performance.

Q: With agents like Muse and Instinct taking over the UI, could they price-compare, switch or cancel less-used subs, making Spotify a replaceable back-end? Is churn from agents a real risk?

A: Integration is a growth driver and churn risk is low—Spotify does not rely on users 'forgetting to cancel'. Spotify has been embedded in Google Maps and Waze for years and kept quiet to avoid tipping off competitors, resulting in major growth rather than commoditization. The prerequisite is differentiated data and experience, and management is confident here. On cancellations, agents help clear 'unused subs', but Spotify has 100 mn+ users active roughly 28 days per month. Surveys show almost no one is subscribed 'unwittingly'. After 20 years and a financial crisis, management believes Spotify will be among the last services users cancel, and agents could highlight Spotify's superior value.

Q: How is demand for Taste Profile, Prompted Playlist and other new features? After adding value, how do you balance pricing vs. sub growth?

A: Use a 'value-for-money' lens: add value first, then raise prices, letting users get a better value-to-price over time. For the first 15 years, investment focused on product and content value, and only after reaching scale did price increases begin, with proceeds reinvested into the product. The company remains on this path. On demand, About the Song users are approaching 100 mn. Tens of millions converse with Spotify, and tens of millions listen to agented playlists.

Q: UMG and WMG deals will be two years old in Q1 2027. How will 2027 pricing and add-ons affect renewals, and will new deal structures match the current ones?

A: No comment on specific partner contract terms. The two-year renewal cadence spans the company's history. Management says incentives are increasingly aligned with labels, relationships are improving and it is easier to reach win‑wins.

<End of text>

Risk disclosure and statement:Dolphin Research Disclaimer and General Disclosure