← back to rankingSPOT · Spotify Technology S.A.
Internet Content & Information · mkt cap $99.5B · calls: Q1 FY2026 vs Q4 FY2025
45.0 conviction · conf-adj 45
conf 4/10 partial
enthusiasm:27.0 · trend:8 · quantifies:0 · impact:0 · under_radar:5 · credibility:0 · business_impact:8 · disruption:-6 · commitment:0 · confirmation:3
Enthusiasm latest 9 / prev 8 (rising)
Spotify frames AI as compounding its historical personalization moat—now via natural-language intent, in-house taste models, agentic/interactive product (DJ, Prompted Playlist, Taste Profile), and faster internal shipping (compute up, headcount flat)—with rising enthusiasm from Q4 to Q1 and more hard user metrics (94M DJ, 52M Song DNA in four weeks). Management does not attribute revenue or margin dollars directly to AI; the credible chain is usage→retention→LTV, plus a strategic bet on rights-cleared derivatives for existing artists rather than competing as a generative-AI DSP.
GROUNDED NEXT-FY IMPACT vs CONSENSUS
Grounded on actual base — revenue $17.2B · net income $2.2B · net margin 12.9% · diluted EPS 10.51
These are next-fiscal-year annual uplift estimates, not next-quarter numbers.
Aggregate next-FY est. rev uplift: % · next-FY EPS uplift: % · vs analysts: inline · priced in: medium (model's call-read: high; verdict above is the hard-data one used for ranking) · confidence: 4/10
| Claim | Figure | Arithmetic | Next-FY Rev % | Next-FY EPS % |
|---|
AI DJ (IDJ) reach 94M subscribers (Q1 FY2026)
engagement · soft | 94M subscribers; 'billions of hours' | Engagement reach, not revenue. 94M/~700M DAU = 13.4% penetration shows scale but Spotify Premium is a flat per-sub fee, so marginal listening hours do not convert to incremental revenue at any disclosed rate. No IDJ→paid conversion %, ARPU lift, or churn-delta disclosed anywhere in the claims → no $ attachable without inventing a take-rate. | | |
Song DNA 52M users in 4 weeks
engagement · soft | 52M users in 4 weeks | Adoption velocity only: 52M/700M = 7.4% of DAU in first month. No monetization, incremental ARPU, or churn figure disclosed → cannot translate to revenue without inventing a per-user value. | | |
Jam usage doubled YoY; >100M monthly listening hours
engagement · soft | 2x YoY; >100M monthly listening hours | Doubling implies prior ~50M hrs/mo, so incremental ≈600M hrs/yr. But flat-fee subscription = no revenue per marginal hour; free-tier hours could carry ad inventory yet no ad-RPM or hour-split disclosed → no $ attachable. | | |
Headcount decreased 65; flat ~3 years
cost · soft | -65 FTE; headcount flat ~3 years | Anchored as a COUNT (65) but NO fully-loaded comp-per-FTE appears in any claim, so the $ saving base is genuinely undisclosed and not derivable — cannot compute after_tax_saving=65*comp*(1-0.21) without inventing comp. Directionally flat headcount on consensus +14.7% 2026 revenue is operating leverage, but that is already inside consensus' ~+80% EPS jump. → soft on $ magnitude. | | |
Internal productivity (DODs/bets doubling)
productivity · soft | 'doubling', not ~10% | Internal throughput proxy with no baseline count, engineering share of opex, or $ value per unit → no opex saving or revenue attachable. | | |
Music catalog 2M → ~250M tracks
other · soft | ~2M (2008) vs ~250M tracks | 125x catalog scale; context for recsys/AI moat. No revenue or cost $ linkage. | | |
IDJ ~90M subscribers; >4B hours (Q4 FY2025)
engagement · soft | ~90M subs; >4B cumulative hours | Same flat-fee logic as Q1 IDJ claim, earlier period; 90/700 = 12.9% DAU penetration. 4B hours is cumulative engagement, not revenue; no per-hour or retention-$ rate disclosed. | | |
50M mixed playlists; >1M transitions/day
engagement · soft | 50M playlists; >1M transitions/day | Interactivity/data-flywheel milestone (~365M transitions/yr). No $/transition or attach to subscription/ad revenue disclosed. | | |
U.S. streaming hours per user +20% in 5 years
engagement · soft | >20% streaming hrs/user / 5yr (U.S.) | Implied ~3.7%/yr engagement growth, U.S.-only, multi-driver (not isolated to AI), and on a flat-fee model does not equal revenue growth. Supports retention but no revenue $ derivable against $17.186B consolidated rev. | | |
>50 new features shipped FY2025
productivity · soft | >50 features in FY25 | Output/ship-count milestone only; no attach rate, ARPU, or revenue/cost per feature. | | |
Taste model stale after ~2–3 weeks (moat)
other · soft | model useless after ~2-3 weeks | Qualitative data-recency moat argument; no quantified churn, LTV, or share-gain $. | | |
700M+ people using platform daily
other · soft | 700M+ daily users | Scale/context for personalization: $17,186M FY25 rev / 700M DAU = $24.55 rev per DAU/yr (descriptive). No incremental user or ARPU delta stated. | | |
Assumptions: FY25 base: rev $17.186B, GAAP NI $2.212B (12.87% net margin), GAAP diluted EPS $10.51. EPS-uplift basis would use consensus/adjusted NI ~$1.388B and adj EPS $7.17 (GAAP is elevated by one-offs vs Street non-GAAP), per the distorted-base rule. Tax on cost saves = 21%; incremental net margin would default to 12.87% IF a revenue claim existed — none does, so it is never applied. No phasing needed (no bookings/revenue $ targets). All claims map to the core Premium/Ad-Supported model, which is flat-fee per sub, so engagement hours do not mechanically convert to revenue. The only revenue-relevant AI lever (good/better/best tiering → 'structural ARPU increase') is explicitly forward-looking and 'too early for specifics', so it carries no number. Supplier-side AI revenue: none (Spotify is purely an adopter).
Top line: No quantifiable incremental topline. Spotify's AI disclosures are entirely engagement/adoption metrics (IDJ 94M subs = 13.4% of 700M DAU, Song DNA 52M, Jam ~+600M incremental hrs/yr, +20% U.S. hrs/user) with no disclosed ARPU lift, take-rate, or churn-delta — and the flat-fee-per-subscriber model means marginal hours are not marginal dollars. Result: est_rev_uplift_pct=null — unquantifiable from disclosure rather than a computed zero.
Bottom line: No quantifiable incremental bottom line. −65 headcount and ~2x internal DOD/bet velocity lack any $/FTE or $-per-unit base in the transcript, so after_tax_saving is not derivable without inventing comp. Illustratively, at consensus NI ~$1.388B a hypothetical $50M pre-tax save would be only ~2.8% EPS (100×$50M×0.79/$1,388M) — but no such figure is disclosed, so est_eps_uplift_pct=null. Flat headcount on consensus +14.7% 2026 revenue is operating leverage already embedded in Street's ~+80% adj-EPS jump.
[impact n/m (all claims soft/unanchored)] Zero dollar-quantified AI claims: every disclosure is engagement/adoption scale or a qualitative moat, and the flat-fee model blocks hours→revenue conversion, so no hard rev_uplift_pct or eps_uplift_pct is derivable on the $17.186B / ~$1.388B adjusted base. Consensus already models strong growth (FY26 rev ~$19.72B, +14.7%; FY26 adj EPS ~$12.89 vs FY25 $7.17), so the operating leverage from flat headcount and AI-driven productivity is already inside Street numbers → priced_in high, inline vs expectations.
MODEL CONSENSUS (impact)
partial
Both agree all 12 claims are soft/engagement with no $ bridge; merged richest basis. Diverge only on aggregate encoding: null where unquantifiable, 0 only where definitively absent.
Conflicts reconciled
- est_rev_uplift_pct: X=0 vs Y=null -> used null because no dollar-quantified adopter claim exists to sum; unquantifiable from disclosure is not the same as a computed zero
- est_eps_uplift_pct: X=0 vs Y=null -> used null, same reasoning (no anchored $ saving or revenue base)
- supplier_rev_uplift_pct: X=0 vs Y=null -> used 0 because Spotify is purely an adopter, so supplier AI revenue is definitively absent, not merely unknown
| Field | Opus 4.8 | GPT-5.5 |
|---|
| Rev uplift % | – | – |
| EPS uplift % | – | – |
| Priced in | high | – |
| vs analysts | inline | – |
| Confidence | 4 | – |
| Top line | No quantifiable incremental topline. Spotify's AI disclosures are entirely engagement/adoption metrics (IDJ 94M subs, Song DNA 52M, Jam >100M hrs, +20% U.S. hrs/user) with no disclosed ARPU lift, take-rate, or churn-delta to convert into revenue — and the model is flat-fee per subscriber, so marginal hours are not marginal dollars. The only revenue-relevant AI lever (good/better/best tiering -> 'structural ARPU increase') is explicitly forward-looking and 'too early for specifics', so it carries no number. Result: est_rev_uplift_pct = null, not zero-by-conviction but unquantifiable from disclosure. | – |
| Bottom line | No quantifiable incremental bottom line. The cost story is real (headcount -65 and flat ~3 years while productivity 'doubles') but management disclosed no per-FTE comp base anywhere, so 65 FTE cannot be turned into an after-tax $ saving (65 * comp * 0.79) without inventing comp. Sizing against the thin/depressed adjusted base (~1.39B) would also inflate any EPS% as an artifact. So est_eps_uplift_pct = null. Qualitatively this is an operating-leverage thesis already embedded in consensus' assumed margin expansion. | – |
| Reasoning | Consensus already assumes the exact flat-cost/operating-leverage outcome the AI claims describe: revenue 17.19B->19.72B (+14.7%) for 2026 and ->22.55B (+14.4%) for 2027, while EPS jumps 7.17->12.89 (+79.7%) into 2026 and ->15.95 (+23.8%) into 2027, with net income 1.39B->2.83B->3.58B. That ~80% 2026 EPS step on ~15% revenue growth IS margin expansion from flat headcount + rising engagement — i.e., the AI/productivity benefit is already in the numbers. Because no AI claim carries a dollar figure that would let me show math ABOVE this trajectory, there is no demonstrable un-priced upside; the engagement metrics corroborate consensus rather than beat it. | – |
Rows highlighted where the two models disagreed.
QUANTIFICATIONS
AI DJ (IDJ) subscriber reach: 94 million subscribers (Q1 FY2026, both)
“IDJ is now used by 94 million subscribers, closing in on 100 million, driving billions of hours of engagement.”
Song DNA adoption: 52 million users in 4 weeks (Q1 FY2026 (feature ~4 weeks old at call), both)
“Also something we released only 4 weeks ago, Song DNA, is now up to 52 million users, in just 4 weeks.”
Jam social listening: doubled year-over-year; exceeds 100 million monthly listening hours (Q1 FY2026, both)
“features like Jam, where usage has doubled year-over-year and now exceeds 100 million monthly listening hours.”
Headcount change: decreased 65 people; headcount roughly flat ~3 years (Q1 FY2026 / prior quarter, bottomline)
“we have not increased our headcount, actually we slightly decreased our headcount”
Internal productivity proxies (DODs/bets): doubling (not ~10%) (Q1 FY2026 current run-rate, bottomline)
“And all of these keep increasing. And they're increasing several times. They're not increasing 10%. They're increasing -- they're doubling, that kind of increase.”
Music catalog scale (context for recommendation/AI): ~2 million tracks (2008) vs ~250 million now (Historical to Q1 FY2026, both)
“When I joined, I think the music catalog was about 2 million tracks, and now something like 250 million tracks.”
AI DJ (IDJ) subscriber reach and engagement: ~90 million subscribers; over 4 billion hours (Q4 FY2025, both)
“About 90 million subscribers have used IDJ so far, driving over 4 billion hours of time spent on Spotify, and this keeps growing.”
Mixed playlists / user interactivity data: 50 million mixed playlists; more than 1 million transitions per day (Q4 FY2025, both)
“We recently hit a milestone of 50 million mixed playlists and listeners are now making more than 1 million transitions per day”
U.S. streaming hours per user: grown more than 20% in the last 5 years (Q4 FY2025 (U.S. only), topline)
“monthly streaming hours per user have grown more than 20% in the last 5 years”
2025 product ship count: more than 50 new features (FY2025, both)
“In 2025, we launched more than 50 new features and innovations”
Taste-model staleness (competitive moat argument): model pretty useless after about 2–3 weeks without ongoing scale data (Q1 FY2026 (theoretical), both)
“that model is pretty useless after about maybe 2, 3 weeks as culture moved on”
Platform scale for personalization models: 700 million plus people every day (Q1 FY2026, both)
“You literally need 700 million plus people every day using the platform to be able to say what is trending in a certain region in India right now.”
PAST (realized)
- Q4: Daniel Ek — Echo Nest acquisition (2014) gave personalization core to everything.
- Q4: Gustav — Sonantic acquired 2022 after seeing 2021 human-level AI potential; IDJ launched 2023.
- Q4: Gustav — Rebuilt company ~2 years ago for agentic age; Honk system for Claude-driven mobile dev workflow.
- Q4: Alex — Investments into personalization and AI are paying off in engagement/retention formula.
- Q1: Gustav — Machine learning and personalization long core from discovery to recommendations.
- Q1: Gustav — When joined (2008) catalog ~2M tracks vs ~250M now—catalog growth not new.
- Q1: Alex — Organization resizing years ago; no employee growth ~3 years; -65 headcount last quarter.
- Q4: Gustav — 2025 launched 50+ features including Prompted Playlists, Page Match, About the Song.
CURRENT (now)
- Q1: Global rollout of more personalized free tier increasing days per month in key markets.
- Q1: Song DNA and About the Song introduced; Song DNA 52M users in 4 weeks.
- Q1: IDJ used by 94M subscribers; Taste Profile in beta; expanded Prompted Playlist to podcasts.
- Q1: Jam usage doubled YoY, exceeds 100M monthly listening hours.
- Q1: Integrating AI across Spotify; shipping faster, lower cost per feature; inference costs in OpEx.
- Q1: Headcount flat/down, compute per employee up; DODs/bets board metrics doubling.
- Q1: ChatGPT and Claude integrations; capturing plain-language intent data; training in-house taste/personalization model.
- Q1: Christian — Q2 OpEx reflects R&D on strategic AI initiatives already driving engagement.
- Q4: IDJ ~90M subscribers, 4B+ hours; Prompted Playlist taken off with power users.
- Q4: 50M mixed playlists; 1M+ transitions per day; monthly U.S. streaming hours per user +20% over 5 years.
- Q4: Engineers using Honk/Claude on commute for iOS changes before office.
- Q4: Working with industry on creation-metadata and spam/fraud for AI-upload scale.
FORWARD (guidance)
- Q1: Investor Day May 21 to show next chapter; Gustav — more to share on what they're building.
- Q1: Gustav — derivatives of existing IP for existing creators (rights/attribution problem to solve).
- Q1: Christian — Elevated OpEx next quarter or two; confident in LTV returns from AI R&D.
- Q1: Gustav — iPhone 2009-scale opportunity; disciplined investment with usage/retention payoff.
- Q1: Gustav — Agentic moment unlocking new growth vectors; climb to new mountains.
- Q4: Gustav — World's first truly intelligent agentic media platform; more at Investor Day.
- Q4: Gustav — Work with industry on artist monetization of catalog via AI derivatives (rights framework).
- Q4: Gustav — Can increase shipping pace further via technology evolution.
- Q4: Gustav — Planning more important in AI era so agents stay utilized.
- Q1: Gustav — Early tiering tests (good/better/best) showing structural ARPU increase (too early for specifics).
TRACK RECORD — PROMISE vs DELIVERY
—/100 (no quantified promises) no-quantified-promises 6 calls reviewed
Across six calls (Q4 FY2024–Q1 FY2026), Spotify repeatedly cites AI/ML for personalization, DJ, prompted playlists, and faster engineering—but management does not state forward AI targets with both a number and a deadline (e.g., % cost cut by year, $X AI revenue, N features by quarter). Disclosed metrics (10×/6× ubiquity rollout, 50+ features in 2025, ~90M+ DJ users) are retrospective achievements or operating stats, not auditable quantified promises.
PRICED-IN (REFINED)
MEDIUMEst. revisions flat · Fwd P/E 67.5 · EV/Sales 4.6x
AI claim maps to Premium, Ad-Supported
Rating counts migrated slightly more bullish (buy/strongBuy up, holds and sells down since Dec-25), but price targets fell from ~689 (LTM) to ~579 (last month), so revision momentum is mixed, not clearly rising. At ~67.5x next-FY EPS and ~4.6x EV/Sales, valuation is rich while consensus already embeds fast revenue growth and a sharp EPS step-up (7.17 to 12.89 by FY26). AI upside (personalization, retention, ad targeting/monetization) maps to Premium and Ad-Supported; rich multiples with flat revisions imply much of the operating-leverage story is in the price, but falling PTs and non-extreme EV/Sales keep this from a full high priced-in call.
COVERAGE — ENTHUSIASM TRAJECTORY + CATALYSTS
4Q4 FY20245Q1 FY20257Q2 FY20254Q3 FY20257Q4 FY20258Q1 FY2026
AI enthusiasm across 6 calls — trend ↗ rising
AI story grew from light automation and Wrapped personalization to generative DJ metrics, then platform-wide AI under new co-CEOs.
RECENT AI CATALYSTS & NEWS
BUSINESS IMPACT - QUALITATIVE MATERIALITY
8/10 qualitative impact material medium-term · mixed evidence
Where AI matters: personalization, retention, engineering productivity
AI is deployed at scale in the core product (94M DJ users, 52M Song DNA in four weeks, taste/agentic features) and flat headcount with faster shipping, but management cites no ARPU, churn, or margin dollars—only engagement proxies in a flat-fee subscription model.
Caveats: No disclosed revenue or margin lift; hours and adoption may not convert under flat Premium pricing; Rising inference/compute OpEx can offset productivity gains; Generative-music flood or off-platform AI discovery could commoditize listening and pressure royalties; UMG/rights-cleared derivatives still face attribution and label economics hurdles
AI DISRUPTION / CANNIBALIZATION RISK two-sided · 5/10
Generative and synthetic music can flood supply, compress catalog scarcity, and weaken the premium case, yet Spotify’s licensed-aggregator model and discovery at ~250M tracks make curation more—not less—valuable if rights/participation are solved.
OPTIONS / MARKET STRUCTURE
option liquidity: good
proxy inputs — dollar-ADV $1.1B · beta 1.554 · px $484.46
source: proxy (no options chain on FMP)
FMP /stable/ exposes no options-chain endpoint on this key, so ATM IV, bid-ask spread and open interest are unavailable. Liquidity below is a PROXY from dollar-ADV, beta and price level (a stand-in for option depth), not measured option-market data.
CONFIRMATION — INSIDERS · 13F · LANGUAGE
Mixed — insiders selling, institutions adding, management language 6/10 measured.
INSIDERS selling 7 open-market sell(s) vs 0 buy(s) — net distribution
INSTITUTIONS (13F) adding as of 2026-03-31: 205 new / 257 closed positions; 725 increased / 405 reduced; institutional ownership -0.87pp; -43 net 13F holders
MGMT LANGUAGE 6/10 measured Substantive AI section with integration claims and DJ scale metrics, but forward vision leans on positioning, beta features, and Investor Day teases.
commit “We're integrating AI across every part of Spotify, accelerating how we build and deliver at a pace we haven't seen before.”
commit “IDJ is now used by 94 million subscribers, closing in on 100 million, driving billions of hours of engagement.”
hedge “Combined with our ads plus subscription model, deep expertise in personalization and scale operations, we think this positions us very strongly for the AI era.”
VERBATIM AI QUOTES
“These 3 levers are rooted in our personalization efforts and act to reinforce one another. And AI just takes this to a whole new level. Essentially, we are unlocking your Spotify, your way for 0.75 billion users around the world.”
— Alex Norström, Q1 FY2026
“With AI expanding our opportunities, we're building a system that understands our deeply engaged, passionate 0.75 billion users, one that adapts to them and improves the more they use it.”
— Gustav Söderström, Q1 FY2026
“As I shared last quarter, and more recently at South by Southwest, AI isn't new for us. Machine learning and personalization have long been core to Spotify, from discovery to recommendations. What's changing is what the technology is now allowing us to better understand, develop and deliver: creating differentiation and unlocking a very different experience and a new level of personalization.”
— Gustav Söderström, Q1 FY2026
“We're integrating AI across every part of Spotify, accelerating how we build and deliver at a pace we haven't seen before. We're shipping more, faster and with greater efficiency, lowering the cost per feature while increasing the impact. You can see some of the inference costs behind that acceleration in our OpEx.”
— Gustav Söderström, Q1 FY2026
“IDJ is now used by 94 million subscribers, closing in on 100 million, driving billions of hours of engagement.”
— Gustav Söderström, Q1 FY2026
“We're particularly excited about [Taste Profile] now in beta. It gives users a clear view of how Spotify models they are listening across music, podcasts and audiobooks, and puts them in the driver's seat, allowing them to directly edit and refine their profile.”
— Gustav Söderström, Q1 FY2026
“Together, these features point to something bigger, a transition from a world where Spotify recommends things to you, to a world where you can actively shape, guide and interact with our platform, from passive to interactive, from static to adaptive, and from single player to multiplayer.”
— Gustav Söderström, Q1 FY2026
“We're well positioned because of our large engaged user base, our deep create relationships and years of investment in personalization and infrastructure scale to arrive at this agentic moment.”
— Gustav Söderström, Q1 FY2026
“This also reflects R&D related to strategic AI initiatives that we already drive -- that is already driving engagement.”
— Christian Luiga, Q1 FY2026
“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.”
— Gustav Söderström, Q1 FY2026
“Also something we released only 4 weeks ago, Song DNA, is now up to 52 million users, in just 4 weeks.”
— Gustav Söderström, Q1 FY2026
“The generative market, for example, for music is really 2 things. It's net new music, which is happening at scale and quickly increasing the catalog. And that, we think, is good for a company that aggregates content, because it makes the recommendation prompt even more important.”
— Gustav Söderström, Q1 FY2026
“What we think there is a unique opportunity is that, right now, existing creators are largely left out of the AI opportunity altogether.”
— Gustav Söderström, Q1 FY2026
“We want to take this opportunity to existing creators as well, with derivatives of existing IP.”
— Gustav Söderström, Q1 FY2026
“For the first time in the Spotify history, this ability for users to actually tell us in plain English or actually whatever language they want, what they want. We were always guessing. Old-school machine learning was a statistical activity based on clicks and streams.”
— Gustav Söderström, Q1 FY2026
“I talked last time about the large personalization model, which is a model that we're training from based on open source models. But it's trained on our proprietary data. This is not something that we rent from someone. This is something we're building in-house. And the casual name for the large personalization model is a taste model.”
— Gustav Söderström, Q1 FY2026
“If AI increases engagement for us, it generally means that it increases personalization for us, right? And increased personalization engagement, to Gustav's points, are going to lead to -- well, they are going to be the best proxies for the increase in retention that we're going to see over time with these investments.”
— Alex Norström, Q1 FY2026
“We are training rather large models in-house, because we have lots and lots of unique data that no one else has. For example, the large personalization model, which is not something that you can rent or buy off the Internet.”
— Gustav Söderström, Q1 FY2026
“It turns out luckily for us, taste is not easily commoditized because it's not a fact, it's an opinion.”
— Gustav Söderström, Q1 FY2026
“We acquired Echo Nest back in 2014 when most people didn't understand why a streaming company needed a machine learning AI company. And that bet gave us personalization, something that's now core to everything we do.”
— Daniel Ek, Q4 FY2025
“It means our investments into personalization and AI are paying off.”
— Alex Norström, Q4 FY2025
“There is no question in my mind that we will continue to be one of the big beneficiaries of AI.”
— Gustav Söderström, Q4 FY2025
“Back in 2021, we saw the potential of AI that would be able to think and speak at the level of a human. So we acquired AI voice platform, Sonantic, in 2022. And this put us on an early path to introduce agentic experiences to Spotify users.”
— Gustav Söderström, Q4 FY2025
“About 90 million subscribers have used IDJ so far, driving over 4 billion hours of time spent on Spotify, and this keeps growing.”
— Gustav Söderström, Q4 FY2025
“So all of this teases the next evolution of Spotify, delivering the world's most intelligent agentic media platform, one that you can literally talk to, that fully understands each individual listener and puts them in the driver's seat.”
— Gustav Söderström, Q4 FY2025
“We are building a dataset that never existed, which is the data set of language to music, language to podcast and language to books.”
— Gustav Söderström, Q4 FY2025
“It turns out that taste is not a fact, it's an opinion.”
— Gustav Söderström, Q4 FY2025
“An engineer at Spotify on their morning commute from Slack on their cell phone can tell Claude to fix a bug or add a new feature to the iOS app. And once Claude finishes that work, the engineer then gets a new version of the app pushed to them on Slack on their phone so that he can then merge it to production, all before they even arrived at the office. We call this system internally Honk”
— Gustav Söderström, Q4 FY2025
“AI leads to better personalization, better personalization leads to more engagement, more engagement leads to more retention, more retention leads to lifetime value and, boom, more lifetime value leads to more enterprise value.”
— Alex Norström, Q4 FY2025
ANALYST QUESTIONS ON AI
Q (Q1 FY2026, Jessica Reif Ehrlich): Q1 had higher marketing cloud and AI spend. Can you discuss the pace of investment for the balance of the year and how you would define a successful outcome for this investment spend?
A: Gustav Söderström: Headcount slightly decreased while compute per employee rose due to productivity returns; shipping accelerated since late fall; more sales/marketing to drive awareness of new features; opportunity as big as or bigger than 2009 iPhone/App Store—Spotify is taking it with discipline; DJ nearing 100M users; Song DNA at 52M users in 4 weeks; usage is proxy for retention and long-term revenue.
Q (Q1 FY2026, Doug Anmuth): Can you update us on your progress towards new AI products that would empower users to create new content and enable derivatives of existing music? What are the hurdles to launching these products? And do you expect that they would impact your cost structure or margin trajectory in any meaningful way?
A: Gustav Söderström: Generative music is net-new catalog (good for aggregators—recommendation becomes more important) and derivatives of existing IP where existing creators are left out; copyright/attribution are hard; Spotify has capabilities and is the right company to solve derivatives so existing creators participate in AI.
Q (Q1 FY2026, Justin Patterson): We're seeing many companies wrestle with headcount investment versus rising AI costs. How is Spotify approaching this problem in gauging employee productivity?
A: Gustav Söderström: Middle path—roughly flat headcount, doing much more, shipping more value; proxies include pull requests, DODs, bets board—all doubling not +10%; usage from features like Song DNA translating to retention/revenue. Alex Norström: No employee growth for 3 years; decreased 65 people last quarter.
Q (Q1 FY2026, Deepak Mathivanan): You have integrated Spotify and leading AI applications already, ChatGPT last year and Claude more recently. Can you talk about what type of traffic you're seeing and how consumers are using Spotify in AI applications at this time? And how are AI applications helping KPIs such as MAUs and time spent?
A: Gustav Söderström: Ubiquity strategy—be wherever users are; tracking usage/engagement/cost diligently and seeing what they want; users now tell Spotify in plain language (vs old click/stream ML); training proprietary large personalization/taste model on that data; advanced usage in Prompted Playlist and DJ. Alex Norström: AI→personalization→engagement→retention→LTV→enterprise value framework.
Q (Q1 FY2026, Eric Sheridan): Can you frame the key platform and product initiatives that are driving incremental operating expense trends? How should investors think about the trajectory of operating margins going forward?
A: Gustav Söderström: OpEx mix is increased compute (not headcount) and sales/marketing—compute for code acceleration/proprietary systems like Honk and in-house training of large models (e.g. large personalization model) on unique data; strategic plus productivity investment. Christian Luiga: Next 2 quarters elevated; operating margin still improves year-over-year.
Q (Q1 FY2026, Steven Cahall): Does Spotify believe in an AI music creation tier? And if so, what are the sticking points with content partners and how might it be priced to Premium users?
A: Gustav Söderström: Big opportunity is existing artists using AI on derivatives of existing IP, not left out; existing IP is most valuable in other media but not addressable in music today due to rights—problem Spotify wants to solve for creators, Spotify, and investors.
Q (Q1 FY2026, William Packer): Could you outline the key moats for Spotify that limit the risk from stand-alone low-cost free AI music alternatives, large platform peers with free AI music, and AI-first alternatives integrating label content?
A: Gustav Söderström: ~20 years listening history; taste is opinion not fact—large in-house taste/personalization model on proprietary data; model stale in ~2–3 weeks without scale; China soda-style paid-gate risk differs—Spotify never gated on content in Western markets; can still benefit from catalog growth and serving existing creators.
Q (Q4 FY2025, Jessica Reif Ehrlich): How is Spotify planning to use AI tools and applications for new and evolving product offers? And will this eventually lead to new tiers of service?
A: Gustav Söderström: Macro-change opportunity; right ads+subscription model; investing for years; building first intelligent agentic media service (DJ casual talk, Prompted Playlist deep research); unique language-to-music/podcast/books dataset; taste is opinion not canonical fact—excited about retraining improving models.
Q (Q4 FY2025, Eric Sheridan): Discuss latest thoughts on AI: role in product/platform evolution, effect on internal processes, and broader audio content creation/distribution landscape.
A: Gustav Söderström: Rebuilt company for agentic age—English queries formerly needing senior developers; Honk lets engineers use Claude on commute to fix iOS bugs and merge before office; retooling entire company; more at Investor Day.
Q (Q4 FY2025, Rich Greenfield): What percentage of music on Spotify today is AI-generated? How much AI-generated content is being uploaded daily? And what is your policy on the uploading of AI music?
A: Gustav Söderström: Won't share AI-generated percentage; won't decide creators' tools but will surface creation metadata via industry work and About the Song; spammy AI tracks scale existing abuse problem where Spotify leads; more catalog makes personalization more important.
Q (Q4 FY2025, Rich Greenfield): Is Spotify playing to win in AI? Bear thesis that Udio, Suno, Klay, Stability become DSPs taking share while Spotify is cautious.
A: Alex Norström: Industry optimistic; Spotify is scaled distribution/monetization with working business model; whole industry lined up behind vision; controlled approach respecting artists.
Q (Q4 FY2025, Justin Patterson): Expand on Spotify's role in AI music—need to invest in content creation tools? How helping human creators build audiences and income?
A: Alex Norström: AI→personalization→engagement→retention→LTV chain. Gustav Söderström: Has all technology/capabilities needed; working with industry to enable opportunities.
Q (Q4 FY2025, Steven Cahall): Market implies Spotify negatively impacted from AI. What is the market missing? Top priorities so you don't fall behind?
A: Christian Luiga: AI hard to grasp; Spotify started years ago—late starters have tougher time. Gustav Söderström: Invest with discipline when clear opportunities and returns.
Q (Q4 FY2025, Justin Patterson): How is agentic coding changing product velocity? What could GenAI mean for engineer productivity and R&D investment needs?
A: Gustav Söderström: December threshold—senior engineers only generate/supervise code; Honk built for this; beginning of change; companies will produce massively more software until consumer change tolerance is limiting factor; aggregation not disaggregation like Internet.
Q (Q4 FY2025, Batya Levi): Back in October, you had announced partnership with the major labels to develop artist-first AI products. With all the hype about competition and disruption, can you talk about how you plan to differentiate with these products? And is there an urgency [transcript truncated]
A: [Transcript ends before management answer.]