Deck
Duolingo runs a freemium mobile app where roughly 9% of monthly users pay about $84 a year for a gamified, ad-free language-learning experience; subscriptions are 84% of revenue.
Priced like Chegg in slow motion — but the leading indicator says otherwise.
- The market's framing. EV/Sales compressed from 18.4× at the FY24 close to 3.8× today, short interest hit 20.2% of float, and bulge-bracket targets cluster between $94 and $101 — all built on the bet that free generative-AI chatbots will hollow Duolingo the way they hollowed Chegg, where paid subscribers fell from 8.1M to 2.9M in three years.
- What the data actually shows. Paid % of MAU rose from 8.8% in Q4 FY24 to 9.3% in Q1 FY26; paid subscribers grew 21% YoY to 12.5M in Q1 FY26 (and 28% YoY for FY25 to 12.2M) — paid % is widening even as DAU growth also runs 21%. The Chegg fingerprint is engagement holding while monetization erodes; Duolingo's monetization line is moving up, not flat.
- What the moat is, and isn't. Sensor Tower puts Western-market churn at 28% for Duolingo versus 58% at Babbel; DAU/MAU sits at ~40% in the gaming and social-leader band; S&M is 12% of revenue because the brand is viral. The undefended part is curriculum — vocabulary and grammar drills are exactly what free chatbots do at zero marginal cost.
An AI-first memo strained the brand engine — the FY25 10-K names it in its risk factors.
Before April 2025: Duolingo was a Spotify-template compounder. DAU grew 49% in Q1 FY25, revenue grew 41%, and the stock crossed $540 at a 26× EV/Sales multiple. From Q2 FY22 through Q3 FY25 the company raised guidance every single quarter — fourteen consecutive prints.
The pivot: In April 2025 a CEO memo said AI would replace contractors and become a performance-review input. The post went viral the wrong way. DAU growth then decelerated from +49% (Q1 FY25) to +21% (Q1 FY26) over the next four quarters. By February 2026 management guided FY26 bookings growth to ~+10.5% (down from +33% in FY25), the CFO was replaced by the sitting Audit Committee Chair, and a $400M buyback was authorized — all in the same week.
Today: The FY25 10-K risk factors say the April memo "may have contributed to ... a deceleration in user growth" — companies rarely flag their own communications as a contributing risk event. Six plaintiff law firms have opened investigations. The next chapter answers a single question: does brand damage from one LinkedIn post amortize, or compound?
Cash generation is genuine — headline earnings are not.
FY25 GAAP net income of $414M includes a $232M net tax benefit (driven by a $257M release of the deferred-tax-asset valuation allowance in Q3) — pretax income of $182M is the cleaner number. Strip $137M of stock-based comp from FCF and the 35.6% margin becomes 22.4%; on that honest basis the stock trades around 18× owner earnings on an EV basis for a mid-teens grower, which is fair, not cheap. The reported cash gap exists because subscription billings are collected upfront and recognized monthly; $496M of deferred revenue cushions FY26 reported revenue, but the cushion drains by FY27.
Pay structure is best-in-class — the audit and disclosure functions are stretched.
- The capital structure works for shareholders. CEO Luis von Ahn took $768K in 2025 and has taken no new stock grants since 2023 by design. His Founder PSU framework caps new CEO/CTO equity through July 2031, so every dollar of the $400M buyback shrinks share count rather than refilling the comp pool. Eight of ten PSU price hurdles already cleared.
- The audit function is concentrated. Gillian Munson moved from Audit Committee Chair (and designated financial expert) directly to CFO effective February 23, 2026 — the transition was announced January 8 and she vacated her audit seat the day she started the executive role. Sara Clemens now serves as both Audit Committee Chair and the designated financial expert, sitting on top of a classified board and dual-class shares that concentrate ~76% of voting power in the two co-founders.
- The disclosure is under attack. Six plaintiff firms — Faruqi, Pomerantz, Portnoy, Levi & Korsinsky, Bronstein, Johnson Fistel — opened investigations between August 2025 and May 2026 alleging undisclosed Q2 FY25 DAU deceleration. Adjusted EBITDA and Free Cash Flow definitions have each been amended twice in twelve months, and MAUs were demoted to a 'supplementary' metric the same quarter MAU growth fell below 6%.
Watchlist — one variable decides this, and August 5 is when it gets to vote.
- For. Best-in-class consumer-subscription economics — the only listed name combining 30%+ growth, 70%+ gross margin, 35% FCF margin, and 50%+ ROIC in a single year. Net cash is 22% of market cap; the Founder PSU plus $400M buyback structure shrinks share count permanently through 2031, an alignment no peer can replicate.
- For. Paid % of MAU rose from 8.8% to 9.3% across the four quarters in which the multiple compressed by ~80%. The Chegg substitution thesis requires this line to fall; through six quarters of generative-AI mainstream adoption, it has risen instead.
- Against. A 23-point bookings deceleration in twelve months is what thin-moat franchises produce, not wide-moat ones — and the 10-K cites a single LinkedIn memo as a contributing risk in the user-growth slowdown. Brand moats can fade in one news cycle; that channel has now been proven open.
- Against. On SBC-honest cash earnings the stock trades around 18× for a mid-teens grower. The 'cheap on 12× FCF' bull leg relies on a metric that excludes 13% of revenue paid in stock — that is fair value, not bargain.
Watchlist to re-rate: (1) Paid % of MAU in Q2 FY26 (August 5) and Q3 FY26 — flat-or-down YoY for two consecutive quarters would convert the call to Avoid. (2) Constant-currency bookings re-acceleration above 13% in Q2 FY26 would convert it to Lean Long. (3) UK Home Office £816M migrant English-test tender, resolving Q3–Q4 2026 — currently carries zero embedded value in consensus targets.