Why I am long Duolingo: durable compounding from product-led growth, AI flywheel, and a second act in testing and new subjects
I am long Duolingo. Here is the summary of my thesis and the near term setup.
Thesis in one line
A product that gets better the more it is used, with a flywheel across free learners, paid tiers, and the English testing business. Execution has accelerated, guidance is moving up, and new subjects and AI features create fresh monetization vectors.
What has actually changed in 2025
1. Growth is still brisk at scale. Q2 FY25 revenue grew 41 percent with record profitability and raised full year guidance. Daily active users grew about 40 percent and paid subscribers were up more than 35 percent to roughly 11 million exiting Q2.
2. Mix is improving. Management called out higher ARPU from users moving into the Max tier that bundles advanced AI features. That uptier is already visible in Q2 commentary and coverage.
3. Optionality is turning into product. Duocon 2025 announced major app updates across language, math, and music. Earlier in the year the company continued to invest in Max and new subjects and even picked up a music gaming team to deepen capability.
Fundamentals snapshot
• Topline: Q2 FY25 revenue approximately 252 million, up 41 percent year over year. Bookings up 41 percent as well. Full year revenue outlook lifted to about 1.01 to 1.02 billion per press and coverage.
• Users: DAUs up roughly 40 to 50 percent year over year in 1H25. Paid subscribers over 10 million as of Q1 and ~10.9 million by Q2. These are large absolute numbers that still post high growth.
• Profitability: Record profitability with Q2 EPS beating expectations and adjusted EBITDA guide moving up. The model shows operating leverage even while the company funds AI features.
• Dilution: Management reiterated a roughly 1 percent fully diluted share count increase for 2025 which is modest for a company at this stage.
uolingo English Test and new subjects
The English Test is now accepted by more than 6,000 programs worldwide, including 98 of the top 100 US universities by U.S. News, plus broad adoption in Canada and the UK. This penetration matters because it makes the test a real alternative to legacy incumbents and drives steady, countercyclical demand from international applicants.
Upcoming catalysts over the next 6 to 12 months
• Feature rollout from Duocon announcements across language, math, and music that can lift engagement and conversion to paid tiers. These updates tend to show up quickly in DAU and bookings.
• Continued mix shift into Max as conversational AI improves. Coverage highlighted ARPU tailwinds from Max adoption in Q2. As the experience gets more natural, attach should rise.
• Seasonal subscriber adds around holiday and New Year resolution windows. The company has historically seen strong Q4 to Q1 momentum which can pair with the higher guide.
• Further English Test acceptance growth and regional initiatives such as DETcon China that expand institutional buy in and brand credibility.
• Expansion of new subjects like chess and continued math and music iteration which broaden the funnel beyond language learners and create cross sell.
Why the edge is durable
Duolingo’s edge is a compounding data and engagement loop. Millions of daily exercises feed item response models and personalization. Better lessons increase streaks and retention which increases the pool of potential subscribers. The Max tier adds real time AI practice that closes the gap between study and conversation. The test business adds a paid credential that is hard to displace once institutions adopt it. The result is a consumer subscription with SaaS like unit economics at internet scale.
Risks I care about
• Macro or app store platform changes that affect conversion or take rates.
• AI cost of goods sold if usage of advanced models outpaces pricing. Management has acknowledged near term gross margin impact in exchange for faster innovation.
• Competition from focused verticals or from large platforms bundling language.
• Volatility around quarterly user comps given very large DAU bases.
What I am watching
• Max attach and ARPU trend each quarter.
• English Test acceptance growth and any new regions added.
• Bookings growth versus revenue as a clean signal of forward demand.
• Share count and stock based comp staying disciplined at about 1 percent dilution for 2025.
Positioning
I am long and adding on drawdowns that are not thesis related. I size it as a core growth holding given the mix of product velocity, recurring subscription revenue, a growing testing franchise, and expanding TAM from new subjects. Execution in 2025 has earned the right to underwrite continued compounding.
Not investment advice.