In partnership with |  |
|
Duolingo closed February 26, 2026 at ~$92 per share — a market cap of roughly $4.3 billion on basic shares (46.6M), or ~$4.6 billion fully diluted (49.8M shares). Nine months ago, this stock traded above $544. |
Let that sink in. The stock has shed roughly 83% of its value from peak to trough. And after last night's Q4 earnings report — which beat on both revenue and EBITDA — shares dropped another 22% because management guided to slower growth and lower margins in 2026. |
The market is treating DUOL like a terminally disrupted business. I think it's one of the most interesting setups in software right now. |
Want to get the most out of ChatGPT? |
|
ChatGPT is a superpower if you know how to use it correctly. |
Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done. |
Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI. |
Download the free guide |
What Actually Happened |
Three things have converged to obliterate this stock: |
1. The Broader SaaS/Software De-Rating |
Software has been in a rolling bear market since mid-2025. The "AI will eat software" trade has become the consensus fear. Every consumer app, every SaaS company, every subscription-based software business is getting repriced on the assumption that AI agents and commoditized intelligence will destroy pricing power. DUOL has been caught in this downdraft, but the irony is that Duolingo is arguably a beneficiary of AI, not a victim. |
2. The T-Mobile "Live Translation" Panic (Feb 11) |
T-Mobile announced a real-time AI translation feature embedded into its 5G network. DUOL dropped 10%+ in a day. The logic: if AI can translate in real time, why would anyone learn a language? |
This is a profoundly stupid thesis. Duolingo is not a translation app. It is a learning app. Translation and education serve fundamentally different human needs. Nobody picks up Duolingo because they need to translate a phone call — they use it because they want to learn a language. The product is gamified, habit-forming, and deeply social. The comparison is like saying Google Translate killed university language departments. It didn't. People who want to learn still learn. If anything, the explosion of AI translation tools increases exposure to foreign languages and may increase the desire to actually learn them. |
3. The 2026 "Investment Year" Guidance Shock (Feb 26) |
This is the real catalyst for the latest leg down. Management guided to: |
Bookings growth of 10-12% (vs. 33% in 2025, and ~20% consensus expectation) Revenue growth of 15-18% (vs. 39% in 2025) Adjusted EBITDA margin of ~25% (vs. 29.5% in 2025)
|
CEO Luis von Ahn was blunt: Duolingo is deliberately throttling near-term monetization to re-accelerate user growth. The company estimates it is investing $50M+ in foregone bookings by removing friction from the free-user experience. The medium-term goal? 100 million DAUs by 2028, up from 52.7M today. |
Wall Street hated it. But let's look at what the numbers actually tell us. |
The Business: Better Than You Think |
|
This is a company that just crossed $1 billion in revenue growing 39% year-over-year, with 72%+ gross margins, ~30% EBITDA margins, and ~35% free cash flow margins. It ended the year with $1.04 billion in cash and zero debt. |
Read that again. The enterprise value (basic market cap minus cash) is roughly $3.3 billion. |
The Unit Economics Are Elite |
Duolingo's business model is enviable among consumer internet companies: |
72% gross margin — comparable to the best pure-play SaaS companies. This is a high-margin subscription business. 34.7% FCF margin — the company converts over a third of its revenue into free cash flow. Among consumer apps at this scale, this is exceptional. ARPU is expanding — subscription ARPU grew mid-to-high single digits in 2025, driven by mix-shift toward higher-priced tiers (Max, Family Plan). Even as the company deliberately eases monetization friction, the users who do subscribe are spending more.
|
SG&A as a Percentage of Market Cap: The Punchline |
Here's what caught my eye. Based on FY 2024 financials (the most recent standardized annual data available): |
|
In FY 2025, GAAP operating expenses grew 19% YoY in Q4, implying full-year GAAP OpEx was likely in the ~$570-580M range. But a huge component of GAAP OpEx is stock-based compensation. In 2025, SBC was ~13% of revenue (~$135M). Stripping that out, cash operating expenses are dramatically lower than the headline GAAP figures suggest. |
Now consider: at a ~$4.3B basic market cap, estimated FY2025 total GAAP operating expenses represent roughly 13% of market cap. But the non-GAAP cash operating expenses (ex-SBC) are closer to ~$435-445M, or about 10% of market cap. That's for a company still growing 39% revenue, 33% bookings. In other words, the market is assigning almost no value to the operating leverage embedded in this business. |
Even more telling: Free cash flow of $360M on a ~$3.3B enterprise value = ~11% FCF yield. For a company growing at this rate, that's extraordinary. Many mature, no-growth businesses trade at lower FCF yields. |
The AI Bull Case: Duolingo Is an AI Company |
The market's AI-fear narrative gets DUOL exactly backwards. Duolingo is not being disrupted by AI — it is one of the most successful deployers of consumer AI products outside of the frontier model companies themselves. |
Consider: |
Video Call with Lily is one of the most commercially successful consumer AI features. It allows learners to have real-time conversations with an AI character, with corrections, in the target language. This is something only possible because of generative AI. AI powers content creation — Duolingo has used AI to dramatically scale its lesson content. This is a direct margin benefit. The company has been open about using AI to reduce headcount in content creation roles and redirect resources to product engineering. AI reduces costs at scale — As LLM inference costs continue to fall (and they are falling rapidly), Duolingo's AI-powered features become cheaper to deliver, which improves margins over time. Management noted that AI costs came in lower than expected through much of 2025. AI features are the monetization driver — Duolingo Max, priced at a premium, is built entirely on AI features. The upcoming move of Video Call to the Super tier will bring AI-powered learning to 10x more paid users.
|
The bear case assumes AI makes Duolingo obsolete. The bull case — which the fundamentals support — is that AI makes Duolingo's product dramatically better and more defensible. Nobody else has 52.7 million daily active users, petabytes of learning data, and a decade of gamification expertise. ChatGPT doesn't have streaks, leaderboards, hearts, or a green owl guilt-tripping you at 9pm. |
Forecasting Free Cash Flow |
Let me build a simple framework for what 2026 and 2027 could look like, using management guidance and reasonable assumptions: |
|
Management explicitly guided to "over $350 million" in free cash flow for 2026, even in an investment year. This is a business that generates cash even when deliberately suppressing monetization. |
2027E (Illustrative — Post-Investment Reacceleration) |
If the user growth strategy works and DAUs approach 80-100M: |
|
The point: even in the base case, Duolingo is on a trajectory toward $400M+ in annual free cash flow within 18 months. In the bull case, you're buying a $400M+ FCF machine at a ~$4.4B EV. |
Backing Into Valuation |
Let's look at this from multiple angles. |
|
A 10%+ FCF yield on a 15-18% revenue grower is a deep value setup in software. |
EV/Revenue |
|
For context, the average consumer internet/subscription software company at these growth rates trades at 6-10x forward revenue. DUOL is priced as if growth is going to zero. |
P/E |
At ~$92/share with ~$1.88 in FY 2024 EPS (before the one-time tax benefit), the trailing P/E is around 49x. But on a 2026 normalized earnings basis (stripping out the 2025 DTA release), if we assume GAAP earnings of ~$4-5/share (Adj. EBITDA of $302M, less ~$180M SBC, less ~$15M D&A, plus some tax benefit), the forward P/E is closer to 18-23x. For a 15-18% grower, that's very reasonable. |
The $400M Buyback Puts a Floor |
The company just announced a $400M share repurchase authorization. At $92/share, that represents roughly 9.3% of the basic market cap. With $1.04B in cash and $360M+ in annual FCF, they have the firepower to execute aggressively. This is a meaningful capital return signal, and at these levels it's highly accretive to remaining shareholders. |
The Risk Framework |
Let's be honest about the risks: |
User growth deceleration is real. DAU growth went from 40%+ to 30% in Q4 2025, and management guides to ~20% in 2026. If the $50M investment in free-tier quality doesn't reaccelerate growth, the 100M DAU target looks aspirational. Monetization giveback could be permanent. Removing friction and dropping Video Call from Max to Super are revenue headwinds. There's a world where Duolingo grows users but never fully re-monetizes them. AI competition is real (long-term). While I think the T-Mobile narrative is silly, the broader question of whether purpose-built AI tutors (from Google, OpenAI, Apple, etc.) could eventually replicate Duolingo's experience is legitimate. The moat is behavioral and brand-driven, not purely technological. SBC dilution is non-trivial. At ~15% of revenue in 2026 and 3.5-4% diluted share count growth, the SBC burden is real. The buyback partially offsets this, but it's worth watching. CFO transition. Matt Skaruppa stepped down; Gillian Munson (board member, audit committee chair) took over as CFO effective Feb 23, 2026. Transitions introduce uncertainty.
|
The Bottom Line |
Duolingo is a rare thing in software right now: a genuinely high-quality business trading at a genuinely distressed valuation because of a confluence of macro fear (SaaS de-rating), misplaced thematic panic (AI = death), and a voluntary near-term growth slowdown. |
The business generates $360M in free cash flow, growing double digits, with 72% gross margins, $1B+ in cash, no debt, and a $400M buyback authorization. It has 52.7 million daily active users who voluntarily spend time learning on their phones every day. The CEO — a Carnegie Mellon professor who invented reCAPTCHA and sold it to Google — is making a deliberate long-term bet on user growth over near-term earnings optimization. |
At ~$92/share: |
|
This is a setup where the downside feels well-defined (cash + buyback + FCF generation) and the upside depends on whether you believe Duolingo can continue to be the world's leading education platform in an AI-powered world. |
I do. |
Disclaimer: This is not financial advice. Do your own due diligence. Positions may be held by the author. |
