Duolingo Stock: The AI Casualty Tilson Names

Duolingo stock appears in Whitney Tilson’s new N.E.W. System presentation as a name on the wrong side of the market’s AI sorting, and the citation is factual: Duolingo (NASDAQ: DUOL) has been one of the most visible AI-era repricings in consumer software, falling roughly 75% from its highs as the market rethought what large language models do to a business built on human-constructed lessons. Understanding why the market sorted it there is genuinely useful for reading this moment. We read the presentation end to end. Our full AI Civil War teardown covers it in detail. For the full backstory, see our Gartner IT Stock: The Consulting AI Angle article. There is more on this in our Gartner Stock: The Hidden AI Beneficiary? article.

What Happened to Duolingo

Duolingo built one of the great consumer products of the app era: gamified language learning with hundreds of millions of downloads, a freemium funnel, and a subscription business that converted engagement into revenue. The AI question hit the thesis from two directions. On the product side, conversational AI can tutor a language learner directly, with infinite patience and no lesson library, which puts the core differentiator, the carefully constructed learning tree, in competition with a model that improvises. On the cost side, Duolingo itself adopted AI aggressively for content production, which improved margins while raising the uncomfortable question of what the moat is if the content is machine-made.

The market’s answer over the past year was the repricing Tilson cites. Whether that answer is final is a separate question, and honest analysis holds both: the drawdown is real, documented market history, and the long-term verdict on AI-native education competition is genuinely unknown.

Where the Name Fits the Pitch

Duolingo appears in the presentation’s casualty catalog alongside Gartner, Oracle, and Salesforce, companies whose models the market decided AI exposes rather than rewards. The catalog is the mirror of the pitch’s winners list, and the pairing is the product’s core claim: that the AI economy is bifurcating, that “We’ve hit the ripping point,” and that a systematic score (the Stansberry Score, 0 to 100 across 5,300-plus stocks) sorts the arms in advance. The free picks demonstrate the sorting: Garmin (GRMN) high, Rivian (RIVN) low.

The honest note on the casualty list itself: the presentation’s “Ripping Point Casualties” report promises “the seven low-scoring stocks you almost certainly own,” and Duolingo’s appearance in the surrounding catalog is as an example of the phenomenon, not necessarily as a report name. The distinction matters, and the pitch’s citation of real, documented drawdowns is fair as evidence of bifurcation.

The Deeper Lesson for AI-Exposed Companies

The Duolingo case generalizes into a framework worth keeping, because the next name on a “casualty list” may not be Duolingo at all. For any company whose product is information work, translation, summarization, image generation, tutoring, code, the honest question is not “will AI replace it” but “which layer does the company own.” If the company owns a proprietary data asset, a brand, a distribution moat, or a regulatory position, the AI wave can land on its cost structure without taking its revenue, and the stock’s AI story is about margin, not extinction. If the company owns only the workflow layer, the interface between a user and a task that a frontier model can now do directly, then the moat is thin and the casualty-list framing has real teeth. Duolingo’s bears argue its gamified habit loop is workflow; its bulls argue the habit loop and brand are the asset. That specific debate is the one worth following, and it is a better template for reading every future “AI kills X” pitch than any single verdict on one stock would be.

The Broader Pattern

Duolingo’s repricing belongs to the same 2025-2026 market episode as the software and consulting damage the presentation catalogs: the SaaSpocalypse repricing, Gartner’s collapse (including the one-day plunge after February earnings), the pressure across Accenture, S&P Global, FactSet, and Morningstar, and Oracle’s reversal. The pattern’s honest lesson: the market has started charging AI-exposure rent on every business model, charging it to the models’ competitors and paying it to the models’ suppliers. A consumer subscription business whose content AI can replicate pays the rent. That is not a moral verdict on Duolingo’s product, which remains excellent; it is a market verdict on the moat, and the market reopens that verdict every quarter.

What You Get

The N.E.W. System: quarterly portfolios, monthly updates, the Stansberry Score tool, Stansberry GPT, “Ripping Point Riches” and “Ripping Point Casualties,” and a mystery bonus. Price behind the order button, 30-day full refund with keep-the-reports. Details in the presentation. For the framework, our AI Civil War explainer covers the sorting thesis, and our Gartner piece covers the other big casualty name.

The Honest Read

Duolingo is the cleanest consumer-software exhibit of the AI repricing: a beloved product, a real subscription business, and a market that decided the moat was vulnerable anyway. Tilson’s presentation cites it accurately and uses it correctly, as evidence that the sorting is real, not as a guarantee that the verdict is permanent. For investors, the lesson is the one the pitch sells and the market keeps teaching: the question is never whether the product is good. The question is what AI does to the numbers. Duolingo answered that the hard way, and the honest reader holds the example with the respect it earned.

NewsletterVetter is an independent publication. We receive compensation from some of the services we review through affiliate links. Nothing on this site is investment advice. Always do your own research.