Chegg, Duolingo, and other edtech companies insist generative AI has the potential to enhance their products, despite the threat of being undercut by cheap AI
Bethan Staton / Financial Times : Tweets: @wavesblog Tweets: @wavesblog : Duolingo: ""We have more data on how people learn a language than any entity in history," https://twitter.com/...
Context & Ripple Effects
This piece lands weeks after educators first began cautiously experimenting with generative AI in classrooms, and it captures edtech's defensive posture at that moment: Chegg and Duolingo publicly arguing that AI enhances their products rather than replaces them, while conceding that cheap AI rivals could undercut their paid offerings. Duolingo's counter is a data-moat claim — "more data on how people learn a language than any entity in history" — positioning proprietary learner behavior, not model access, as the asset worth paying for.
The surrounding arc sharpens why this matters: Stanford's late-2023 survey found high-school cheating rates flat, deflating one feared substitution mechanism, but the undercut threat later took institutional form when [[a:1173758|AI frontier labs started building free or cut-price tailored learning tools directly with schools and edtech startups]]. Against that backdrop, Duolingo's reported quarter — revenue and subscriber counts still growing double digits yet both below analyst estimates, with soft forward guidance — reads as growth intact but investor patience priced around AI risk.
First-order effects
- Duolingo's latest reported quarter shows the squeeze concretely: $298.5M revenue (up 18%) and 12.7M paid subscribers (up 17%) still missed estimates, and its next-quarter forecast came in below consensus — markets are discounting an otherwise healthy subscription business for AI substitution risk.
- Chegg, whose homework-help product is the most directly undercuttable of the named players, faces immediate pressure to demonstrate AI-enhanced differentiation or watch its core paid Q&A offering be replicated by general-purpose chatbots.
Second-order effects
- Frontier labs' response is to skip competing on features and compete on distribution: by partnering with schools and edtech startups to give away cut-price learning tools, they convert the undercut threat from a consumer choice into an institutional procurement decision.
- As tech companies deepen school-channel investment — including partnerships with major teachers' unions — the battleground shifts from app-store marketing to classroom integration, raising the cost of customer acquisition for standalone edtech incumbents.
Third-order effects
- If the pattern holds, standalone tutoring and study subscriptions get unbundled into general AI assistants, leaving durable pricing power with whoever owns longitudinal learner data — Duolingo's explicit bet — while content-and-answer providers like Chegg face structural margin erosion rather than a cyclical dip.
- Whether edtech survives as a distinct category may hinge less on model quality than on whether schools adopt AI through incumbent platforms or through lab-run partnerships — an open question the current coverage does not resolve.
The trend: Consumer edtech is shifting from selling proprietary study subscriptions to defending distribution channels as AI labs give learning tools away, with longitudinal learner data as the last defensible moat.