How GenAI Costs Sank Duolingo’s Stock 20% (A Non-AI Generated Analysis)
A human analysis of financial documents reveals what AI content farms missed: Duolingo’s GenAI features are undermining its own profit margins.

Last month, Duolingo announced record-breaking earnings—yet, surprisingly, its stock dropped nearly 20%. The company is currently valued at $13 billion.
The culprit? Surprisingly, it’s the very technology that’s supposed to be revolutionizing the company: Generative AI.
Trying to Google for answers, all you get is AI-generated articles on why Duolingo’s stock dropped without really answering the question. So I set out to figure out why and read through their Form 8-K and earnings call transcript.
Just a disclaimer: this discussion is explicitly about Duolingo’s stock price—the single metric that decides the company’s future direction. The company itself is profitable and projected to earn almost a billion dollars in 2025, surpassing its edtech peers like Coursera and Udemy.
In my experience of observing companies for more than a decade, stock price or valuation is the lens that best explains a company’s decisions. Mission is fungible; stock price isn’t.
That is why Duolingo is trying to expand beyond languages and into math and music. It’s searching for the next “big market.”
Duolingo’s Record-breaking 2024 Results

Before I get into my theories, let me summarize their 2024 results for you (alright, I am using GenAI for this one).
Daily active users jumped to 40.5 million, up 51% from the previous year, while monthly active users reached 116.7 million, a 32% increase.
The company added almost 1 million paid users, taking the total number of paid subscribers to 9.5 million—a 43% increase year-over-year.
This growth translated into total revenue of $748 million, growing 41% compared to 2023. Most impressively, the company made $88.6 million in profit, a massive increase from $16.1 million in 2023.
These impressive results and new offerings like Duolingo Max set up the company to potentially exceed $1 billion in bookings in 2025.
Why Duolingo’s Stock Price Dropped

Based on the questions asked during the earnings calls and my own understanding, I believe there is one major and one minor reason why Duolingo’s stock price dropped.
Duolingo Max Impact on Margins
The first question (and one of many) that was asked by investors in the earnings call was about Duolingo Max. Duolingo Max is the newest and most expensive subscription that Duolingo has to offer. It includes all the features of Super Duolingo and includes features powered by GenAI like Video Call with Lily.
It’s clear from the questioning that investors consider Duolingo Max as a key part of Duolingo’s business. Duolingo has announced that Max accounts for 5% of paid subscribers, or almost half a million users. Here is a key point on the impact of Max on margins from their 8-K:
“In Q4, gross margin decreased by approximately 120 basis points year over year to 71.9% due to lower subscription margins from increased generative AI costs due to the increased adoption of Duolingo Max, and lower advertising gross margin during the period from lower advertising revenue per DAU. This was offset by an increase in subscription revenue as a percentage of total revenue.”
In simple English, Duolingo’s margins have decreased because of costs related to using GenAI. Duolingo uses OpenAI APIs for its GenAI features and was one of the featured partners when GPT-4 was announced two years ago.
According to CEO Luis von Ahn, “Most of our AI costs are tied to Video Call, and they scale based on Max subscriber growth. We’re prioritizing using the latest AI models to deliver a high-quality experience. What that means is that we’re not concentrating on cost optimization right now, even though we’re confident these costs will come down over time.”
In short, the more users use the video calling feature, the lower Duolingo’s profit margins. Heavy users of the app might not be profitable at all. This does not bode well for a company that built a language learning behemoth on gamification.
Duolingo believes that they will be able to get better margins in the latter half of the year due to two reasons:
First, they believe GenAI costs will keep going down. According to the CEO, “the cost of querying large language models is coming down kind of every month.”
“Every month” is a bit of an exaggeration, but it is true that GenAI costs have gone down drastically since launch – from $60 per million tokens to $5 per million tokens in the case of OpenAI for their latest models. However, in 2024, the pricing seems to have stabilized, especially for flagship models.
There are other models that are cheaper than the flagship ones, but in my experience at Class Central testing various models across different providers, the flagship models tend to perform the best. For Duolingo, the requirements are even more intense: they require real-time Audio APIs which are significantly more expensive.

There are signs that OpenAI itself is feeling the pinch. They recently launched a $200/month subscription, and their latest flagship model, GPT-4.5, costs $75 per million tokens.
It is not certain that OpenAI will keep cutting costs, as the company itself is a big cash guzzler. The question is how tied Duolingo is to OpenAI’s APIs. Is it even possible for them to switch to other providers if they are cheaper?
The second reason Duolingo expects improved margins is that they think there are many optimizations they can make to reduce costs. But these optimizations might also impact user experience and eventually user revenues.
If Duolingo Max is the future of Duolingo, then investors are right to be cautious.
Slower Projected Growth for 2025
Duolingo is projecting revenues of $962.5M – $978.5M in 2025, around a ~30% increase over 2024, which is lower than the 40% growth it achieved in 2024. Both are really impressive, but one of them is slightly less impressive.
In Duolingo’s defense, they had projected their 2024 revenues to be in the range of $717.5 – $729.5 and ended up achieving $20 million more than its high end.
But I have also seen things go the other way, as it did with Coursera, eventually leading to the departure of Coursera’s longtime CEO Jeff Maggioncalda. Last year, Coursera repeatedly lowered its earnings predictions for 2024. Ultimately, Coursera’s 2024 earnings grew by only 9%, reaching $695 million, far below the 15% growth the company had predicted at the start of 2024.
Tags
