https://duolingodata.com

2026/09/14:

Extended clarification about the prediction from 2026/09/12: The prediction is not that Duolingo will abandon the language pairs that already exist. Quite the opposite: existing courses will continue to receive updates and improvements, increasingly generated with the help of AI.

What I believe is going to change is the creation of new language pairs. In other words, after the current massive expansion, Duolingo may have reached the point where it no longer makes sense to keep creating new combinations such as Spanish-from-Marathi, Japanese-from-Kannada, Italian-from-Bulgarian, etc.

The exception would be courses in which English is the target language. I believe Duolingo will continue expanding English from additional source languages, because English represents a fundamentally different commercial opportunity and is the one language that can realistically be offered as a global target.

Therefore, the future model I am predicting is not:

"Duolingo stops teaching Spanish, French, Chinese, Japanese, etc."

but rather:

"Duolingo stops creating new language pairs, except English-from-X."

The existing language pairs will remain available and will continue evolving. However, the crucial difference is that human resources will no longer be allocated specifically to each individual course. AI will increasingly be responsible for generating, adapting, expanding and updating the existing content.

This would explain why Duolingo can maintain its enormous current catalogue without having to maintain a corresponding number of human teams. Once a course has reached a sufficiently advanced and stable state, AI can keep it alive at a fraction of the previous cost.

This interpretation also makes the transition much less dramatic for users. Someone learning Spanish from French, Japanese from Spanish, or Chinese from German would not suddenly lose their course. They would simply be using a course whose future development is increasingly automated.

Duolingo China appears to be following a somewhat different path. Its activity suggests that it may continue creating language courses both from and to languages relevant to the Chinese market. This makes sense commercially: the Chinese version of Duolingo may have its own strategic priorities and a sufficiently large local market to justify investments that would not necessarily make sense for the global platform.

So, to avoid any misunderstanding, my prediction is about Duolingo's global strategy, not about the disappearance of existing courses.

If this is indeed the strategy Duolingo is moving toward, the enormous expansion of 2025-2026 may therefore be understood not as the beginning of an endless multiplication of language pairs, but as the completion of the global language catalogue before Duolingo moves its human resources toward English, AI-powered learning, and non-language subjects such as Chess, Math, Music, Paint, Science, etc.

These are predictions I publish based on what I see in the code and on the company's commercial viability. This does not mean that I like it or that I want it to happen, nor does it mean the opposite.