DeepSeek has publicly claimed that, under idealized conditions, its cost of running inference (the phase where the AI processes user queries) is so low compared to its theoretical revenue that it would result in a “cost profit margin” of 545%[1][3][4][5]. In other words, if all token usage were billed at its premium rates, the revenue generated in a 24‑hour period would be about 545% greater than the operating costs.
However, DeepSeek itself has clarified that this figure is purely theoretical. They note that actual revenues are “substantially lower” because not all services are monetized (for example, free web and app access is maintained), and various discounts—such as those for off‑peak usage—and differential pricing between its models reduce the real-world amounts earned[1][2][4][5].
So, while the 545% figure is based on their internal cost–to–revenue calculations, it does not mean that the company is actually realizing a 545% profit in practice.
Get more accurate answers with Super Search, upload files, personalized discovery feed, save searches and contribute to the PandiPedia.
Let's look at alternatives:
- Modify the query.
- Start a new thread.
- Remove sources (if manually added).
- Request a manual search from our human research team.
