Chinese artificial intelligence startup DeepSeek has disclosed financial estimates for its AI models, claiming a theoretical daily cost-profit ratio of up to 545%. However, the company clarified that actual revenue figures are significantly lower due to various factors.
This is the first time the Hangzhou-based company has revealed profitability data for its inference processes—the stage where trained AI models perform tasks like chatbot responses. The announcement could further impact global AI stocks, which experienced a sell-off in January when DeepSeek’s R1 and V3 models gained widespread popularity.
DeepSeek’s Competitive Edge
DeepSeek previously stated that training its AI models cost less than $6 million, far below the massive investments made by U.S. AI firms such as OpenAI. The Chinese firm primarily used Nvidia’s H800 chips, which are less powerful than the high-end chips utilized by U.S. competitors. This has raised questions about the necessity of massive spending on advanced hardware in AI development.
Cost and Revenue Estimates
In a GitHub post, DeepSeek estimated that renting one H800 chip costs around $2 per hour, leading to a total daily inference cost of approximately $87,072 for its V3 and R1 models. Theoretical daily revenue is projected at $562,027, which translates to an annual revenue of over $200 million.
However, DeepSeek acknowledged that actual earnings are considerably lower due to several reasons:
The V3 model is more cost-effective than the R1 model.
Some services remain free for web and app users.
Developers receive discounted rates during off-peak hours.
Implications for AI Market
DeepSeek’s efficiency and lower costs have intensified scrutiny on U.S. AI firms that have committed billions to chip purchases and infrastructure. Investors are now questioning whether such large-scale spending is necessary for profitability in the AI sector.
As competition in the AI space continues to evolve, DeepSeek’s cost-effective approach could reshape industry expectations on operational expenses and profitability.
