DeepSeek’s AI Models Claim Theoretical 545% Daily Profit Margin

Chinese artificial intelligence startup DeepSeek has unveiled data suggesting its V3 and R1 models could achieve a theoretical daily cost-profit ratio of 545%. This revelation has significant implications for the AI industry, particularly concerning operational efficiencies and cost structures.

According to DeepSeek’s disclosure, the daily operational cost for these models amounts to $87,072. This figure is based on the rental expense of Nvidia’s H800 graphics processing units , priced at $2 per hour. In contrast, the models are projected to generate daily revenues of $562,027, leading to the stated 545% cost-profit ratio.

However, DeepSeek has cautioned that actual revenues may be lower due to various factors, including different model costs, free access via web and app platforms, and reduced developer fees during off-peak hours.

The company’s pricing strategy has been notably aggressive. DeepSeek recently announced discounts of up to 75% for developers during off-peak hours, specifically from 1630 GMT to 0030 GMT. This move aims to attract a broader developer base and challenge global competitors by offering more affordable integration options.

Beijing has expressed support for DeepSeek’s advancements, emphasizing that the company seeks to complement, rather than compete directly with, U.S. AI giants like OpenAI and Google DeepMind. The Chinese embassy in Washington highlighted the collaborative potential of global AI developments, reflecting a preference for cooperation over competition.

DeepSeek’s models have seen rapid adoption across various sectors in China, including hospitals, local governments, car manufacturers, and state-owned enterprises. This widespread integration is attributed to the models’ low deployment costs and the company’s open-source strategy, which have facilitated seamless adoption. Notably, endorsements from President Xi Jinping and major Chinese corporations like Tencent, BYD, and Great Wall Motor have further propelled DeepSeek’s prominence in the AI landscape.

The company’s cost-efficient approach has raised questions about the high expenditure models prevalent among U.S. AI firms. DeepSeek’s ability to deliver competitive performance using less powerful and more affordable hardware challenges the existing paradigms of AI development and deployment costs.