DeepSeek says it could earn 5 times more than what it spends: What does it mean

The Chinese IT firm DeepSeek, which unveiled its ground-breaking R1 model in January, has disclosed that its "theoretical" profit margin may exceed five times its expenses. This is one of the few occasions when the public has been given access to something that closely resembles the true costs of creating and maintaining an AI model.

With its creative and low-cost method of creating AI models, the new startup caused a stir and a stock market meltdown in the global technology sector. On February 28, it said that the cost of inferencing to sales during a 24-hour period for its V3 and R1 models put profit margins at 545%.

The processing power, electricity, data storage, and other resources required to enable real-time AI model operation are referred to as inferencing.