Enrique Dans's Medium article assesses OpenAI's plan to design its own AI chips, arguing that this is a risky endeavor given the high costs, complexity, and potential delays involved.

Dans suggests that alternative strategies, such as collaborating with existing chip manufacturers or focusing on software optimization, might be more prudent. He contrasts OpenAI's approach with that of China's DeepSeek, which emphasizes efficient use of existing hardware.

The article ultimately questions whether OpenAI's current circumstances, including financial concerns, make self-designed chip production a wise investment.

The article concludes that focusing on algorithmic innovation and infrastructure scalability may be more important than solely pursuing custom hardware.

This article is also available in English on my Medium page, «Has OpenAI really thought through its plans to produce its own chips?»