OpenAI turns to Google's AI chips to power its products, source says | Reuters

According to a source, OpenAI has begun utilizing Google's AI chips to power its products, including ChatGPT. This marks a significant shift for the company, which has primarily relied on Nvidia's GPUs for both training AI models and inference computing. This move signifies OpenAI's first meaningful use of non-Nvidia chips and indicates a strategic move away from exclusive reliance on Microsoft's data centers, its primary backer.

This collaboration between OpenAI and Google, two major competitors in the AI field, involves OpenAI renting Google's in-house tensor processing units (TPUs) through Google Cloud. Google has been expanding the external availability of its TPUs, which were previously reserved for internal use, and this deal helps solidify its position in the AI hardware market.

This move also allows Google to add a major player like OpenAI to its customer base, demonstrating its ability to leverage its in-house AI technology to expand its cloud business. The report suggests that OpenAI hopes to leverage Google's TPUs to potentially lower the cost of inference, a key aspect of running AI models.

However, the article also notes that Google is not renting its most powerful TPUs to OpenAI, according to The Information. This collaboration represents a strategic move for both companies, with OpenAI diversifying its hardware resources and Google expanding the reach of its AI chip technology.

The implications of this partnership are significant, potentially impacting the competitive landscape of the AI hardware market. By using Google's TPUs, OpenAI is exploring a potentially cheaper alternative to Nvidia's GPUs. This shift could influence the pricing and availability of AI chips and further accelerate the development and deployment of AI technologies.

The collaboration highlights the evolving dynamics of the AI industry, where competition and collaboration are becoming increasingly intertwined.