OpenAI CFO says cloud giants have been 'learning on our dime,' in warning shot over data center ambitions

In a recent statement, OpenAI's Chief Financial Officer, Sarah Friar, expressed concerns about the company's reliance on major cloud providers, asserting that these tech giants have been 'learning on our dime.' This remark was made during a presentation at a Goldman Sachs conference in San Francisco, where she emphasized the need for OpenAI to safeguard its intellectual property amidst increasing competition. OpenAI has significant computing demands due to its popular ChatGPT service and expanding enterprise AI initiatives. Currently, the startup collaborates with leading cloud providers such as Microsoft, Google, Oracle, and CoreWeave. However, Friar highlighted the importance of ensuring that OpenAI’s proprietary knowledge is not inadvertently shared with these partners as they assist in shaping AI data center operations. To address this issue, OpenAI is considering constructing its own AI data centers, a move Friar referred to as 'first-party builds.' She outlined a three-step strategy for this transition. Initially, OpenAI plans to purchase computing resources from existing cloud service providers. The next step involves closer collaboration with these partners to learn the intricacies of data center design and management. Ultimately, Friar anticipates that OpenAI will increasingly transition to building its own data centers. This strategic shift could pose a long-term threat to cloud providers, as OpenAI's growing independence may transform it from a customer into a competitor. Presently, AI companies are eager for computing power and are willing to invest heavily in cloud infrastructure. However, the prospect of OpenAI establishing its own facilities may not materialize for several years due to the lengthy construction timeline and limited availability of essential resources like AI chips, energy, and labor.