OpenAI Plans Own Data Centers Amid Ballooning Infrastructure Costs

OpenAI is transitioning toward building its own AI data centers, known as first-party builds, to protect its intellectual property. Chief Financial Officer Sarah Friar announced the shift during a Goldman Sachs conference in San Francisco, stating that cloud providers have been learning on the company's dime while helping design infrastructure for ChatGPT and enterprise services. Friar outlined a transition moving from buying off-the-shelf compute to independent buildouts, a move that could eventually position OpenAI as a competitor to partners like Microsoft, Oracle, and CoreWeave.

Financial scaling remains a primary focus as the company expects to spend $115 billion by 2029 on chips, computing power, and data centers. Despite high cash burn, OpenAI has nearly doubled its revenue to a $12 billion annualized run rate. Chief Revenue Officer Ashley Kramer stated at the Fortune Brainstorm Tech conference that the company is prioritizing a research-led product strategy over marketing buzz to drive enterprise value through actual usage.

Friar also warned that professionals and companies failing to adopt AI quickly will be left behind by those using the technology deeply. Meanwhile, OpenAI faces legal scrutiny in California and Delaware, where attorneys general are investigating whether the company's proposed transition to a for-profit structure violates charitable trust laws.