OpenAI, the company behind advanced artificial intelligence, is planning a significant change in how it gets the computing power it needs over the next five years. According to a report from The Information, OpenAI expects that by the year 2030, a large portion—about 75%—of its computing resources will come from a new project called Stargate. This project is likely to receive substantial investment from SoftBank, a major financial supporter of OpenAI. Currently, most of OpenAI’s computing power comes from Microsoft, which is also a big investor in the company.
However, this transition to relying more on Stargate will not happen immediately. In the coming years, OpenAI still intends to increase its spending on Microsoft’s data centers to meet its current needs.
At the same time, OpenAI anticipates that its overall expenses will rise sharply. The report indicates that by 2027, the company expects to lose about $20 billion in cash, which is significantly higher than the $5 billion it reportedly spent in 2024. Additionally, by 2030, OpenAI believes that the costs associated with running its AI models—referred to as inference—will surpass the expenses related to training those models. This means that while OpenAI is shifting its computing sources, it is also preparing for much higher costs as it continues to develop and operate its AI technologies.
