OpenAI and Oracle Expand AI Infrastructure with 4.5GW Capacity Boost

OpenAI and Oracle are accelerating their AI infrastructure buildout with a new 4.5-gigawatt capacity expansion, pushing OpenAI's total planned capacity beyond 5 gigawatts. The project, part of OpenAI's Stargate initiative, will deploy over 2 million chips across undisclosed locations, with Texas, Michigan, Wisconsin, and Wyoming under consideration.

The collaboration builds on a $500 billion joint commitment made with SoftBank in January, which aimed to deliver 10 gigawatts of U.S. AI infrastructure within four years. OpenAI now expects to exceed that target, though SoftBank—reportedly facing financing challenges—is not participating in this phase. Oracle declined to comment on the development.

Meanwhile, OpenAI's Abilene, Texas facility is already operational, with partial compute capacity online. The rapid scaling underscores the insatiable energy demands of frontier AI models, a trend that could indirectly benefit energy-efficient blockchain projects as tech giants seek sustainable power solutions.