OpenAI and Oracle are partnering to add 4.5 gigawatts of datacenter capacity in the U.S. as part of the Stargate project. Announced in a recent blog post, this moves Oracle’s total provision of capacity to over five gigawatts, supporting more than two million GPUs. While OpenAI plans significant investment in AI infrastructure, the financial logistics remain unclear. Current revenue projections don’t match the scale of investment, with the total cost potentially soaring to $100 billion, considering GPUs and energy needs.
Despite this, over 100,000 jobs are anticipated from the project, though many roles may be short-lived or phased out by AI advancements. Oracle has already delivered Nvidia GB200 racks for the Stargate I datacenter in Abilene, Texas, indicating initial progress. With Oracle’s reported $40 billion expenditure, a phased approach is expected, dependent on power availability and demand.
While Oracle plays a significant part, SoftBank’s involvement in this phase is uncertain. Notably, OpenAI’s ambitious developments continue amidst broader AI industry trends and market pressures.