OpenAI has recorded its first month with more than $1 billion in revenue, but Chief Financial Officer Sarah Friar warned the company is still “constantly under compute,” citing an acute shortage of graphics-processing units needed to train and run its artificial-intelligence models. Friar told CNBC that demand for GPUs is “voracious” as the industry remains in what she called the early infrastructure era of AI. Her remarks echo Chief Executive Officer Sam Altman’s push to amass more than 1 million GPUs by the end of 2025, underscoring the strategic importance of hardware supply in the intensifying race to develop larger and faster models. To address the bottleneck, OpenAI has launched Stargate, a roughly $500 billion data-center project being built in partnership with Oracle and SoftBank. The initiative is designed to secure long-term compute capacity—largely powered by Nvidia chips—and support the company’s ambition to achieve artificial general intelligence. The scale of the investment highlights how access to high-performance silicon has become the pivotal constraint for leading AI developers.