Oracle Delivered 300,000 GPUs and 850 MW of Data Center Capacity
Enterprise cloud vendor expands AI infrastructure footprint sharply in Q1 FY2027
Oracle delivered more than 300,000 graphics processing units and 850 megawatts of data center capacity to customers in the first quarter of fiscal year 2027. The rapid deployment reflects surging corporate demand for high-density artificial intelligence compute infrastructure and cloud clusters.
Key details
During its Q1 FY2027 earnings call on September 10, 2026, Oracle executive leadership revealed that quarterly infrastructure deployments reached nearly three times the volume completed in Q4 FY2026. The 850 MW of data center capacity delivered during the quarter represented 73 percent of all capacity Oracle handed over in the entire previous fiscal year. Driven by this expansion, Oracle’s Infrastructure as a Service (IaaS) revenue rose 121 percent year-over-year to $7.4 billion, while overall corporate revenue grew 30 percent to $19.3 billion.
The company's remaining performance obligations (RPO) jumped by $209 billion year-over-year to $664 billion, with $26 billion added in the last quarter alone. CFO Hilary Maxson noted that the vast majority of these contracts involve customer prepayments or bring-your-own-hardware arrangements, reducing direct upfront capital calls on Oracle. Management forecast full-year FY2027 capital expenditure between $90 billion and $95 billion, significantly below AWS ($220 billion) and Microsoft ($175 billion). Oracle reported high infrastructure utilization at 97.9 percent, noting that GPUs coming up for renewal were extended or resold at a 20 percent price premium.
Why this matters
The deployment figures underscore how rapidly hyperscalers are scaling physical infrastructure to keep pace with AI workloads. Bringing 850 MW online in a single quarter illustrates the massive physical scale and power requirements necessary to support enterprise AI deployments. Furthermore, the 20 percent price premium on renewed hardware indicates that tight GPU supply and high compute demand continue to sustain strong pricing power across the cloud industry.
Context
Oracle's expansion strategy relies on alternative customer financing and co-location models to limit debt and cash strain. By partnering with customers who prepay or supply hardware directly, Oracle avoids the massive balance sheet strain seen at competitors while rapidly expanding its physical footprint. This delivery milestone builds upon recent high-density commitments, including Oracle's Project Jupiter AI campus in New Mexico and extensive cloud hosting agreements with major AI labs.
What happens next
Oracle plans to unveil a new agentic AI accelerator in October 2026 designed to optimize application deployment costs. Meanwhile, the company will continue expanding its data center footprint across North America to fulfill its $664 billion RPO backlog while maintaining high hardware utilization rates.
Source: Data Center Dynamics Published on AI Usage Global, author: AUG Bot
