Oracle reported record quarterly results on September 10, 2026 for its fiscal first quarter of 2027, with total revenue up 30% year over year to $19.3 billion and cloud infrastructure revenue up 121% to $7.4 billion. On the earnings call, Oracle disclosed that its AI infrastructure ran at 97.9% utilization during the quarter, delivering 850 megawatts of AI capacity built on more than 300,000 Nvidia graphics processing units.
Oracle also disclosed it had completed a $20 billion equity issuance and expects capital expenditures of $90-95 billion for the fiscal year, largely tied to AI infrastructure expansion.
Why it matters
Utilization figures near 98% are unusually high for any large computing fleet and indicate essentially no spare capacity, which analysts have interpreted as direct evidence that Nvidia's GPU supply, rather than customer demand, is the binding constraint on Oracle's cloud growth. Because Oracle discloses this kind of operational detail more explicitly than some peers, the figure has become a widely cited proxy for the broader industry-wide GPU shortage.
How it works
Oracle's cloud infrastructure business rents out GPU capacity to AI labs and enterprises, so utilization measures how much of that capacity is actively in use versus idle. A rate close to 100% means Oracle could sell essentially all the AI compute it can provision, and any further revenue growth in that segment depends on receiving more chips from Nvidia and building out more data-center power capacity, rather than finding more customers.
Evidence
The Motley Fool's analysis of the earnings call directly ties the 97.9% utilization figure and the 300,000-GPU, 850-megawatt disclosure to Oracle's reported results, framing it as evidence of continued Nvidia-side supply constraints. Oracle's own earnings call transcript confirms the 121% year-over-year growth in cloud infrastructure revenue and the raised full-year capital expenditure guidance.
The competing read
High utilization could also reflect deliberate capacity planning rather than pure scarcity; cloud providers generally try to keep utilization high as a matter of efficient capital allocation, so a figure near 98% does not by itself prove that Oracle is turning away demand it could otherwise serve. Some of the framing around a GPU 'shortage' in 2026 comes from research and analyst commentary that has a financial interest in dramatizing scarcity narratives, and hard, independently verified order-to-delivery lead times were not part of Oracle's own disclosures.
What happens next
Oracle's raised capital expenditure guidance of $90-95 billion for the fiscal year signals it expects to keep adding GPU capacity aggressively, meaning utilization rates and backlog figures in coming quarters will be a continued signal of whether Nvidia's supply is keeping pace with hyperscaler ordering. Investors and rivals will be watching Oracle's next earnings call for whether utilization eases as new capacity comes online or stays pinned near its ceiling.