Oracle has spent the past few years trying to convince the technology industry that it belongs in the same conversation as the biggest cloud infrastructure providers. Its latest quarter gives that argument considerably more weight.
The company reported on September 10 that Oracle Cloud Infrastructure revenue rose 121% year over year to $7.4 billion in its fiscal 2027 first quarter. More strikingly, Oracle said it booked more than $30 billion in additional AI cloud contracts during the quarter, taking its remaining performance obligations — contracted revenue that has not yet been recognised — to $664 billion.
That backlog was up $209 billion from a year earlier. Oracle’s total quarterly revenue climbed 30% to $19.3 billion, while overall cloud revenue, including infrastructure and applications, reached $11.6 billion, up 62%.
Those are unusually large numbers even by the standards of the current AI infrastructure boom. They also tell a broader story: the race to supply computing power for AI is no longer simply a contest over who can design the best model or sell the most GPUs. It is increasingly about who can actually build enough data-centre capacity to run them.
Oracle is becoming an AI infrastructure company
Oracle said demand for AI training and inference continues to grow faster than supply. Since the end of its previous quarter, the company says it has delivered more than 300,000 GPUs to AI cloud customers and nearly tripled the capacity delivered in the preceding quarter.
It also brought another 850 megawatts of data-centre capacity online during the quarter.
That physical build-out is important. Generative AI services can feel almost weightless from the user’s side of a browser or phone, but behind every model sits an enormous collection of GPUs, networking equipment, cooling systems, power infrastructure and data centres. The companies capable of putting that infrastructure online quickly have become a crucial layer of the AI economy.
For years, Amazon Web Services, Microsoft Azure and Google Cloud have dominated discussions about hyperscale cloud computing. Oracle is still smaller, but 121% infrastructure growth makes it increasingly difficult to treat OCI as an afterthought — particularly for large AI workloads.
The $664 billion number needs some context
A backlog is not the same thing as revenue sitting in Oracle’s bank account today. It represents contracted future business, and converting those commitments into recognised revenue depends partly on Oracle delivering the infrastructure customers have ordered.
That distinction matters because the AI data-centre race is extraordinarily expensive.
Oracle reported negative free cash flow of roughly $5 billion for the quarter as it continued investing in cloud infrastructure. It also completed a $20 billion common-stock sale during the quarter as part of its capital investment programme.
Investors have consequently been watching not only how many AI contracts Oracle can sign, but whether the company can finance the infrastructure needed to fulfil them without continually putting more pressure on its balance sheet.
The latest contracts offer an interesting answer. Oracle says their structure does not add to its planned capital-raising requirements. Reuters reported that many of the new orders involve mechanisms such as customer prepayments or customers supplying their own hardware, reducing the amount Oracle itself must spend upfront on chips.
That may be almost as significant as the headline backlog. Selling enormous amounts of future AI capacity is impressive; finding ways to build it without carrying the entire hardware bill is potentially much more sustainable.
The AI boom is becoming a construction race
There is a tendency to measure the AI competition through model benchmarks: which chatbot reasons better, which model codes faster, or which company has the newest release. Oracle’s quarter is a reminder that another competition is happening underneath all of that.
AI companies need somewhere to run those models.
Oracle is betting heavily that it can become one of the companies providing that place. The company expects roughly half of its current backlog to turn into sales within the next 36 months, according to Reuters. For fiscal 2027, Oracle now expects at least $90 billion in total revenue.
There are still risks. Data-centre projects face constraints ranging from electricity supply and permitting to construction timelines, cooling and access to advanced chips. A $664 billion backlog therefore represents both an extraordinary opportunity and an enormous delivery obligation.
For everyday technology users, the effects will mostly be indirect. More infrastructure competition can give AI developers more places to train and serve models, potentially improving capacity and putting pressure on cloud pricing. It can also determine which companies become indispensable behind services consumers use every day without ever seeing the data centres underneath them.
Oracle’s latest numbers suggest that the next phase of the AI race may be decided as much by megawatts, GPUs and concrete as by algorithms.
Sources: Oracle Q1 FY27 results; Reuters.
