Oracle’s AI Bet Starts Paying Off With $30 Billion in New Cloud Deals

Oracle Finally Gives Wall Street the Number It Wanted

For months, Oracle has been spending like a company building a small digital country. Data centers, networking equipment and mountains of AI hardware do not come cheap. Investors understood the ambition. They simply wanted proof that customers would eventually show up with equally enormous checks.

Oracle’s latest quarterly results delivered that proof—or at least a convincing first installment.

The company booked more than $30 billion in new AI cloud contracts during its fiscal first quarter. That helped push its remaining performance obligations, commonly treated as a measure of contracted backlog, to a record $664 billion. Wall Street had expected roughly $640 billion, according to Visible Alpha figures cited by Reuters.

That is not pocket change hiding behind the sofa. It represents a gigantic queue of business that Oracle expects to turn into revenue over time.

The company also reported quarterly revenue of approximately $19.3 billion, up 30% from a year earlier. Adjusted earnings reached $1.92 per share, beating the $1.74 analysts expected.

The Cloud Division Hits the Accelerator

The most eye-catching growth came from Oracle’s cloud infrastructure business, which rents computing capacity to customers running demanding workloads.

Revenue from that unit reached $7.4 billion during the quarter. That marked a 121% increase from the same period a year earlier, according to Barron’s. A business that once lived in the shadow of Amazon Web Services, Microsoft Azure and Google Cloud is suddenly moving at sports-car speed.

AI created much of that momentum.

Training frontier models requires huge clusters of advanced processors. Running those models for millions of users—known as inference—creates another voracious source of demand. Oracle supplies the data centers, networking and cloud platform needed to keep both activities moving.

The company said demand for its AI cloud training and inference services continues to grow faster than available supply. That detail matters. Oracle is not trying to invent demand through clever accounting or optimistic forecasts. It is racing to install enough capacity to serve business already knocking on the door.

The company also brought 850 megawatts of data-center capacity online during the quarter, according to the Financial Times. This was an industrial-scale expansion, not a few server racks wheeled into a basement.

A $664 Billion Backlog Changes the Conversation

Backlog does not equal cash in the bank. It consists of contracted revenue that Oracle has not yet recognized, and delivery can stretch across several years. Even so, $664 billion dramatically improves the visibility of Oracle’s future business.

The company expects roughly half of that amount to convert into revenue within 36 months. If Oracle delivers the required computing capacity on schedule, its recent growth could have a long runway.

The composition of the latest contracts may be even more important than their headline value. Oracle said most of its new AI orders use arrangements such as customer prepayments or bring-your-own-hardware structures. Under those deals, customers help fund construction or provide some of the equipment installed in Oracle facilities.

That reduces the amount Oracle must spend before it can start earning revenue. It also gives customers some skin in the game—a useful feature when data-center projects involve billions of dollars and enough electricity to make a utility executive reach for a second calculator.

Chief Financial Officer Hilary Maxson said the vast majority of the new orders would not require incremental capital from Oracle. The business can therefore grow without automatically forcing the company to raise its already towering investment budget.

This is the part investors had been waiting to hear. Big contracts are exciting. Big contracts that arrive with financing attached are considerably easier to love.

OpenAI Is Important, but It Is Not the Whole Plot

Oracle’s relationship with OpenAI remains central to its AI transformation. The companies are involved in the Stargate infrastructure effort, and OpenAI has agreed to purchase vast amounts of computing capacity from Oracle over several years.

That concentration also creates risk. If a major customer delays deployments, changes suppliers or struggles to finance its commitments, Oracle could feel the impact. Investors have repeatedly questioned whether the company depends too heavily on a handful of AI laboratories.

The latest results offered a partial answer. Oracle’s non-OpenAI backlog has more than doubled over the past year, according to Barron’s. In other words, the growth story is getting broader.

That breadth matters because enterprise AI demand looks different from frontier-model training. Banks, retailers, healthcare groups and manufacturers often need secure access to databases, applications and specialized computing. Oracle already supplies many of those organizations with critical business software. It can now pitch AI infrastructure alongside products customers already use.

OpenAI may be the enormous rocket strapped to Oracle’s cloud business. But Oracle increasingly wants to prove that it has more than one engine.

The Spending Is Still Enormous

Now for the number that prevents this earnings report from turning into a full confetti parade: $28.5 billion.

That was Oracle’s capital expenditure during the quarter, up from $8.5 billion a year earlier. The company still expects to spend between $90 billion and $95 billion across fiscal 2027, much of it on data centers and AI infrastructure.

To put the quarter in perspective, Oracle spent more on long-lived assets than it generated in total revenue. This is what happens when a software veteran decides it also wants to become a hyperscale infrastructure powerhouse. The transformation requires land, buildings, power agreements, chips and networking gear. None accepts payment in enthusiasm.

The investment drove free cash flow to negative $5.4 billion. That sounds alarming in isolation, but analysts had expected an outflow of roughly $9.6 billion, according to Reuters. Oracle therefore burned far less cash than feared.

Customer prepayments covered about $11.36 billion of the quarter’s capital spending. That disclosure helped explain why Oracle maintained its annual investment forecast even while adding tens of billions of dollars in contracts.

The spending remains a genuine risk. The difference is that Oracle now has better evidence that revenue can catch up.

Wall Street Sees the Beginnings of an AI Return

Investors responded quickly. Oracle shares rose in extended trading after the results, with reports placing the initial gain between roughly 4% and 8% as trading moved.

The market was not celebrating spending for spending’s sake. It was reacting to signs that Oracle’s infrastructure investments had started producing measurable returns.

Revenue and earnings beat expectations. The backlog surpassed forecasts. Cloud infrastructure more than doubled, while negative free cash flow came in billions of dollars better than analysts had modeled.

Oracle shares had faced sustained pressure as investors worried about rising debt, weak cash generation and delays affecting large data-center projects. S&P Global downgraded Oracle’s credit rating in July, citing cash-flow weakness and increased business risk. Questions surrounding labor, permitting and power availability added another layer of uncertainty.

Those concerns have not vanished. One strong quarter cannot magically build a data center or erase a debt balance. Yet the results showed that Oracle’s AI strategy is progressing from promises and construction schedules toward operating revenue.

As Investopedia noted, the company delivered record quarterly revenue and an outlook that exceeded expectations. Oracle now forecasts fiscal 2027 revenue of at least $90 billion and adjusted earnings of $8.10 per share, up from $8.05. For the current quarter, it projects revenue growth of 30% to 34%.

Why Oracle Has Become an Unexpected AI Powerhouse

Oracle did not enter the cloud era as the obvious winner. Amazon established an early lead. Microsoft used its enterprise relationships to expand Azure. Google brought formidable engineering and AI research credentials.

Oracle, meanwhile, remained best known for databases, business applications and licensing negotiations capable of aging an IT manager several years before lunch.

AI changed its opportunity.

Frontier laboratories need enormous, specialized computing clusters. They also want alternatives to the largest cloud platforms, both for extra capacity and bargaining power. Oracle designed its cloud infrastructure around high-speed networking and large GPU deployments, making it attractive for model training.

Oracle’s smaller starting position may even offer an advantage. Its infrastructure revenue can grow rapidly because it begins from a lower base than AWS or Azure. A few massive contracts can visibly reshape the entire company.

The latest quarter captured that transformation in one statistic: cloud infrastructure grew 121%, while the remainder of Oracle expanded much more slowly. The old software machine still pays the bills. The AI cloud business increasingly determines where the company goes next.

The AI Boom Is Moving From Training to Deployment

Oracle’s results also reveal something larger about the AI market. Demand is no longer limited to a few headline-grabbing training runs.

Companies now need infrastructure for inference—the process of serving trained models to actual users. Every chatbot response, coding suggestion, automated workflow and AI search consumes computing resources. A successful product can generate relentless demand, one prompt at a time.

That favors cloud providers capable of supplying processors, networking and data services at scale. It also creates recurring usage instead of a single burst of training activity.

Oracle specifically highlighted growth in both AI training and inference services. That pairing strengthens the story. Training produces spectacular contracts, but inference can create durable consumption as models enter daily business operations.

This does not guarantee that every AI project will succeed. Plenty will remain stuck in pilot programs, and some expensive models will discover that their main talent is generating meeting notes nobody reads. But Oracle’s backlog indicates that major customers are moving real money into the market.

The AI boom is becoming an infrastructure business. Oracle has positioned itself near the electrical socket.

The Risks Have Not Left the Building

The upbeat quarter deserves attention, but optimism needs guardrails.

First, backlog is not revenue. Oracle must build the capacity, deliver the service and keep customers satisfied before those contracted dollars flow through its income statement.

Second, the company carries substantial financial obligations while funding one of the technology industry’s most aggressive construction programs. Heavy borrowing, customer concentration and negative free cash flow leave less room for major errors.

Third, physical limits can interrupt even the grandest AI strategy. Power shortages, permitting disputes, equipment delays and construction bottlenecks have already complicated data-center expansion across the industry.

The Financial Times described Oracle’s shift as both expensive and risky, even as the latest results showed clear progress. That tension belongs in any honest account of the company’s quarter.

Oracle has not completed its AI transformation. It has simply produced its strongest evidence yet that the enormous wager might work.

From Database Giant to AI Infrastructure Contender

Oracle’s fiscal first quarter did not settle every argument surrounding the AI investment boom. It did, however, move the conversation forward.

The company generated record revenue. Its cloud infrastructure unit more than doubled. Customers signed over $30 billion in additional AI cloud contracts. The backlog reached $664 billion, and customers helped fund a meaningful portion of the infrastructure needed to serve them.

Most importantly, Oracle showed that its AI expansion can produce growth before the spending spree ends.

The next challenge is delivery. Oracle must finish data centers, install equipment, secure power and convert contracted demand into revenue without allowing debt and cash burn to overwhelm the business. That is a formidable to-do list. Larry Ellison probably will not be clearing it with a quiet afternoon and a fresh cup of coffee.

Still, the latest numbers justify measured optimism. Oracle is no longer merely promising that AI will reshape its cloud division someday. The reshaping has begun, and it is appearing in revenue, earnings and signed contracts.

For years, the AI infrastructure contest looked like a three-company race among Amazon, Microsoft and Google. Oracle has now barged into the frame carrying a $664 billion backlog and asking everyone to make room.

Subtle? Not remotely. Effective? This quarter, absolutely.

Sources

The Kingy Brief

One consequential launch, one pricing, limit, or shutdown change, one hands-on test, one exact prompt or Test Pack, and one try / watch / skip verdict.

Free · Choose your subjects · Double opt-in · Unsubscribe anytime

Form not showing? Subscribe on Beehiiv.