Oracle’s latest results reveal a striking contradiction at the heart of the AI boom. Demand for the company’s cloud infrastructure is surging — but the cost of supplying that computing power is rising just as fast.
The company reported $553 billion in contracted cloud backlog, an extraordinary figure that reflects how aggressively companies are locking in AI computing capacity. But meeting that demand is forcing Oracle into one of the largest infrastructure spending cycles in its history.
The effect is already visible in the company’s cash flow.
Over the past year Oracle’s free cash flow has swung from roughly $11 billion to negative $24 billion, as capital spending on AI data centers accelerates. For a company long associated with high-margin enterprise software, it is a dramatic shift.
The reason lies in the scale of the AI build-out now sweeping the technology industry. Training and running large AI models requires vast quantities of specialized chips, electricity and data-center capacity — turning cloud computing into one of the most capital-intensive businesses in tech.
Oracle’s latest earnings suggest that transformation is already well underway.
A $553B signal from the AI market
Oracle reported third-quarter revenue of $17.2 billion, up 22% year-over-year, while cloud revenue rose 44% to $8.9 billion.
But the most revealing number in the release was the company’s remaining performance obligations, which surged to $553 billion, up 325% from a year earlier.
That figure represents cloud services already contracted but not yet delivered — effectively a pipeline of future revenue tied largely to AI infrastructure demand.
The scale is striking. Oracle expects about $67 billion in total revenue for fiscal 2026, meaning its contracted backlog is roughly eight times its annual revenue. Such commitments suggest companies developing AI systems are securing computing capacity years in advance rather than relying on on-demand cloud access.
Oracle also disclosed an unusual feature of some of those deals: customers may prepay for hardware purchases or supply GPUs themselves, helping guarantee access to scarce AI computing power.
The scramble for compute is beginning to reshape how cloud infrastructure is financed.
Cloud growth is now an infrastructure story
Oracle’s fastest-growing business is no longer its software — it is infrastructure.
Cloud infrastructure revenue jumped 84% to $4.9 billion, driven largely by demand for AI training and inference workloads.
Traditional software growth remains far slower. Software revenue rose just 3% in US dollars and declined 1% in constant currency, underscoring how much of Oracle’s recent momentum now comes from computing capacity rather than enterprise software licensing.
For decades, Oracle’s competitive strength lay in databases and enterprise applications. Today, the company’s growth increasingly depends on the data centers, chips and power capacity required to run artificial intelligence systems.
AI is shifting the economics of the cloud industry. What was once a software-driven business is becoming an infrastructure race.
The cost of the AI arms race
The transformation is already showing up in Oracle’s cash flow. Despite rising revenue and profits, free cash flow has swung sharply into negative territory as the company accelerates investment in AI infrastructure.
Oracle expects capital expenditures of roughly $50 billion in fiscal 2026, an extraordinary level of spending for a company whose historical business model depends on high-margin enterprise software rather than large-scale infrastructure.
The company’s cash flow statements show capital expenditures rising rapidly over the past year as Oracle expands data centers to support AI workloads.
To finance that expansion, Oracle has moved quickly into the capital markets. The company recently announced plans to raise up to $50 billion in debt and equity, and has already secured $30 billion through investment-grade bonds and mandatory convertible preferred stock.
The shift highlights a broader reality across the technology industry: the AI boom is turning cloud computing into a capital-intensive infrastructure race.
The entire industry is running the same race
Oracle is not alone in confronting the costs of the AI boom.
Across the technology sector, cloud providers are committing unprecedented sums to new infrastructure. Companies including Amazon, Microsoft and Google have all sharply increased capital spending as demand for AI computing accelerates.
The challenge is no longer purely technological. Running large AI models requires vast quantities of electricity, cooling systems and specialized data-center capacity. In some regions, the biggest constraint on new AI facilities is no longer computing hardware but power availability, a factor increasingly shaping where new data centers can be built.
Oracle’s results illustrate how access to computing infrastructure — not just software platforms — is becoming the decisive battleground in the AI race.
The risks behind the spending boom
The surge in AI demand may be transforming the cloud industry — but it also raises the stakes.
Building hyperscale infrastructure requires enormous upfront investment and years of development. Data centers must secure land, power supply, cooling systems and specialized hardware long before revenue is generated.
If AI demand slows or industry capacity expands too aggressively, cloud providers could find themselves competing on price for computing power that was expensive to build.
Oracle also faces competition from larger cloud providers that still dominate the global infrastructure market.
For now, however, the company’s $553 billion backlog suggests demand for AI computing capacity remains extremely strong.
What happens next?
Oracle expects the momentum to continue. The company forecasts fourth-quarter revenue growth between 19% and 21%, with cloud revenue expected to rise between 46% and 50%.
Management also raised its fiscal 2027 revenue outlook to $90 billion, signaling confidence that demand for AI computing capacity will continue expanding.
But the deeper message in Oracle’s earnings is structural.
Artificial intelligence is transforming cloud computing from a software-driven business into an infrastructure race. Companies that once relied on high-margin software are now pouring tens of billions into data centers, specialized chips and power capacity.
Oracle’s $553 billion backlog — and its rapidly shrinking free cash flow — illustrate just how expensive that transformation is becoming.
