- Oracle is restructuring AI infrastructure deals so that customers prepay or provide their own hardware instead of Oracle funding everything on its balance sheet.
- This approach eases pressure from more than $50 billion in planned data center investments while still supporting rapid AI cloud expansion.
- The model has fueled strong growth in cloud infrastructure revenue and a sharp jump in remaining performance obligations to $553 billion.
- Despite ongoing concerns about debt, margins and execution in Oracle Cloud Infrastructure, investor sentiment has improved as risk perceptions cool.
Oracle is quietly reshaping how AI infrastructure gets paid for, moving a big chunk of the financial burden off its own books and onto its customers. Instead of taking on all the upfront costs of building massive data centers full of high-end chips, the company is structuring deals so that clients either pay in advance or directly fund the hardware they will run on.
Behind this shift lies a simple reality: building out the infrastructure needed for large-scale AI is enormously expensive. Oracle has signaled plans involving more than $50 billion in data center-related investment, a figure that had been raising eyebrows among investors worried about leverage, cash flow and long-term risk if AI demand were to cool unexpectedly.
How Oracle’s customer-funded AI infrastructure model works
At the heart of the strategy is a structure often described as a “bring your own cloud” style model, where the traditional funding roles are flipped. Rather than Oracle fully financing and carrying the cost of new AI infrastructure on its balance sheet, clients play a much more direct role in bankrolling the underlying capacity they intend to use.
In practical terms, the company is putting together agreements where customers prepay for access to AI capacity or even provide their own hardware, while Oracle supplies the platform, software, cloud environment and operations. This lets the firm expand its footprint of AI-ready data centers without having to shoulder every dollar of capital expenditure itself.
Under these arrangements, clients finance the hardware they depend on, either through upfront payments or by integrating their own equipment into Oracle’s cloud infrastructure. Those cash inflows reduce the need for Oracle to pile on additional debt, helping the company manage its leverage at a time of extremely capital-intensive growth.
Because the inflows are locked into long-term contracts, the structure also helps Oracle support a larger wave of AI infrastructure expansion without putting as much strain on its cash flow and reported debt metrics. The company can keep building capacity for customers while softening the impact on its own financial statements.
According to management commentary, this approach has already produced more than $29 billion in new contracts where the underlying AI infrastructure is essentially financed by clients, rather than being fully funded out of Oracle’s capital budget.
Financial impact and key metrics from the new AI funding approach
The effects of this customer-funded structure are starting to appear clearly in Oracle’s financial results for the third quarter of fiscal 2026. The company reported solid headline growth while emphasizing how AI-related contracts are reshaping its order book.
Total revenue for the quarter came in at around $17.19 billion, representing roughly 20% year-over-year growth. That topline expansion was supported in large part by a surge in cloud infrastructure activity, where demand for AI workloads is concentrated.
Within that segment, Oracle highlighted an 84% annual increase in cloud infrastructure revenue, underscoring how quickly its AI and GPU-centric offerings are scaling. This is where the new funding model is most visible, because many of these agreements include mechanisms where the client, not Oracle, covers much of the underlying infrastructure cost.
One of the most striking figures was the jump in remaining performance obligations (RPO), a key indicator of contracted but not yet recognized revenue. RPO climbed by about 325% year over year to approximately $553 billion in the third quarter, reflecting an expanding backlog of long-term commitments tied to cloud and AI services.
Within its evolving infrastructure-centric business, Oracle also reported that gross margins in this area reached around 32%. While not comparable to traditional high-margin software licensing lines, that level suggests the model can be economically viable even as the company continues to ramp up AI capacity at scale.
A shift from pure software toward AI infrastructure and GPU-based services
For decades, Oracle was known primarily as a software-focused company, centered on databases, applications and enterprise platforms. The current AI wave is pushing the business model toward something more hybrid, where infrastructure, GPUs and cloud services play an increasingly central role.
To compete in the AI era, Oracle has been rapidly building new data centers outfitted with high-end processors designed for training and running advanced AI models. Customers include prominent names such as large social platforms and cutting-edge AI labs that require massive compute resources.
These facilities are geared toward what is often described as GPU-as-a-service (GPUaaS), where clients rent access to powerful chips through Oracle Cloud Infrastructure rather than buying and maintaining the hardware themselves. The goal is to give organizations on-demand capacity for AI workloads without forcing them to construct their own specialized data centers.
That said, the shift is not without trade-offs. Infrastructure-heavy operations tend to involve lower margins and higher capital intensity compared with core software licensing. By pushing customers to finance more of the hardware, Oracle is attempting to preserve economic flexibility while still embracing the AI buildout that the market is demanding.
As this strategy unfolds, the mix of Oracle’s business is gradually tilting further toward infrastructure, cloud services and long-term AI capacity contracts, even though software remains a major pillar of its portfolio and value proposition.
Investor reaction, debt concerns and changing risk perception
Oracle’s aggressive push into AI data center expansion initially stirred unease among investors, particularly when the company signaled earlier in the year that it could raise up to $50 billion in debt and equity to expand capacity. That kind of funding need raised questions about leverage, credit risk and what might happen if AI spending slowed.
Those concerns were reflected in both the share price and the cost of insuring Oracle’s debt. Over the course of the year leading up to the latest results, the stock had fallen roughly 23%, with an even steeper decline of about 54.5% over the previous six months before sentiment began to shift.
Credit markets told a similar story. The cost of five-year credit default swaps (CDS) on Oracle debt had spiked to record levels of around 166 basis points earlier in the month, up sharply from roughly 40 basis points a year before, reflecting heightened concern about the company’s perceived credit risk.
Following the latest earnings and the clearer emphasis on customer-funded AI infrastructure deals, that risk premium started to ease. CDS spreads fell back to about 150 basis points, their lowest point since mid-February, suggesting investors were becoming a bit more comfortable with Oracle’s funding strategy and long-term commitments.
On the equity side, the stock finished one trading session at about $149.40 per share, down roughly 1.4% on the day, but then climbed close to 10-12% in pre-market and subsequent trading as investors digested the revenue outlook, the AI contract structure and the implications for future cash flow.
Analyst views on AI contracts, margins and Oracle Cloud Infrastructure
Market analysts have been dissecting Oracle’s AI strategy and funding arrangements to gauge how sustainable they might be. One key theme in their commentary is that shifting infrastructure costs to customers can help Oracle commit to more future revenue without taking on the full financial burden.
Some equity research voices have highlighted that, with many new AI contracts structured around prepayments or customer-supplied hardware, Oracle can grow its backlog and revenue visibility while limiting the amount of capital it needs to deploy upfront. That trade-off is particularly important given the multibillion-dollar scale of its AI buildout.
At the same time, analysts caution that the conversation about how all this expansion is funded is far from over. Questions remain about the long-term impact on profit margins, free cash flow and the overall risk profile of Oracle Cloud Infrastructure as the company moves deeper into GPU-centric offerings.
Commentary from major investment banks has noted that investors still want more evidence that the emerging GPU-as-a-service business will turn into a consistently positive contributor to earnings and cash generation rather than simply driving up scale and capital needs.
Valuation is another part of the debate. Oracle shares currently trade at a forward earnings multiple just above 19 times 12-month profit estimates, a discount to peers such as Microsoft, which trade at higher multiples. That gap reflects a combination of growth expectations, perceived execution risk and differing business mixes between software and infrastructure.
AI tools, SaaS risk and Oracle’s product strategy
Alongside the infrastructure and funding questions, there is an ongoing discussion about how AI-powered coding and automation tools might reshape demand for traditional software and cloud services. Some investors have worried that rapid advances in AI could undermine parts of the SaaS model by changing how software is built and consumed.
Oracle’s leadership has pushed back on the notion of a looming “SaaS apocalypse”. The company’s executive chairman has argued that new AI development tools are not expected to erode demand for Oracle’s products; instead, Oracle is using those same tools internally to create additional software offerings and enhance existing ones.
Research firms following the stock point out that, while these reassurances are plausible, it will still take time to observe whether AI-driven changes affect license volumes, pricing structures or renewal dynamics across Oracle’s application and database franchises.
For now, many analysts believe investors are less fixated on a dramatic collapse in SaaS demand and more focused on the execution risks within Oracle Cloud Infrastructure. That includes questions about scaling AI capacity efficiently, maintaining healthy margins and handling the complex financing mix required to keep up with hyperscale competitors.
In other words, the biggest debate is not whether Oracle can participate in the AI boom, but how well it can balance growth, profitability and financial risk while shifting more infrastructure costs onto its clients.
The evolving picture around Oracle shows a company trying to thread a needle: it wants to be a major player in AI infrastructure and GPU-based services without overburdening its own balance sheet. By designing contracts where customers prepay or bring their own hardware, Oracle is pushing much of the cost of AI data centers toward users while still capturing long-term revenue streams. The early financial results, the spike in contracted obligations and the recent improvement in market sentiment suggest this approach is starting to gain traction, even as investors keep a close eye on debt levels, margins and how Oracle Cloud Infrastructure performs over the next few years.
