Research Finder
Find by Keyword
Is Oracle's $523 Billion AI Backlog Worth the $50 Billion Price Tag?
Record AI backlog, a commitment to chip neutrality, a CapEx hike, while posting huge OCI growth and health in core franchises.
12/11/2025
By the numbers
- Q2 Total Revenue: $16.1 Billion, up 14% year over year
- Q2 Non-GAAP EPS: $2.26, up 54% year over year
- Remaining Performance Obligations (RPO): $523 Billion, up 438% year over year
- Cloud Infrastructure (IaaS) Revenue: $4.1 Billion, up 68% year over year
- Q2 Free Cash Flow (FCF): -$10 Billion
Key Highlights
- The massive $523 billion Remaining Performance Obligation is strong validation of high demand for AI infrastructure capacity.
- The 54% jump in non-GAAP EPS was heavily inflated by a one-time gain from divesting the Ampere chip unit.
- Management’s decision to exit in-house chip design for "chip neutrality" highlights the rapidly shifting GPU landscape.
- The company is dramatically accelerating CapEx, raising the fiscal year 2026 spend to an alarming $50 billion.
- Negative free cash flow of $10 billion in the quarter underscores the heavy costs of capacity build-out required to satisfy backlog.
The News
Oracle announced mixed fiscal second-quarter 2026 results, reporting an EPS beat that was significantly boosted by a one-time divestiture gain. Total revenue slightly missed analyst estimates, but cloud infrastructure revenue growth accelerated sharply. The standout number was the $523 billion Remaining Performance Obligation, which signals massive, multi-year AI deals with major players like Meta and NVIDIA. This extraordinary backlog requires an equally extraordinary commitment, with the company raising its fiscal 2026 capital expenditure forecast to $50 billion. Find out more.
Analyst Take
My analysis of this quarter is straightforward: Oracle is executing on its core franchises perfectly. Database growth, check, ERP growth, check, Cerner growth, check, core OCI growth, market leading growth, check, Netsuite growth, check - The TL;DR - There is a lot to like in the core business.
People are mistakenly conflating Oracle with being a "neocloud" provider, a narrative that fails to capture the company's established strengths and diversified strategy. This oversimplification misses several key facts that define Oracle’s current position and future potential.
Oracle possesses highly profitable, long-term franchises—including Database, ERP, Finance, and Healthcare—that generate significant and predictable Annual Recurring Revenue (ARR). This stable financial foundation enables the company to strategically build out its faster-growing cloud and AI businesses.
A critical point often overlooked is the allocation of Oracle's substantial capital expenditure (CapEx). While the company is spending an impressive amount, the investment is not solely dedicated to building speculative GPU farms for AI. Much of the CapEx is directed towards sovereign build-outs and deploying Exadata infrastructure within other Hyperscalers, catering to existing, burgeoning demand for core OCI and database offerings. The infrastructure expansion is rooted in building capacity against existing demand and backlog, not just a speculative AI gamble.
Furthermore, Oracle holds powerful, defensible moats in key software markets. The company has a nearly 50-year history of providing data curation, presentation, and orchestration for the world's largest enterprises. This expertise is expected to be the defining factor for successful AI projects in 2026 and beyond. In the rapidly emerging AI use case of healthcare, Oracle owns Cerner, the market leader in electronic patient records, positioning it ideally to capitalize on this sector.
In net, Oracle is fundamentally different from newer "neocloud" providers like CoreWeave. It is emerging as a genuine fourth Hyperscaler that operates with less technical debt and is leveraging its stable, profitable software franchises to fund a strategic cloud and AI build-out. The market narrative does not currently sustain the underlying facts, and Oracle should be valued as a diversified, established technology powerhouse strategically investing in existing and future demand.
Now to the AI Bubble Froth
While servicing core franchise demand, where the company has market-leading positions, Oracle is also undertaking a bold, high-stakes bet on the AI infrastructure boom, and the bet is currently consuming cash at an unprecedented pace. The headline figures paint a stark picture. On one hand, you have a staggering $523 billion in RPO, a figure that is nearly four times the backlog from the previous year. This RPO number is not merely large, it is transformative, positioning Oracle as a serious player in the global AI cloud ecosystem, even if the revenue conversion still lags. On the other hand, the financial quality of the quarter was certainly mixed. The 54% non-GAAP EPS growth is impressive on paper, but removing the $2.7 billion pre-tax gain from selling the Ampere chip unit leaves core profitability looking potentially below market expectations.
The market reaction post-earnings suggests a clear concern about the financing of this transition. Oracle reported a negative $10 billion free cash flow for the quarter, the third consecutive quarter of cash burn, coupled with the announcement that fiscal 2026 capital expenditures will now reach $50 billion. This level of spending is breathtaking. Management is essentially telling investors that to capture that $523 billion backlog, they must undergo a temporary but profound shift in financial structure, moving from a cash-generating machine to a massive infrastructure developer. They are betting that the long-term, high-margin revenue stream from cloud customers will eventually justify the upfront debt and build-out costs.
The strategic shift regarding chip design is another critical detail. Larry Ellison explicitly stated the company is moving toward a policy of "chip neutrality." By selling its interest in Ampere, Oracle signaled that dedicating resources to designing its own chips is no longer strategic. This move is architected to allow them to remain agile in a hardware ecosystem dominated by NVIDIA and, increasingly, AMD. It shows a realization that to meet the immense and heterogeneous demands of AI customers, particularly those running large language models, the smartest move is to become a neutral platform that can deploy whatever chips clients require. This decision is a major validation for the leading GPU duopoly, and it simplifies Oracle’s hardware sourcing, focusing capital on data center construction rather than processor R&D.
The growth in the cloud is undeniably strong. Huge, in fact. Cloud Infrastructure revenue grew 68% year over year, accelerating sequentially. To put that in context, the Hyperscaler peer group is growing at roughly half that rate, albeit off larger numbers. This is the core engine of the RPO growth and the most encouraging sign that the strategy is working. Furthermore, the push for "Cloud Neutrality" and the multicloud database business is paying off handsomely; management reported the Multicloud database business grew an incredible 817% in the quarter. Read that again 817% growth. This is a brilliant strategic maneuver designed to appeal to CIOs who are determined to avoid vendor lock-in with AWS or Azure. By embedding services directly into competitor clouds, Oracle bypasses the traditional lift-and-shift problem and brings its legacy database customers into the cloud world on their own terms.
The Applications (SaaS) side is less spectacular but remains stable. Fusion Cloud ERP is growing at a healthy 18%, and NetSuite is also performing well. However, the traditional Software segment experienced a 3% decline, illustrating the inexorable march of customers away from legacy license models. This shift only reinforces the high-stakes nature of the cloud infrastructure expansion. Oracle needs OCI to maintain rapid acceleration to compensate for the gradual erosion of its highly profitable, albeit decelerating, legacy license base. This is in line with other businesses of Oracle’s vintage and doesn’t concern me one bit.
The debate around this quarter is whether the quality of earnings and the sheer scale of the capital commitment undermine the significance of the RPO growth. The market volatility proves investors are split. My view is that the RPO validates demand, but the CapEx figure validates risk, manageable, but still noteworthy. Oracle must now execute its ambitious build-out flawlessly and secure long-term, profitable utilization of its new data center capacity. This is no small feat.
What was Announced
The core announcement detailed the financial results and the major jump in RPO due to new, large-scale AI infrastructure deals, including those with Meta and NVIDIA. Management announced the immediate divestiture of its interest in the Ampere chip company, pivoting to a strategy of chip neutrality where Oracle is architected to utilize the best available CPUs and GPUs from all suppliers to meet customer needs.
In terms of physical expansion, the company confirmed it is halfway through building out 72 Oracle Multicloud datacenters that are designed to be embedded directly within the cloud environments of major competitors like Amazon, Google, and Microsoft. This physical deployment reinforces the "Cloud Neutrality" commitment. The AI strategy aims to deliver embedded AI functionality across three distinct layers: the underlying Cloud Datacenter software, the Autonomous Database and analytic software, and the expansive Applications portfolio. Specific examples given included using AI to automate complex multistep processes like loan origination for banks and patient diagnosis and reimbursement management for healthcare providers.
Looking Ahead
Based on what I am observing, the market is giving Oracle partial credit for the massive RPO increase, but it is deeply concerned about the underlying costs. The $523 billion backlog is real demand, not just speculation. That is the good news. The key trend that I am going to be tracking is the rate of RPO conversion into actual revenue and the profitability of these hyper-scale AI deals. Based on my analysis of the market, the CapEx surge is a bold bet, but, it is a necessary evil if Oracle intends to be a top-tier provider of AI capacity alongside the established hyperscalers.
The earnings print confirms that the AI infrastructure arms race is heating up, and capital intensity is the new barrier to entry. Going forward, I am going to be looking for how the company performs on reducing its negative free cash flow and articulating a clear timeline for when the capital spending curve flattens. HyperFRAME will be closely monitoring how the company does at financing this expansion without compromising its commitment to maintaining an investment-grade debt rating in future quarters.
Put simply, Oracle has profitable, defendable moats in core software offerings that are largely growing. OCI is growing 2x its peer group, and the company has a clear strategy to be a winner in AI.
Steven Dickens | CEO HyperFRAME Research
Regarded as a luminary at the intersection of technology and business transformation, Steven Dickens is the CEO and Principal Analyst at HyperFRAME Research.
Ranked consistently among the Top 10 Analysts by AR Insights and a contributor to Forbes, Steven's expert perspectives are sought after by tier one media outlets such as The Wall Street Journal and CNBC, and he is a regular on TV networks including the Schwab Network and Bloomberg.