Oracle Earnings Call: Delivered 850 MW of computing capacity in a single quarter; GPU renewal premiums reached 20%; RPO to be substantially converted into recognized revenue.

September 10, $Oracle (ORCL.US)$ The first-quarter financial report was released, with revenue reaching $19.3 billion, a 30% year-on-year increase that set a new historical high. The cloud infrastructure division reported revenue of $7.4 billion, surging 121% year-on-year and exceeding the market expectation of $7.19 billion.

Following the earnings release, Oracle's stock price rose nearly 8% in after-hours trading before pulling back to a 4% gain. During the subsequent analyst conference call, Chief Financial Officer Hilary Maxson summarized the quarter with a single word: "acceleration."

Addressing external concerns regarding delays at data centers in New Mexico and Wisconsin, Oracle's management dismissed these issues one by one, stating that "the two sites will not impact FY27 performance" and emphasizing that "not all eggs are in one basket."

Compute Delivery: 850 MW in a single quarter, nearly triple the previous quarter's volume

During the conference call, when asked about progress on data centers, CEO Clay Magouyrk provided a set of specific figures.

In Q1, Oracle delivered 850 MW of AI compute capacity to customers, including over 300,000 GPUs. Clay stated, "Q1 deliveries were nearly triple the full-quarter deliveries of Q4, equivalent to 73% of the total deliveries for the entire previous fiscal year."

Specifically, the Abilene hyperscale campus delivered 131,000 GPUs in Q1, 1.9 times the volume in Q4. Six of the eight campus buildings have been completed and delivered, accounting for 75% of total capacity. Clay specifically noted that OpenAI's latest release, GPT-6 Astra, was trained at Oracle's Abilene site.

GPU Renewals: Older cards used for more than four years can still command a 20% price increase

Magouyrk stated, "All GPUs with contracts expiring in Q1 were renewed or resold at a 20% premium over the original contract price. Most of these GPUs had been in use for four years or longer." He also disclosed that GPU utilization in Q1 reached 97.9%.

In response to analysts' follow-up questions regarding the useful life of hardware assets, Magouyrk said, "I have spent almost my entire career in the infrastructure industry and have been with Oracle for 12 years. One trend has remained unchanged so far: demand for server-side computing and data centers will only increase. This is even more true for AI use cases. Achieving a 20% premium upon renewal is a very positive signal for demand sustainability, business growth, and profitability."

RPO: Approximately half is expected to convert into revenue within 36 months.

During the conference call, Maxson disclosed that Remaining Performance Obligations (RPO) increased by $26 billion quarter-over-quarter in Q1, and provided explicit guidance on conversion pace for the first time: "We now expect approximately half of our RPO to convert into sales revenue over the next 36 months."

Meanwhile, new AI contracts exceeded $30 billion in Q1. Magouyrk emphasized that these new contracts "do not require additional capital outlay from Oracle," as capital expenditure pressures have been partially shifted through various models such as upfront payments and customer-provided hardware.

Maxson further explained the characteristics of this incremental RPO: "This portion of new RPO will not impact our capital expenditures, and the associated revenue will not be recognized until fiscal year 2028 or later."

Data Center Construction Controversy: Management Responds Directly

During the conference call, analysts directly questioned potential delay risks regarding data centers in New Mexico and Wisconsin.

Magouyrk responded that neither of these sites was included in the 850 megawatts delivered in Q1, and they "will have no impact on our previously announced revenue or earnings guidance for fiscal year 2027."

He also stated, "Our data centers under construction are widely distributed across the United States and globally. We do not have all our eggs in one basket." He added that large-scale sites are not launched all at once but are delivered in phases, meaning delays at a single site will not cause a cliff-like drop in any given quarter.

Regarding New Mexico, he indicated that construction is proceeding smoothly, with air permit applications underway, and plans to use Bloom fuel cells for on-site power generation, citing "extremely low emissions and minimal water consumption." Construction of the Wisconsin data center is also progressing "very smoothly," with coordination ongoing with local electricity regulators on grid power supply solutions.

Applications Business: AI Is Not a Replacement, But an Accelerator

Mike Sicilia, CEO of Applications Business, provided a systematic explanation of the growth logic for the SaaS business during the conference call.

"The introduction of AI is an accelerator, not a replacement for suite applications," he said. Overall SaaS business growth in Q1 was 10%, with Fusion growing by 14% and industry-specific applications exceeding 20% growth.

He also revealed that an AI-assisted implementation tool will be launched at the AI World conference in October, "compressing SaaS deployment cycles from years to months, and from months to weeks." He cited examples where the complex go-live cycle for healthcare applications has been reduced from "high double-digit months" to "single-digit months," and for some NetSuite customers, from "double-digit months" to "single-digit weeks."

Full-year guidance raised; Q2 cloud business growth expected to exceed 65%

Maxson announced that Oracle has raised its full-year total revenue guidance to at least $90 billion, representing a year-on-year increase of approximately 34%; the full-year non-GAAP earnings per share (EPS) guidance is $8.10.

Regarding Q2 guidance, total revenue growth is expected to be 30%–34%, and cloud revenue growth is expected to be 65%–71%.

In terms of capital expenditure, the full-year forecast is $90–95 billion, with net cash capital expenditure not exceeding $70 billion. Maxson stated, "Each such project is essentially a strong free cash flow generator. Once it enters the ramp-up phase, it can quickly achieve a free cash flow conversion rate approaching 100% of after-tax EBITDA."

Full transcript of the conference call follows:

Minutes of Oracle's Fiscal Year 2027 Q1 Earnings Conference Call

Conference Date: September 10, 2026 Company Name: Oracle Conference Type: Fiscal Year 2027 Q1 Earnings Conference Call

I. Opening Remarks

Moderator: Good day, everyone, and welcome to Oracle's Fiscal Year 2027 First Quarter Earnings Conference Call. This call is being recorded. Please limit each question to one inquiry. I will now turn the call over to Mr. Ken Bond, Head of Investor Relations. Please proceed.

Ken Bond, Head of Investor Relations:

Thank you, Miriam. Good afternoon, everyone, and welcome to Oracle's Fiscal Year 2027 First Quarter Earnings Conference Call. Joining me on today's call are Mike Sicilia, Chief Executive Officer; Clay Magouyrk, Chief Executive Officer; and Hilary Maxson, Chief Financial Officer.

The press release, which includes financial results, data tables, supplementary financial metrics, and performance guidance, has been posted on our Investor Relations website. The slides used during this conference call, as well as the GAAP to non-GAAP reconciliation tables, are also available on the website.

Please note that today's discussion will include forward-looking statements, and we will also address certain important factors related to our business. These forward-looking statements are subject to risks and uncertainties, and actual results may differ materially from those stated today. Therefore, we caution you against placing undue reliance on these forward-looking statements and recommend reviewing our latest reports, including Forms 10-K, 10-Q, and related amendments. Furthermore, we assume no obligation to update our performance figures or forward-looking statements due to new information or future events.

Before moving to the Q&A session, we will provide some prepared remarks. I will now turn the floor over to Hilary.

II. Overview of Financial Performance

Hilary Maxson, Chief Financial Officer:

Thank you, Ken. Hello, everyone. It is a pleasure to speak with you today. As Ken mentioned, you can follow our presentation in real-time via the earnings slides and webcast available on our website.

If I were to describe this quarter in one word, it would be 'acceleration'—we have accelerated execution across the entire company, translating into strong performance in both revenue and profit.

The first quarter set another all-time high, driven by the dual engines of cloud infrastructure and cloud applications.

• Total revenue reached a record $19.3 billion, representing a 30% year-over-year increase in U.S. dollars.

• This marks the first time total revenue has achieved quarter-over-quarter growth in the first quarter, serving as a significant signal of our continued progress in scaling infrastructure. Historically, a strong fourth quarter has often been followed by a softer first quarter, but this pattern no longer applies as we accelerate across our entire technology stack—from infrastructure and databases to software.

• Cloud infrastructure revenue amounted to $7.4 billion, up 121% year-over-year, reflecting our strong execution in bringing record new megawatt capacity online, underpinned by sustained robust demand for compute and database services.

• Cloud applications revenue grew 10% year-over-year, with Fusion and industry-specific applications growing significantly faster than the overall average. Mike and Clay will provide further details on these business segments later.

• Non-GAAP operating profit increased 31% year-over-year in U.S. dollars to $8.2 billion, driven by both strong revenue growth and improved operating leverage.

• Gross margin declined as expected, primarily due to the accelerated expansion of data centers and the rapid growth of infrastructure revenue. However, this pressure was offset in the current quarter by lower operating costs and strong operating leverage resulting from simplification and efficiency initiatives. Overall, the non-GAAP operating margin remained essentially flat at approximately 42%.

• Non-GAAP earnings per share (EPS) surged 30% year-over-year in U.S. dollars, reaching $1.92 for the quarter.

A final point regarding financial highlights: Remaining Performance Obligations (RPO) increased by $26 billion compared to the fourth quarter, which carries two implications:

First, we continued to expand our RPO base in the current quarter to support future revenue. The vast majority of new contracts utilize prepayment, bring-your-own-hardware, or similar mechanisms, requiring no additional capital investment from Oracle. The impact of this incremental RPO on capital expenditures or revenue will not be realized until fiscal year 2028 or later.

Second, this quarter we began to see a strong conversion of Remaining Performance Obligations (RPO) into revenue, driving impressive performance in our cloud infrastructure business. We have added several charts to the slides to clearly illustrate the significant inflection point in the conversion of RPO into revenue and operating profit.

In detail:

• Cloud infrastructure revenue grew 121% year-over-year this quarter, up from 93% in the previous fourth quarter. We expect this accelerating momentum to continue throughout the remainder of fiscal year 2027 as more RPO converts into revenue.

• We currently expect approximately half of our RPO to convert into sales revenue within the next 36 months.

• On a trailing twelve-month basis, the company's total revenue growth rate increased by 5 percentage points in the first quarter compared to the fourth quarter, with the contribution from cloud infrastructure continuing to rise.

• Also on a trailing twelve-month basis, operating profit growth accelerated from 16% in the fourth quarter to 21% in the first quarter, demonstrating strong momentum.

Regarding the balance sheet and cash flow:

• Operating cash flow reached a record $23 billion this quarter, reflecting our strong execution amid robust demand, as well as benefits from customer advance payments.

• Capital expenditures for the quarter amounted to $28 billion, resulting in negative free cash flow of $5 billion.

• Net cash capital expenditures, excluding advance payments, were $18 billion.

It should be noted that capital expenditures are not evenly distributed throughout the year. We continue to forecast full-year capital expenditures between $90 billion and $95 billion, with net cash capital expenditures not exceeding $70 billion.

Finally, we are pleased to announce that the previously disclosed $20 billion at-the-market equity offering program was fully completed in the first quarter.

I will now hand over to Mike, and then to Clay, for a detailed discussion of our cloud applications and infrastructure businesses.

3. Overview of Cloud Applications Business

Mike Sicilia, Chief Executive Officer:

Thank you, Hillary.

First, I would like to provide additional insights into our applications business.

We continue to see strong customer endorsement of our application suites. Customers are investing in trusted, comprehensive solutions that seamlessly integrate AI agents with applications to drive business operations. The introduction of AI serves as an accelerator for our application suites, not a replacement.

Over decades, we have accumulated deep expertise and capabilities in managing business process operations across industries, regions, and organizations of all sizes, giving us a profound understanding of how to help customers succeed.

Even before the advent of AI, our application suites had proven their effectiveness. End-to-end automation and standardized, efficient business processes enabled enterprises to improve profit margins, far outperforming customized point solutions. However, this required employees to strictly adhere to workflows designed within the system, a consistency that many organizations struggled to maintain across different functions, teams, and regions.

AI has changed this dynamic. AI agents can automatically execute tasks based on an organization’s existing workflows and business rules, while employees transition to roles focused on overseeing agents, handling exceptions, and applying human judgment at critical junctures. By combining application AI with decades of accumulated complex business rules, regulatory compliance requirements, security models, data models, and customer-specific configurations, we help customers continuously unlock the value of AI while ensuring data security and maintaining robust operational guardrails.

This makes it easier than ever for organizations of all types to harness the full power of the application suite. We are confident in the potential of this new paradigm to deliver substantial ROI to customers rapidly.

At the AI World conference this October, we will launch a new AI Agent Accelerator, which will redefine how customers deploy Oracle applications—faster, simpler, and at a significantly lower cost. AI agents will collaborate with Oracle and customer teams to automate and orchestrate implementation processes at an unprecedented scale, compressing SaaS deployment cycles from years to months, and from months to weeks. It is the compounding effect of these capabilities that convinces me our applications business growth will continue its upward trajectory.

First Quarter Performance:

This quarter, driven by the demand environment described above, several of our SaaS products performed strongly. Overall SaaS business grew by 10%, with Fusion growing by 14%. Oracle Health business continued to accelerate, and industry applications (though not disclosed separately) achieved year-over-year growth exceeding 20% in the first quarter.

As mentioned last quarter, some decision cycles for NetSuite slowed in the previous fiscal year, resulting in growth slightly below other business segments; however, we have launched significant new products that we look forward to detailing later.

Selected customer highlights this quarter (excerpted from a longer list):

• Uber Technologies, Stanford University, and MUFG Bank (Japan) have all gone live with and/or accelerated their use of Fusion;

• Pye-Barker Fire & Safety selected the complete Oracle application suite, ranging from industry-specific applications to Fusion, including Fusion Agent Applications;

• Johnson Controls, Saudi National Bank, Guidewell Mutual Holdings Corporation (a health solutions company serving over 45 million people), and Petronas all added Fusion Agent Applications this quarter to achieve better business outcomes.

AI Usage Statistics:

• Customers utilized our embedded AI capabilities more than 150 million times this quarter, representing a 42% quarter-over-quarter increase;

• AI agents were executed over 3.5 million times in production environments, nearly doubling compared to the previous quarter;

• The number of AI agents used by customers in production environments exceeded 2,300, marking a 90% quarter-over-quarter growth;

• In the Fusion application alone, AI production usage consumed a total of 900 billion tokens.

It is evident that customers are using our AI capabilities embedded within the Fusion application and the entire application stack on a daily basis.

Regarding NetSuite:

We announced the general availability (GA) of NetSuite Next. This is a new AI-driven product that offers a simpler and more powerful agent experience, with AI deeply integrated into the workflows customers rely on daily. The product is easier to adopt, enhances productivity from day one, and continues to deliver greater value as customers grow.

Furthermore, the NetSuite AI Connector service has become one of the fastest-adopted features in NetSuite’s history—enabling customers to securely connect NetSuite data to their preferred mainstream AI assistants (including ChatGPT and Claude). Currently, over 10,000 customers are using this service. Every Man Jack, a personal care company, estimates that this service alone saves them approximately $350,000 and nearly 5,000 hours of work annually.

Regarding Oracle Health:

This month, at our Oracle Health and Life Sciences customer event, we will launch the new Magenta Care Management System, featuring a distinguished lineup of external speakers. This system is more than just an electronic health record (EHR) system; it organically connects clinical research with care delivery, reducing the burden of patient record management and allowing healthcare professionals to use AI as their user interface, thereby focusing more energy on their core responsibilities. Among all AI missions, helping to improve healthcare quality and enabling healthcare professionals to focus on patients rather than computer systems is one of the most meaningful directions. The effectiveness of Oracle’s AI products has been thoroughly validated, and we are confident about the future.

I will now hand over to Clay.

4. Overview of Cloud Infrastructure Business

Clay Magouyrk, Chief Executive Officer:

Thank you, Mike.

Oracle Cloud Infrastructure (OCI) continues to grow rapidly to meet customer capacity demands.

Regarding capacity delivery:

• Since the end of the fourth quarter, we have delivered 850 megawatts of AI computing capacity to customers, comprising more than 300,000 GPUs;

• Delivery volume in the first quarter was nearly three times that of the fourth quarter, equivalent to 73% of the total delivery volume for the entire previous fiscal year;

• This achievement is attributable to our sustained investment over the years across all stages of infrastructure, including data center design, supply chain and manufacturing, as well as installation and operations.

Customer demand remains robust:

• New AI contracts signed in the first quarter exceeded $30 billion, with no additional capital expenditure required from Oracle;

• Our capability to operate large-scale multi-tenant clusters continues to demonstrate significant advantages, with GPU utilization reaching a high of 97.9% in the first quarter;

• The lifespan and value of GPUs continue to exceed expectations—all GPUs up for renewal in the first quarter were either renewed or resold at a 20% premium over the original contract price, with the majority having been in service for over four years. This fully demonstrates that our deployed AI computing power possesses long-term utility and continuously improving economic benefits.

Regarding the Abilene campus:

• In the first quarter, 131,000 GPUs were delivered at Abilene, representing 1.9 times the delivery volume of the fourth quarter;

• Six of the eight buildings in the campus have been delivered, representing 618 megawatts of capacity, which accounts for 75% of the total capacity;

• Customer acceptance has been compressed to just 24 hours, indicating that the systems are ready to handle customer workloads upon arrival;

• GPT-6 Astra, recently released by OpenAI, was trained at our Abilene facility.

Regarding Shackleford:

• Progress on our next gigawatt-scale campus, Shackleford, is proceeding smoothly;

• The NVIDIA Vera Rubin system has exceeded expectations in hardware quality, manufacturing yield, and performance. We will deliver the first batch of Vera Rubin systems to customers in the second quarter.

Regarding the database cloud business:

• Multi-cloud database revenue increased by 353% year-over-year, and the number of multi-cloud database customers grew by 180% year-over-year;

• We completed our planned expansion of Azure and AWS regions, currently covering 70 multi-cloud database regions and 119 availability zones, providing a consistent way for customers to run Oracle AI Databases in their chosen clouds, close to their applications and data;

• Oracle Interconnect for AWS was officially launched. OCI now has private, high-speed connections with all major hyperscale cloud providers with no data transfer fees, facilitating customer development of distributed applications and workload migration across different clouds.

Regarding AI ecosystem innovation:

• We expanded our partnership with OpenAI, offering access to the OpenAI API, Enterprise ChatGPT, and Codex via the Oracle Cloud Marketplace, including GPT-6 Astra;

• We are integrating Gemini models into Oracle enterprise applications;

• We released new Grok inference, multimodal, and text-to-speech models;

• The open-source model catalog continues to expand, with new additions from NVIDIA, Qwen, Google, DeepSeek, and others;

• We have announced a multi-year partnership with Quentinium to provide its Helios quantum computer via Oracle Cloud Infrastructure (OCI). Helios will operate within OCI AI data centers in the United States, delivering hybrid quantum and AI workload capabilities for enterprise research and AI customers.

Regarding developer tools:

Oracle APEX 26.1 integrates infrastructure, database, and modeling capabilities to empower application developers. APEX currently supports over 2 million active applications, with thousands of new applications added daily.

APEX Lang is a new technology that represents APEX applications as structured, human-readable application definitions. These definitions can be stored in source code management systems for validation and governance. AI coding agents can generate and modify these definitions, while the APEX engine continues to provide the security, reliability, and operational control required for enterprise applications. This enables developers to achieve the efficiency of generative development while avoiding the maintenance challenges associated with opaque application code.

We are applying the same approach to the Oracle AI Data Platform. The platform is now integrated with Codex and Cloud Code, allowing developers to access the data, knowledge, and capabilities of the AI Data Platform directly within their preferred programming environments. We are also enhancing advanced MLOps capabilities and collaborating with early adopters to deploy business knowledge models, semantic region engines, and deep insight agents, thereby deeply integrating AI with enterprise context, semantics, and governance frameworks. These capabilities will also be introduced into Fusion Data Intelligence and Oracle Analytics Cloud, platforms where many customers already manage their most critical business data.

Overall, the first quarter demonstrated the mutual reinforcement of our capabilities. We delivered data center and GPU capacity at a pace that seemed nearly impossible a year ago; customers continued to sign new contracts, renew at higher prices, and keep clusters operating near full capacity; we deployed the Oracle AI Database across major cloud platforms and offered an increasingly rich portfolio of proprietary and open-source models on OCI; and we connected these models to enterprise data, applications, and developer workflows. It is this synergistic effect that drives sustained demand growth and reinforces our strong conviction in the long-term value of the technologies we are building.

I will now hand it back to Hilary.

V. Guidance

Hilary Maxson, Chief Financial Officer:

Thank you, Clay. I will now present the guidance for the second quarter and full fiscal year 2027.

Second Quarter Guidance:

• Total revenue growth is expected to be between 30% and 34% year-over-year (in USD);

• Cloud revenue growth is expected to be between 65% and 71% year-over-year (in USD);

• Non-GAAP earnings per share are expected to be between $1.85 and $1.93, representing a year-over-year increase of 21% to 25% (in USD); this guidance excludes the Ampere-related gains recognized in the second quarter of last year.

Full-Year Guidance:

• Given the strong execution and accelerating momentum in the first quarter, we have raised our full-year guidance, with total revenue expected to reach at least $90 billion, representing approximately 34% year-over-year growth;

• The full-year non-GAAP earnings per share target is $8.10.

Finally, please be sure to mark October's Investor Day on your calendars. I will now turn the call back to Ken for the Q&A session.

VI. Q&A Session

Moderator: Thank you. Please prepare your questions.

Question 1 (Brad Zelnick, Deutsche Bank)

Brad Zelnick: Thank you very much. First, Ken, I understand that this is your last earnings conference call, and I would like to congratulate you on your retirement. For nearly two decades, you have been Oracle's spokesperson to the investment community, consistently reliable and acting with integrity. Your departure will surely be deeply felt by all.

Regarding my question: I firmly believe that Oracle will continue to be among the few leaders in the AI infrastructure market. You previously indicated that fiscal years 2027 and 2028 would be peak years for capital expenditures; however, other industry players appear to be continuously investing hundreds of billions of dollars in capacity expansion without any signs of slowing down. How should we interpret this? If competitors do not decelerate their spending, is it possible for Oracle to reduce its expenditures after the next two years? Is there a possibility of an even higher peak in capital spending beyond fiscal year 2028? What factors will guide your investment decisions? Additionally, based on the current backlog, when can we expect the company to return to positive free cash flow? Thank you.

Clay Magouyrk: Thank you, Brad. I will address the parts I am best suited to answer, and then I will hand over the remaining questions to Hillary.

Over the past few quarters, we have consistently highlighted our ongoing exploration of various innovative ways to finance our business. One approach naturally involves utilizing Oracle's own capital directly. However, we have also made significant investments in supplier and partner relationships and created new business models, including bring-your-own-hardware arrangements, which diversify the sources of capital expenditure in different ways.

Therefore, I believe we need to conceptually decouple 'capital expenditures directly invested by Oracle' from 'business growth potential.' From our perspective, the capital required to drive business growth does not necessarily have to come entirely from Oracle, nor does it all need to be recorded as Oracle's capital expenditure. This is not a bottleneck for business growth but rather a reflection of our continuous evolution of business models as we expand in the AI sector. Hillary, would you like to address the other questions?

Hillary Maxson: Certainly.

Regarding the question on free cash flow: We have not provided specific guidance on timing, nor do we intend to do so today. However, I would like to emphasize that every project we are advancing is inherently highly conducive to free cash flow generation. Once these projects enter their acceleration phase, they can quickly achieve a free cash flow conversion rate of approximately 100% of after-tax EBITDA. Essentially, this business possesses a considerable degree of 'self-sustaining' capability at a certain stage.

We are not providing specific timelines at present, but the situation will become clearer as we continue to expand over the coming quarters. We anticipate that the timeframe will be relatively reasonable following the current phase of large-scale expansion. The only variable factor, as you pointed out, is further growth-related capital expenditure—this is the portion we will continue to deploy at appropriate levels to drive business growth.

Brad Zelnick: Thank you very much. I look forward to seeing everyone at the AI World conference.

Question 2 (Siti Panigrahi, Mizuho Bank)

Siti Panigrahi: Thank you. Ken, I would also like to extend my congratulations on your retirement; your departure will certainly be missed.

Regarding my question: There are many rumors in the market about delays in the data center projects in New Mexico and Wisconsin. However, you delivered on schedule in the first quarter and raised your full-year revenue guidance to at least $90 billion. Could you provide an update on the current progress of these two projects? Are there any risks in the delivery timelines of these two data centers that could impact the revenue expectations for fiscal year 2027? Additionally, with RPO continuing to grow, how confident are you in your ability to secure the necessary capacity supply and maintain the rollout pace? Thank you.

Clay Magouyrk: Thank you, Siti. Before addressing the specific questions, I would like to provide some background information.

New Mexico and Wisconsin are indeed very important large-scale sites for us, with each site having a scale of approximately 1 gigawatt. However, it is important to note that we just delivered 850 megawatts in the first quarter, and these deliveries did not come from Shackleford, New Mexico, Wisconsin, or Michigan. This demonstrates that we have a large and diversified portfolio of data center development projects across the United States and globally, enabling us to continuously provide computing power to our customers. While New Mexico and Wisconsin have received significant attention, please understand that we do not have all our eggs in one basket.

Furthermore, bringing large-scale sites online is not an overnight process. Taking a gigawatt-scale site as an example, even if deliveries are scheduled to begin in January, it does not mean that 1 gigawatt of computing power will be available all at once in January. Instead, delivery occurs in phases across multiple quarters. Therefore, even if a particular site experiences delays, it will not cause a significant impact within a single quarter.

In addition, in our industry, we typically refer to a plan that requires 100% of everything to go exactly as scheduled as a 'bad plan.' We fully understand the complexity of large-scale infrastructure construction. Therefore, we do not assume that everything will proceed without any issues. We always maintain contingency plans and do not presuppose that every task will be completed perfectly on time.

Regarding New Mexico: Construction is progressing smoothly, and the data center project is fully on track. We are advancing the application process for air emission permits. The energy technology we plan to deploy is Bloom fuel cells, which represent one of the most environmentally friendly methods for on-site power generation currently available—offering extremely low water consumption and emissions, superior to any other on-site power generation method. We are confident in driving the project forward through continued communication with local regulators in Doña Ana County, community residents, and various stakeholders in New Mexico. This process is not uncommon for large-scale projects.

Regarding Wisconsin: We do not engage in on-site power generation; instead, we collaborate with various partners to secure energy supply through the grid. The construction of the data center itself is actually progressing very smoothly. We are continuously collaborating with the Public Service Commission, ATC, and WE Energies to optimize energy design and supply solutions. We remain highly confident in both of these sites.

Regarding the impact on fiscal year 2027 revenue: Neither of these two sites will have any impact on our previously announced revenue or earnings guidance for fiscal year 2027.

Overall, while some of our sites are ahead of schedule and others face more challenges, we remain committed to advancing every project. Our forecasting models and target settings have fully incorporated risk management factors.

Regarding confidence in future additional capacity: We are very optimistic about this. Constraints have evolved over time—from GPUs and chip manufacturing initially, to power supply later, and now to data center construction. However, the world is large enough, and the demand for computing power is robust enough, that we are pursuing various avenues comprehensively to bring computing capacity online. We maintain high confidence in our ability to meet current RPO and support future RPO growth.

Siti Panigrahi: Thank you very much for your detailed responses to each point.

Question 3 (Raimo Lenschow, Barclays)

Raimo Lenschow: Thank you. First, Ken, I sincerely wish you a happy retirement, and I also thank the team for making Ken’s final quarter so outstanding.

My question also relates to the data center side. There is considerable market discussion regarding rising component prices. Clay, could you comment on the trends in contract pricing versus spot pricing? Additionally, last year you provided a framework for gross margin expectations. Given the ongoing changes in component pricing, does this framework still hold? Thank you.

Clay Magouyrk: Certainly.

In a market characterized by excess demand over supply, prices typically rise rather than fall. As costs increase, we need to raise our charges to ensure reasonable returns—a practice we apply across all relevant business segments. We do not anticipate this will have a negative impact on gross margins, and the guidance previously provided remains valid.

Regarding the useful life of hardware assets and data center value, I have been in the industry for nearly thirty years, twelve of which at Oracle, and there is one consistent rule: demand for server-side computing and data centers continues to grow. All current indications suggest that AI application scenarios follow this trend, if not exceed it. The renewal of compute capacity expiring this quarter at a 20% premium serves as strong evidence of sustained demand growth, business expansion, and profitability.

Question 4 (Mark Mordler, Bernstein)

Mark Mordler: Thank you very much. Ken, it is regrettable to see you depart, but I am confident you have an exciting bucket list to fulfill.

I would like to focus on RPO: Could you elaborate on the primary drivers behind RPO growth that requires no additional capital expenditure from Oracle—specifically prepayments and the bring-your-own-hardware model? Are the participants mainly AI labs and semiconductor companies, or sovereign cloud customers? By the way, it has been some time since we discussed sovereign cloud, and there has been little mention of sovereign AI. Could you provide an update on the progress of OCI’s sovereign cloud business and its impact on capital expenditures and profit margins? Thank you.

Clay Magouyrk: First, I need to clarify one point: Neither I nor Hillary stated that these projects require no additional capital expenditure. What we said is that these projects do not require additional funding from Oracle. I am aware that I am not an accountant, but the core message is this: While capital expenditures certainly exist, Oracle does not need to raise separate financing for them.

We achieve this through several approaches:

First, supplier financing arrangements: Through financing agreements with suppliers, we make equipment payments in sync with customer receipts;

Second, customer-provided hardware: Customers have expressed willingness to purchase hardware themselves but seek to leverage our operational capabilities, cloud infrastructure technology, and data centers to build AI clusters;

Third, customer prepayments: Customers use their own funds to make advance payments, eliminating the need for Oracle to provide upfront financing.

Regarding customer types: Demand comes from a broad spectrum, ranging from startups to investment-grade large enterprises. All parties recognize that acquiring and allocating capital in diverse ways is the current reality, and the industry is adapting in the most efficient manner possible.

Regarding sovereign cloud: Our Alloy business is performing well. We have numerous partners in Japan and strong partners in the Middle East, covering both commercial sovereign cloud models and government-focused sovereign cloud services. The business continues to expand with robust demand. These operations are closely linked to AI as well—many sovereign cloud customers are also utilizing our GPU capabilities to support their sovereign workloads.

Furthermore, we have a large and rapidly growing general-purpose cloud business. Although we discuss it less frequently, it is experiencing strong growth and healthy margins. While it requires some capital investment, the amount is far lower than that needed for large-scale AI cluster projects.

Question 5 (John DiFucci, Guggenheim Securities)

John DiFucci: Thank you. I have been following Oracle for nearly thirty years, and Ken, you have been here for almost twenty years. In this industry, it is rare to develop professional friendships, and even more precious to become personal friends—and you are indeed my close friend. I have learned much from you over the years, and I sincerely thank you.

My question is for Hillary: Gross margin declined this quarter despite significant capacity coming online. I assume these deliveries incurred substantial costs prior to launch, while revenue had not yet been recognized, leading to a notable drop in gross margin. This has been a frequent topic in our investor communications. However, as you pointed out, operating margin remained largely flat (with slight improvement). I understand that operating margin is the key indicator closest to net income and ultimate cash flow. How should we comprehensively assess the trends in both gross margin and operating margin?

Hillary Macksen: Gross margin is important both internally and externally as a key indicator of business health: it reflects whether pricing is appropriate and whether input costs are effectively managed. Gross margin serves as an important 'health check' metric for us internally, and likewise for investors. Once issues arise in the pricing model, gross margin tends to respond quickly.

However, from a value creation perspective, operating margin is the core metric I ultimately focus on.

The current changes in gross margin are driven by several factors: first, the accelerated expansion of data centers; second, structural adjustments between two business models—the software business has higher gross margins but also higher R&D and sales expenses; the infrastructure business has lower gross margins but significantly lower R&D and sales expenses, particularly for Oracle, where the company's overall R&D investment can be leveraged to empower this segment. Therefore, the core metric for assessing our long-term value creation capability should be operating margin, although gross margin certainly warrants continued attention.

John DiFucci: Thank you for the clear explanation. Could you elaborate further? The market generally appears to have overestimated expectations for Oracle's gross margin while underestimating the resilience of its operating margin. Will gross margin continue to decline, or will it remain at current levels for the remainder of the year?

Hillary Macksen: As mentioned in the fourth quarter, gross margin will experience a step-down decline this year. However, our EPS guidance reflects our expectations for operating margin. Looking ahead several years, it is reasonable to expect gross margin to stabilize as the expansion phase concludes, although we are not providing specific guidance today. More details will be presented at the Investor Day in October.

John DiFucci: Thank you very much. The details provided tonight were very helpful.

Question 6 (Brent Thill, Jefferies)

Brent Thill: Mike, it is encouraging to see double-digit growth in SaaS. Could you share your outlook for the coming years? There is widespread market concern about the impact of AI on the SaaS business, but this does not seem to align with your view. Could you discuss the logic behind sustaining double-digit growth in SaaS?

Mike Sicilia: Certainly, thank you for the question.

I would like to briefly expand on the scope of our SaaS business, as there are several key interrelated factors.

First, we offer highly differentiated end-to-end industry suites covering numerous verticals such as healthcare, retail, telecommunications, and construction, with equal emphasis on horizontal and vertical applications. This quarter, both Fusion and industry applications achieved strong growth. Customers can acquire these as complete suites, which has been Oracle’s key differentiator for many years.

Second, we layer AI on top of this foundation—delivered as embedded features within regular application updates, without additional add-ons. We believe this provides customers with a seamless way to rapidly realize ROI from AI, without needing to integrate separate solutions.

The third layer is the Fusion Agentic AI Studio: Customers and partners can build their own AI agents within the same platform, sharing a common control plane and benefiting from synchronized quarterly updates and security patches. Delivered as a comprehensive cloud service, it uses AI as the interface to navigate complex business rules, highly differentiated security models, and data models refined over many years.

The fourth layer is AI-assisted deployment: This is a major innovation we are driving. In many cases, we are helping customers address highly challenging, industry-specific mission-critical issues that traditionally involve lengthy implementation cycles and complex data migration. AI significantly accelerates this process. Current empirical evidence includes: complex healthcare application deployments compressed from dozens of months to just a few months; and early NetSuite customers leveraging AI tools to go live in single-digit weeks rather than dozens of months.

This delivers dual value: First, it helps customers realize value faster and at lower cost. Second, it accelerates revenue recognition in complex industry projects, where revenue was previously deferred based on implementation progress.

Another critical factor is the AI Data Platform: It automates the creation of enterprise ontologies, serving not only Oracle applications but also third-party applications run by customers. By integrating horizontal applications, vertical applications, and the Agentic Studio into a unified cloud service offering—especially in highly regulated industries and at our current scale—I believe no other player in the market can match this capability.

Finally, our SaaS business serves as a high-quality lead generator for our IaaS business. SaaS customers are also purchasing OCI and migrating non-Oracle workloads to OCI, thereby validating OCI’s performance and scalability as SaaS consumers. Thus, we view SaaS as part of an integrated solution rather than an isolated business line.

We continue to invest, lower barriers to adoption, and expand our AI toolset, giving us strong confidence in the future of our applications business.

Brent Thill: Thank you, Mike.

Question 7 (Kirk Materne, Evercore ISI)

Kirk Materne: Thank you. Ken, since Investor Day has not yet arrived, I will save my congratulations for when we meet in person. Mike, you mentioned the AI Data Platform in the press release, and I would like to delve deeper into this. It appears to be a significant way for customers to deploy AI agents based on proprietary data. Could you elaborate on its business model? Does it drive incremental consumption of Oracle Database and OCI, or does it exist as standalone software license revenue? I am curious about its long-term impact on financial metrics.

Mike Sicilia: You have actually hit upon the answer—the business model is "all of the above." We can run it in workloads entirely outside of Oracle; it is by no means exclusive to Oracle Database or Oracle Applications. The AI Data Platform can connect to any data source, and we are currently automating ontology construction for hundreds of data sources. Whether through pure AI Data Platform consumption, in combination with Oracle Applications, or with OCI, our primary focus is enabling customers and partners to make the choices that best suit their needs.

I would add that Oracle Database remains the world’s most trusted platform for mission-critical data hosting. As we automate ontology construction and leverage Oracle Database as one of the primary data sources for mission-critical data, it will also fuel the continued growth of our multicloud database business. Customers increasingly seek to migrate their on-premises Oracle databases to the cloud with ease, and as we continue to expand our deployment across every cloud and region, the AI Data Platform will further drive this growth.

Kirk Materne: You have a vast services ecosystem, but do you also deploy onsite engineers to help customers unlock the value and new opportunities in these previously underutilized databases? Is there a corresponding market expansion strategy?

Mike Sicilia: Yes, we are already deploying onsite engineers for customers, both for the AI Data Platform and for Fusion Agentic Studio. Both share a single control plane on OCI, and we view them as an integrated platform. We consider this service essential and believe it helps customers achieve go-live status as quickly as possible. As I mentioned, we are already seeing results measured in "weeks" in highly regulated industries, which would have been unimaginable even a year ago.

7. Closing Remarks

Clay Magouyrk: We expect to announce our fiscal second quarter 2027 results on December 14, 2026. Any changes to this date will be publicly notified in advance. A replay of this conference call will be available on our Investor Relations website for 24 hours. Thank you all for participating today. I will now hand the call back to Miriam for closing remarks.

Moderator: Thank you all for participating today. This concludes our meeting. Please disconnect your lines.

Disclaimer: This transcript may contain inaccuracies, including typographical errors and other defects. It is provided "as is" without any express or implied warranties.

AI-powered earnings snapshot highlights key takeaways—build your options strategy! Open Futubull > Stock Page > Click [Company] >Earnings Snapshot

Editor/lambor