The Zhitong Finance App learned that German software giant SAP SE (SAP.US) achieved the quarterly revenue milestone of breaking 10 billion euros for the first time in the third quarter, and that cloud business growth exceeding expectations indicates that the artificial intelligence (AI) boom has not weakened market demand for its products. The company also expects full-year cloud business revenue to reach the high end of the previous guidance range.
According to financial reports, SAP's third-quarter revenue increased 13% year over year to 10.27 billion euros, exceeding 10 billion euros for the first time in a single quarter. In terms of profit, non-IFRS operating profit was 3.03 billion euros, up 18% year on year, breaking 3 billion euros for the first time in a single quarter; non-IFRS operating profit margin was 29.5%, up 1.2 percentage points year on year. IFRS basic earnings per share were 1.96 euros, up 14% year over year; non-IFRS basic earnings per share were 1.89 euros, up 19% year over year, higher than analysts' unanimous expectations of about 1.83 euros. Free cash flow increased 36% to 1.72 billion euros.
Cloud business revenue increased 26% in the third quarter to 6.67 billion euros, better than analysts' general expectations of 6.57 billion euros. SAP said that the increase in cloud business revenue was due to customers resuming investment after getting out of “financial difficulties.”
At a fixed exchange rate, the SAP cloud business order backlog increased 26% to 24.14 billion euros, higher than the 24% to 25% increase previously predicted by UBS. SAP's order backlog is seen as an important indicator of global corporate technology spending, as the company's software suite provides accounting and payroll management services to many of the world's largest enterprises.
SAP develops enterprise-grade software to manage finance, human resources, and other core business functions. In recent years, SAP has encouraged customers to move from legacy software systems installed on local servers to the cloud. Currently, most of SAP's revenue comes from remote access services. The company plans to end regular support for old products next year, after which the cost of maintaining traditional services will rise.
SAP currently expects subscription revenue in 2026 — that is, the right for customers to rent software remotely rather than buy and run the software themselves — to reach the upper end of the expected range of 25.8 billion to 26.2 billion euros, corresponding to a year-on-year increase of 23% to 25%. Taking into account the declining traditional software licensing business, SAP expects annual revenue to reach 36.3 billion to 36.8 billion euros, an increase of 12% to 13% over the previous year. SAP maintained its annual operating profit guideline, which was 11.8 billion to 12.2 billion euros at a fixed exchange rate.
Below is the full text of the SAP performance call (AI tool assisted translation):
I. Executive speeches
CEO Christian Klein
Excellent performance in the third quarter. We took the “independent enterprise” from vision to reality. Customers are experiencing our latest AI capabilities, and the response is unprecedented. The strategy not only resonates, but is also being implemented as a result.
In terms of key data, the current cloud backlog of orders reached 24.1 billion euros, an increase of 26% over the previous year; cloud revenue was 6.7 billion euros, an increase of 25% under a constant exchange rate. Among them, AI data and platform business were the main drivers. Ninety percent of the 50 largest transactions included AI and commercial data cloud (BDC). Total revenue for the quarter was 10.3 billion euros, up 12%. This was the first time in SAP history that quarterly revenue exceeded 10 billion euros; operating profit increased 18% to more than 3 billion euros. According to IDC data, we are growing about 11 percent faster than the market.
AI is the three major driving forces for customers to go to the cloud: accelerate RISE and GROW cloud migration, make the commercial value of the suite more prominent, and customers rely on RISE/GROW to embrace the commercial AI platform. RISE and commercial AI are fully blossoming in various industries, with new customers including financial services (Spanish Foreign Exchange Bank, Fairstone Bank of Canada), technology and research (Infineon, Fraunhofer Association), manufacturing (Japan NTN), logistics (Nagel-Group, US Defense Logistics Agency), etc. On the GROW side, South America's largest railway manufacturer Greenbrier Maxion plans to use Joule to improve real-time operational visibility; customers such as AstraZeneca, Decathlon, Denmark's Salling Group, and Henkel (Henkel) are also using industry AI. Among them, Salling Group has reduced listing time by up to 66% and category management time by up to 70%.
At the solution level, Kao (Kao) uses SAP Signavio, NTT DATA uses procurement solutions, and Helvetia uses SuccessFactors; IDC estimates that our human capital management (HCM) market growth rate is more than 1.7 times that of the industry as a whole. RISE customers launched this season include Mondelez (Mondelez), Panasonic, Roche, Infosys, etc.
The core of the autonomous enterprise strategy is Joule Work, which enables 400 million SAP users to interact with SAP in natural language, based on three major capabilities: one is “dialogue” (dialogue), where users can ask questions, delegate tasks, and call dedicated agents (agents) when necessary. For example, supply chain planners can allow Joule to summarize supplier contracts and delegate agents to resolve supply shortages and directly trigger production adjustments; the second is “Spaces” (Spaces), which aggregate real-time data across systems, so that IBP and S/S/ 4HANA data presents supply, demand and risk on the same screen; the third is “Joule Studio”, which supports everything from no code to professional code extensions or self-built agents. Currently, the autonomous kit includes 40 assistants and 300 agents, and we insist on “quality over quantity”.
Technically, Joule relies on a commercial AI platform to provide more than 70 models (including cutting-edge and open source weight models), without locking down a single model and giving customers the right to choose; the knowledge map is anchored in BDC, maps more than 7 million fields, 500,000 tables, and 400 data products, and extends the context to data outside of SAP through acquisitions of Dremio and Reltio. In the Ariba scenario, Joule achieved twice the accuracy at half the total cost of leading edge models. At the governance level, each agent is operated and governed by SAP, covering compliance and auditor certification in 26 industries and 60 countries; we promise that customer data will never be used for big model training or improvement.
We are also transforming ourselves into an independent enterprise: 40,000 developers work “AI-first”, and code contributions have increased by about 70% since the beginning of the year; the listing strategy has been fully AI; Joule Work Desktop has been launched among over 110,000 employees, and the productivity of finance, manpower, and procurement functions has increased 20%. The Q4 pipeline is strong, but Q4 has always been the biggest and most critical quarter, and the team is in full execution.
CFO Dominic Assam: Financial Data and Outlook
Strong performance continued in the third quarter. With the current cloud backlog of orders exceeding €24 billion, an increase of 26%, SAP is still the undisputed leader in the SaaS/PaaS growth rate for large enterprises. We still expect cloud backlog order growth to slow slightly in 2026 under a constant exchange rate, but market share will continue to increase due to macropolitical uncertainty likely to have the greatest impact on the Q4 order season, and the continued expansion of the base.
Cloud revenue was around 6.7 billion euros, an increase of 25%, including a positive one-time impact of about 1 percentage point (mainly from revenue capture from financially troubled customers). Cloud ERP suite revenue increased 28%, accounting for 89% of cloud revenue; software licensing revenue declined 9%; service revenue declined 5%. Total revenue of 10.3 billion euros, an increase of 12%.
By region, cloud revenue is particularly strong in Asia Pacific Japan (APJ) and Europe, Middle East and Africa (EMEA), while the Americas are steady; Brazil, France, Germany, Italy, India, South Korea, and Spain have performed well; Switzerland, the United Kingdom, and the US are particularly strong.
In terms of profit, IFRS cloud gross profit margin for the quarter was 73.9% and non-IFRS 74.3% (down 0.7 percentage points year on year); IFRS operating profit increased 18% to 2.9 billion euros, and non-IFRS operating profit also increased 18%, exceeding 3 billion euros (both benefited from a one-time negative impact in the same period last year). IFRS effective tax rate 24.4%, non-IFRS 25.8%; free cash flow of €1.7 billion; IFRS earnings per share increased 14% to €1.96, and non-IFRS earnings per share increased 19% to €1.89.
Looking ahead, we expect cloud and “cloud+ software” revenue to fall at the upper end of the original guidance range in 2026, while maintaining the original guidelines of cloud backlog of orders, operating profit, and free cash flow. We are still committed to accelerating total revenue growth in 2027, but there are three tips: First, Q4 accounts for about 50% of the annual order volume, and macro-uncertainty may suppress Q4 orders and affect the 2027 trajectory; second, we intend to use the service business to drive AI adoption, which will further reduce billable hours and service revenue; third, as revenue improves and the base figure rises, the difficulty of accelerating in 2027 will also increase. Overall, we are still planning for revenue acceleration in 2027, but we need to face up to these factors.
II. Q&A session
Morgan Stanley analyst Adam Wood
Q: Why is the Cloud Backlog Order (CCB) guidance still conservative; are they just cautious about the size of Q4, or are they seeing hidden concerns in pipeline or customer decisions?
Dominic Assam: CCB's growth has been strong this year, but half of the orders for the whole year still need to be placed in Q4. The current macro is not ideal. Oil prices have risen from $60 to $110, and the US interest rate has hit more than 5% and is beginning to show strength, so it is not appropriate to assume that the Q4 macro will be very favorable. We did not give an exact point value for “slight deceleration”. This is just a range implied by the previous guidelines. The upper end of the range may be around 24%. Ultimately, it depends on the trend of the remaining time in Q4.
J.P. Morgan analyst Toby Ogg
Question: Has the backlog growth rate exceeded internal expectations, and which products brought unexpected surprises?
Christian Klein: Every RISE/GROW journey is a surprise. We just received a large mining company today. The other party gave our data migration agent extremely high reviews. The feedback from our consultants and partners was just as good. The system integrators are even willing to disrupt their own models to speed up customers to the cloud. Second, most deals now include AI. Two years after BDC went public, Dremio and Reltio acquisitions are just right, and we will be able to automatically expand the ontology (ontology) for non-SAP data in the first quarter of next year. Customers are gradually understanding that data model chaos will slow AI implementation, which requires a “clean core”, just like any ERP implementation. Finally, Q4 is a golden period for our new AI capabilities. Joule Work has covered scenarios such as finance, manpower, and sales, and always has data context and authority governance, which is a value unique to SAP.
Jefferies Analyst Charles Brennan
Question focus: How far is the agent's response time and accuracy from the target; have AI credits affected income statement metrics?
Christian Klein: In terms of accuracy, we'll never “finish”. When the product manager shows 95% accuracy, I ask him to raise the testing standards and solve more difficult problems. This is an ongoing process. In terms of response time, we tracked every query to the data layer, body layer, and API, and the team was very motivated and progressed well.
Dominic Assam: After AI credits are issued, we amortize them according to the contract period, and there will be no cliff effect. The consumer model is a gradual introduction process, and the traditional logic of “CCB to cloud revenue” still holds true in the short term; the future evolves towards results (outcomes), requiring consumption and adoption, and all incentives lean towards adoption, but this is not something to worry about in the 2027 bridge. It is worth mentioning that the transactional business, which has been sluggish for a long time, has basically bottomed out, and achieved good year-on-year positive growth during the quarter.
Jackson Ardle, Capital Markets Analyst at KeyBanc
Question focus: As AI becomes a decision factor, can the growth rate of cloud revenue in the Americas keep up with EMEA and APAC?
Christian Klein: AI and data are equally strong across all regions, and have been highly consistent over the past four quarters. North American customers are often pioneers of AI, and the regulatory environment is more relaxed than Europe. In fact, the US is leading, followed by EMEA and APJ.
Dominic Assam: Precisely because the US is a pioneer, its cloud penetration rate is already higher. This is the main reason for the difference in growth rate.
Bernstein analyst Mark Mudler
Question focus: What are the bridging factors between the first nine months and full-year revenue guidance?
Dominic Assam: The visibility of Q4 cloud revenue is already quite high. This is the reason why we moved to the upper end of the range, implying cloud revenue growth rate of about 23%. There is no mystery. There are several points where the growth rate declined from 25% in Q3: two negative factors of 200 million euros in the same period last year made the base lower; SmartRecruiters only merged on September 11 last year, and the resulting merger and acquisition gains will disappear in the year-on-year period (about 0.5 percentage point); and the CCB growth rate slowed from 27% in Q3 last year to 25% in Q4. License revenue has historically been unpredictable, but its share is getting smaller, and service revenue continues to decline on a quarterly scale of around €1 billion, as the practice of converting billable hours into adoption of support pays off handsomely.
UBS analyst Michael Brist
Question focus: Q4 Confidence in the cash flow target of around €2 billion, and are the migration credits starting to be used by customers?
Dominic Assam: Migrating credits will eventually affect free cash flow, but it won't be enough to shake the €10 billion yearly guideline. Q3 Accounts receivable are too high. We are tightening repayments. In a high interest rate environment, customers sometimes delay payments due to reasons, and we will ensure that the money is actually recovered.
Deutsche Bank analyst Johannes Schaller
Question focus: Can model routing be a potential cost savings source for gross margins?
Christian Klein: We're picking the best “performance and price” model for each proxy use case, and the share of open source weight models is increasing (both externally and internally); this is just the beginning. In terms of gross margin, some one-time investments and hardware prices are high this year, but next year we will promote large-scale AI automation in the deployment, monitoring, event and full application management processes of cloud operations and reduce third-party outsourcing, so we maintain confidence in the healthy development of cloud gross margin.
BNP Paribas analyst Ben Castillo-Bernaus
Focus of the question: The trend of “headless” (headless) software and the differences between “model agnostic” and “exclusive collaboration” strategies. Has SAP's strategy been verified by customers?
Christian Klein: We are using AI to train models, continuously optimize the accuracy and results of agents, and simultaneously optimize the “price-to-result” cost performance ratio. The progress is good from quarter to quarter. Regardless of the model, giving customers the right to choose is our advantage.
Bank of America analyst Frederick Blanc
The focus of the question: What are the factors that change profit margins this year and next, as well as operating leverage and room for continuous expansion?
Christian Klein: We are implementing an “all-in-one” AI program in-house, applying AI throughout the R&D and delivery cycle, and adjusting third-party contracts over the next 12 months. The sales side Joule Work has covered customer briefings and account planning, the financial side uses Tableau AI and Joule Work for simulation and reporting, and the HR side uses TechWolf for skill transformation. We expect efficiency gains of up to €3 billion over the next two years, and AI will have an impact on every cost in the income statement.
Dominic Assam: These levers are extremely strong. At the same time, the sovereign cloud, retail business, and BDC structures are also expanding, and multiple acquisitions will mainly affect operating profit margins. We are committed to controlling the growth rate of total operating expenses to 80% to 90% of revenue growth, and at the same time place more emphasis on the absolute amount of money earned from opportunities rather than simply focusing on profit margins.
Evercore ISI Analyst Kirk Marten
Question focus: After the transactional business stabilizes, can 2027 be a smooth wind?
Dominic Assam: “Tailwind” is an exaggeration; in fact, it is far below the Group's average cloud growth rate. It only returned to a positive range from near zero growth before. Essentially, it is “reducing headwinds” rather than “contributing to tailwind.”
Citizens JMP Securities Analyst Patrick Wallavins
Question focus: How does SAP view Concur's travel expenses business in the face of new competitors?
Christian Klein: Concur is our sizable business and focuses more on the “expense-to-pay” (expense-to-pay) process. The customer has deeply integrated it into the financial plan, and the business is very stable; the reservation side is very small. Among all line of business solutions, Concur's agency adoption is the highest. We will also announce new partnerships with credit card companies and airlines. The ecological interest is strong, and the business is healthy and intact.
Citibank analyst Balaji Tirupati
Question focus: The internal composition of the backlog of orders, and is ECC's maintenance business biased more towards migration or AI/BDC as it approaches recession?
Christian Klein: Today, every RISE journey is not just about going to the cloud, but also using AI to reshape business processes. Customers will use S/4, various business line solutions, and BDC platforms at the same time to build a unified and harmonious data layer between SAP and non-SAP. The RISE contract has become thicker overall, embedding more solutions, process-layer AI and BDC, and customers are also using various AI migration tools to speed up implementation. This trend is also reflected in the strong Q4 pipeline, where every RISE transaction is AI-led and data layer essential.