The bottom line
SAP's Q3 2025 results reframe the cloud migration case around embedded Business AI and the Business Data Cloud rather than support timelines, backed by five straight quarters of 25%-plus cloud growth funding that R&D.
An analysis for ERP decision-makers — read SAP's Q3 2025 call through the lens of what the numbers fund, not how the keynote frames it.
The views below are my own, drawn from SAP's publicly reported Q3 2025 results and earnings commentary and from my work in the ERP ecosystem. Nothing here reflects inside knowledge of SAP's plans, and SAP would frame parts of this differently. I welcome that conversation.
Why the earnings call matters more than the keynote
A quarterly call is written for investors, but it is more useful to customers than any product deck. Public-market scrutiny strips out the aspirational language and leaves the things that actually bind a roadmap: where revenue is committed, what R&D is funded, and which bets the company is willing to defend to analysts.
SAP's Q3 2025 results, reported October 2025, are worth that kind of reading. Cloud revenue grew 27 percent — the fifth consecutive quarter above 25 percent — and the current cloud backlog, a measure of committed future revenue, rose 27 percent to more than €18.8 billion. For anyone running ECC or S/4HANA, those are not vanity numbers. They tell you the innovation budget is funded and where it is pointed. Three themes from the quarter should shape an SAP roadmap over the next one to three years.
AI moves from feature to "key enabler"
The headline strategy SAP repeated is "no apps, no data, no AI" — the argument that a model on its own is worthless without business process context and clean, governed data underneath it. That framing is self-serving, but it is also directionally correct, and it explains where SAP is spending.
The mechanical shift worth noting is the move from individual "AI agents" to persona-based "AI assistants" that act as team leads, pulling in the right agents to run an end-to-end process. SAP put concrete adoption figures behind it, all of which should be read as vendor claims until your own estate proves them:
- A large building-technologies firm: roughly 3,000 hours a year saved automating IT system monitoring.
- A global industrial supplier: about 2,500 hours per customer service center per year via a services agent.
- A cement manufacturer: purchase-related process times cut by 50 percent with a sourcing agent.
- An instruments maker: Business AI used to raise accuracy in a mission-critical spare-part quotation process.
SAP also cited productivity gains of up to 40 percent for planners using these assistants, and announced a partnership with Perplexity to let its Joule copilot reason over both SAP and non-SAP data. SAP further noted that some customers are pulling transformation deals planned for 2026 forward to gain earlier access to Business AI.
The CIO takeaway is not the percentages. It is that SAP is repositioning embedded AI as the central reason to move off ECC and onto the cloud suite. If you treat that as marketing and do nothing, the functional distance between your static footprint and a competitor running these assistants widens every quarter. The work to do is unglamorous: audit where your finance and supply-chain teams still do repeatable manual work — close, reconciliation, sourcing, monitoring — and decide which of it an assistant could absorb.
The Business Data Cloud goes open
In Q3, SAP positioned its Business Data Cloud (BDC) as the data layer underneath every AI deal, and made a meaningful pivot: rather than keep analysis inside its own walls, SAP is opening that harmonized data outward.
The vehicle is BDC Connect, described as a "serial copy service" that links the BDC directly to partner platforms such as Databricks and Google BigQuery, giving customers live access to outside AI ecosystems — including Google's Gemini models — against SAP's business-context-rich data. SAP cited a roster of global enterprises adopting the BDC alongside their cloud transformations.
This is the more consequential signal for architecture. SAP is conceding that customers run multi-cloud data estates and is trying to become the indispensable semantic layer across them rather than the sole processing destination. For a CIO, that reduces lock-in friction on paper — but it also means the BDC becomes a dependency you are funding whether or not you have a coherent plan for it. Decide deliberately what the BDC is for in your estate before it becomes the default answer to every data question by inertia.
A "sovereign cloud on-site" for regulated industries
SAP reiterated that it is an application and solutions provider that partners for infrastructure rather than building its own data-center capacity — a deliberate contrast with vendors that own the full stack. This quarter that posture produced a notable new offering.
SAP announced a "sovereign cloud on-site" model that deploys its full cloud portfolio — including the Business Technology Platform, the BDC, and AI — inside a customer's own data center. It is aimed squarely at highly regulated buyers: public sector, defense, banking, and governments that require the highest levels of data, operational, and legal sovereignty. SAP paired the announcement with a public-sector framework contract with the United States Army and an OpenAI partnership for the German public sector.
The honest read: sovereignty has been the single biggest blocker to cloud adoption in regulated industries, and delivering the SaaS innovation stack inside a customer-controlled environment is a real differentiator against hyperscalers and against staying on legacy on-premise. If you operate under residency or sovereignty constraints, this widens your option set — but it also adds an operating model to run. Sovereign-on-site is still a system someone has to patch, monitor, and keep current; the deployment location changes the control story, not the operational load.
What the financials tell a buyer
SAP framed its results as the foundation that funds long-term innovation, and on the numbers that holds. Total revenue grew 11 percent, non-IFRS operating profit rose 19 percent, and SAP reiterated a goal of growing headcount significantly below revenue — citing internal AI use, including Joule answering thousands of employee questions a day, as the lever.
That last point is the one to sit with. SAP is telling investors it intends to grow output faster than headcount by absorbing routine work into automation. It is a sound strategy, and it is precisely the strategy a customer should apply to its own ERP support layer. The cost in an ERP estate is rarely the license; it is the accumulated manual, repeatable work around the platform — the tickets, the monthly close steps, the reconciliations, the integration babysitting. The vendor is compressing that internally. The question for a CIO is whether your own operating model is doing the same, or whether your run-rate simply holds flat while the vendor's does not.
A decision framework for CIOs and CFOs
Pulling the Q3 signals into a roadmap:
- If you are still on ECC: the migration case SAP is making is no longer primarily about a support timeline. It is about access to embedded AI and the BDC. Evaluate it on whether those capabilities eliminate measurable manual workload in your finance and supply-chain processes — not on the keynote's ROI percentages, which are claims until your estate proves them.
- If you already run S/4HANA Cloud: you are funding the Business AI and BDC R&D. Inventory the manual work the assistants are designed to absorb and pilot them against it. The return is in eliminated effort, not new licenses.
- If you operate under sovereignty constraints: sovereign-on-site widens your options, but scope the operating cost of running the stack in your own environment before treating it as a free unlock.
- For everyone: treat AI assistants as a way to retire repeatable work, not as a reason to add it. The economics only improve if the automation reduces the human effort around the platform rather than layering on top of it.
What I took away from the quarter
- The innovation budget is funded and pointed at AI. Five quarters of 25-percent-plus cloud growth means the R&D dollars are real and concentrated on Business AI and the BDC. A static estate falls behind on capability, not on support dates.
- SAP is opening the data layer. BDC Connect signals SAP accepts multi-cloud reality and wants to own the semantic layer across it. That is good for buyers — provided the BDC is a deliberate choice, not a default.
- Sovereign-on-site is a genuine unlock for regulated buyers — and a genuine operating commitment. The control story improves; the work of running it does not vanish.
- The vendor is compressing its own cost base. SAP intends to grow well above headcount by absorbing routine work into automation. The same logic applies to the support layer around your ERP, and it is the difference between a run-rate that compresses and one that merely holds flat.
Analysis by Ravi Madhavan, offered as one practitioner's reading of SAP's Q3 2025 results. Originally published on LinkedIn. Substantive disagreement or alternative reads welcome. Allari runs the ERP support layer for JDE, SAP, Oracle Fusion, and NetSuite estates — measuring the manual, repeatable work these earnings point to and finding where it can be automated.
