🎁 Black Friday: 50% off Unlimited and 30% off Pro annual plans
SAP SE (ETR:SAP)
Germany flag Germany · Delayed Price · Currency is EUR
187.02
-6.90 (-3.56%)
Nov 19, 2026, 5:39 PM CET
← View all transcripts

Morgan Stanley European Technology, Media and Telecom Conference

Nov 19, 2026

Summary

AI demand supported cloud revenue and results stayed broadly on plan despite macro headwinds and dilutive acquisitions. Management plans a new AI-related revenue KPI after Q4 and expects AI productivity to support migration, monetization and lower headcount over time.

George Webb
Head of European Equity Software Research, Morgan Stanley

All right, I think we'll kick off. Good morning, everyone. My name is George Webb. I'm the Head of European Equity Software Research at Morgan Stanley. I'm delighted to be joined by Dominik Asam, the Chief Financial Officer at SAP. Welcome, Dominik. Good to have you in Barcelona with us.

Dominik Asam
CFO, SAP SE

Well, thanks for having me here. Really appreciate it.

George Webb
Head of European Equity Software Research, Morgan Stanley

Just before we kick off, I'll read the safe harbor statement. During this fireside chat, SAP will make forward-looking statements, which are predictions, projections, or other statements about future events. These statements are based on current expectations, forecasts, and assumptions that are subject to risks and uncertainties that could cause actual results and outcomes to materially differ. Additional information regarding these risks and uncertainties may be found in SAP's filings with the Securities and Exchange Commission, including but not limited to risk factors section of SAP's 2025 annual reports on Form 20-F. Look, with that out of the way, maybe, Dominik, we start with the bigger picture. There's been a lot for businesses out there to navigate this year, from macro to geopolitical to the pace of change of AI.

As we look back over 2026 so far, how do you frame the year that SAP's had from your perspective, and how do you think about the deal pipeline as you look through this final quarter of the year?

Dominik Asam
CFO, SAP SE

For sure, the macro environment was not necessarily very conducive and continues to be not so easy. There are relatively high interest rates in the U.S., even some European countries. I have seen some historic analysis that when you hit that kind of 5% level, interest rates start to get teeth and have some dampening effect. The oil prices are still much higher than at the beginning of the year. Luckily, also, there was no kind of catastrophic meltdown on supply chains in the Strait of Hormuz, so it was contained. Really, I think the key consequence of all the geopolitical tensions is rapidly increasing prices, and that's about it.

On the other hand, we have seen still a very strong structural demand for everything that's paying into the AI story, and I'd say these two taken together have resulted in the fact that we have been pretty much on track of what we have announced at the beginning of the year. I said macro, some headwind, but the credibility of SAP to have a seat at the table when it's about transforming enterprises to be better prepared for artificial intelligence have been giving us a little bit of upside even on the cloud revenues, as you've seen. On the margin side, it has not necessarily boosted the margin because we've also been quite active on embarking some tuck-in acquisitions that were dilutive. So that's how I look back on the year so far.

I also want to stress, as always, 50% of the bookings are coming in Q4, so we are feverishly working on that, and that's why we are still sticking with also the guidance we gave at the beginning of the year on the slight deceleration of the CCB growth.

George Webb
Head of European Equity Software Research, Morgan Stanley

When you think about that slight deceleration, just to touch on that point, I mean, it has been a good first three quarters. We haven't actually seen a deceleration in that CCB yet. Is it purely an abundance of caution? Are you seeing anything in that kind of pipeline as you look into Q4 that actually makes you feel I think we talked to you maybe at 24% being a higher ceiling for the fourth quarter, that makes you kind of feel that that deceleration is the most realistic outcome versus something that's maybe a little bit more stable year-over-year?

Dominik Asam
CFO, SAP SE

I mean, I have, of course, the benefit of really seeing internally all the small puts and takes and how they compound, and I'd say even if you go to that slight deceleration with what you just pointed to, that kind of ceiling of 24%, as you said, that is, that is not a dramatic change. It's really a pretty continuous evolution of what we are seeing. So it's actually very much in line with expectation. Then we also have to be careful on all the puts and takes from these smaller M&A things that can move the needle a little bit. So, if you think about that, also when we guided, we had a little bit of a slight uptick on ratio on the CCB growth. So if you adjust for that, I think it's all pretty much on track.

George Webb
Head of European Equity Software Research, Morgan Stanley

Okay, cool. Well, let's move on to some of the kind of bigger strategic questions. I want to start off on the S/4HANA migration, clearly a key growth driver for SAP as we encourage your customers to move from SAP ECC to S/4HANA. I think what's a little bit less clear sometimes is tracking where we are in that migration curve. Customers are clearly at different parts of the journey. Some will be on the kind of implementation phase, some will be live, some will be yet to commit. Could you kind of talk to us when you look at the overall base, where you feel you are in that journey and how you feel it's going for customers?

Dominik Asam
CFO, SAP SE

Yeah, sure. We're making good progress on that journey. That is one of the legs that gives us the top line growth we have. It has been a very strong contributor in the past and will continue to be so. I think we have to differentiate in terms of what's the urgency with customers. There's, of course, a certain difference between our SAP ECC and older customers, where the extended maintenance will end end of 2030. Then there is S/4 customers, which basically have no time pressure whatsoever, where it's purely on the kind of merits of the system that you can then move them to the cloud, which we do a lot to. Now, there is one very important issue, which is what we're selling is actually a transformation to make the company future-proof for AI and leverage AI in enterprises.

There are two parts to it for the business case. One is the upfront investment to get to that stage, and I think there would be very few customers who would say, I would not love to sit on a highly integrated, clean core S/4 system. Everybody would love to be there. Then the problem is the price tag. The price tag is really not the SAP price tag, it is the whole cost of transforming and the risks around the transformation and all of that.

There is also a very important AI initiative at SAP, and I would say in the early innings, maybe the most important, is to deliver the tools to the customer that automate more of that work and bring that price tag down because it is so big, b ecause the business case lives from bringing that price tag down and augmenting the returns, the rewards you reap once you have installed the system.

On both sides, AI plays a crucial role. But I think there is still a little bit of an underappreciation how important the first part is, and we will really double down on that massively in 2027 because that will be the key to really also unlock the further upgrades. It is a little bit of a two-edged sword because as people see that these tools are working, they say, okay, this is a moving target. So you get a certain productivity this year, but maybe one year later, the productivity will be even higher. So it is a deflationary environment almost.

Currently, we see that these kind of acceleration effects from making the tools powerful and reducing the price tag, that is a positive that would kind of incentivize more customers to go fast. On the other hand, they could also say, maybe in a year's time, it is a little bit bigger. So I would say these two currently are in balance. To cut a long story short, I see that gradual grinding which we have seen over the last years. I think that indeed by creating these tools, we have also created the opportunity for customers who have big estates to move on time.

For instance, a couple of weeks ago, I had a gigantic customer who was before telling me, well, I will not make 2030 anyhow, and this is a big chunk of the business is very ERP-intensive, but it is low margin, and should we invest there? Now that they see what we can do, they say, oh, maybe I can get it done by 2030, and yeah, maybe if you can help me with all that, we can get it done, and we will actually move to S/4. So I see some good stories there, and that is the key because when we do not have the customer on S/4, on RISE, or on GROW, the whole agentic thing does not work anyhow. So that is super important for us to continue that.

Then, of course, we light the next stage of the rocket, and in that regard, what will be important is really now the adoption. We will also not only deliver the product tools as we speak. There's a lot of products going into general availability every day now. Then it's all about driving that consumption. We, by the way, want to make that more tangible for our investors, but do that not in the interim, but post Q4 in January. We want to create a new KPI because the kind of cloud ERP KPI at like around about 90% is not so meaningful anymore. For revenues, we want to really put everything where we basically convert personal expense into agentic IT dollars.

That's not only the runtime, so what the users do in the system with AI, but it's also what we do on the transformation part, plus the BDC, which sits in the middle, because that BDC is kind of the core system to enable that whole journey. This is something which we are going to break out in revenues in January, and I hope you find the numbers interesting in terms of both the magnitude we have already achieved there and the growth we can achieve there, which then leaves the more traditional application system, so application business, which is the runtime of everything that's not really leveraging AI, but it's just traditional software business. What you will see is that also that business is doing better than the market, so give you more feeling about how we are progressing on that journey.

George Webb
Head of European Equity Software Research, Morgan Stanley

Just to be clear, that KPI could be some form of the BDC revenue you're generating plus the-

Dominik Asam
CFO, SAP SE

All the AI stuff, but also all the things we do within the agent-led tool chain for the kind of tapping into that body leasing business and reducing these hours, and because that's another conversion of personal expense, so to speak, into AI, and it's super crucial. It's actually, as I tried to explain, at the base of our success story on the AI transformation, because if we don't get people over that hump, it's anyhow difficult.

George Webb
Head of European Equity Software Research, Morgan Stanley

That is very clear. There is a number of things I want to jump into from those kind of statements. I think the first one is how AI is changing those migration curves, because that is one of the points you talked about. I think earlier this year, you talked about it reducing the effort, the cost, the time of doing these migrations and transformations, which are big processes. Can you give us a flavor of the types of savings, the types of time reductions that you are seeing your customers achieve on migrations today based on these tools?

Dominik Asam
CFO, SAP SE

Yeah. I think the best example is, of course, also what we do internally, and you know that we have committed to extracting about a EUR 2 billion price tag out of that, EUR 2 billion plus. I think not a surprise that a lot of that is on just the very low-hanging fruit of enabling all our developers to use the best tools on vibe coding to massively boost productivity, and that is certainly there. Where it takes longer, but is also coming, is on the big end-to-end processes in admin functions like finance. There, I would say the productivity gains are definitely accelerating massively. If you look at the specific numbers of SAP, if you go in like G&A ratios over many years, you see that our admin functions have been able to generate already like mid-single productivity per annum.

I would say very roughly, there should be a multiplication of that number. What exactly it is depends a little bit on the function, but I would say very roughly 4x, 3x is of that productivity gain. So move it from like mid-single digit to something around 20% over a one-and-a-half-year timeframe. I think that is the type of order of magnitude that we think is realistic. That has already helped us, by the way, this year. We have embarked a lot of dilutive acquisitions, there are three acquisitions, the fourth one now coming, which are J-curves, requiring a lot of investment. We are also leveraging that internally to create the room in the P&L to absorb that and still deliver the bottom line results. That will come. So it is a quite powerful journey. Of course, it is not standing still.

It is going to evolve over time. We think that the way people will work on our tool will be completely revolutionized. Still today, you have the users navigating to some data, extracting the data, making sense out of it, aggregating it with other data, and of course, you can automate that to some degree. Then again, when they have figured out what they want to transact, recall that SAP is all about transacting stuff, yeah? Then they have to trigger another transaction, and what we really want to achieve is put a cockpit with our Autonomous Enterprise, a curated cockpit in front of every single coworker of the company that makes all the stuff travel to the user automatically, spontaneously, so to speak, in anticipation of what that guy needs, and then also delivers recommendations that just can be triggered by hitting a button and say yes.

And of course, if you do that more and more and you see the guy says, Yes, yes, you automate it away. That is a very ambitious vision, but it has a lot of potential. I think when you talk about these type of numbers, I just mentioned it is the tip of the iceberg. Over time, it will go much further than that, because if you think about these two worlds, how different they are in terms of how much work you can get done, it is radically different.

George Webb
Head of European Equity Software Research, Morgan Stanley

That is very clear. I think one of the things you also mentioned was the different kind of there are different pockets of different types of customers that you will be taking to S/4HANA in a cloud environment. There is a cohort of customers who are on-premise, S/4 customers. What is your messaging to them? I guess they also see the carrot of what the platform can offer in the cloud, but how do you get those customers without the same kind of function of—

Dominik Asam
CFO, SAP SE

It is all about the fact that we say your life is so much easier if you are going to RISE on everything we do on AI, and we invest a lot of money and showcase the Autonomous Enterprise and say, if you want to have access to that Autonomous Enterprise, well, you will need to go to the cloud with us. Then there is a TCO debate to say, look, it is not even more expensive for you, because if you look at everything you pay in the managed services, everything you pay on your own infrastructure, and if you bundle all that, we tend to even be, on a pure TCO comparison, quite competitive. So basically, it is kind of an option value you create. What is paramount now is to really, for SAP to get that seat at the table on the agentic game.

I think sometimes there is a perception that maybe every enterprise wants to have one vendor spanning the entire enterprise doing everything for that agentic work. I am deeply convinced that is not the case. I am deeply convinced about substance over form, and that people will gravitate to where the real intellectual property lies, the semantics, the process know-how, how these things all hang together. Luckily, there is a lot of users and enterprises who are extremely focused on SAP and spend a huge share of their time on that system. I do not see the value add of sandwiching a thin postman layer in between, because there is no value add, frankly, from that. On the other hand, there are certain personas in the company which have nothing to do with an SAP system, a guy in marketing maybe, I do not know.

Why the hell would he use Joule to do the work done? What we are convinced, and by the way, every customer wants to avoid lock-in. You don't want to have lock-in with SAP, you don't want to have lock-in with your miraculous agentic orchestration layer provider.

What we see is that these large enterprises who have already, who are probably more advanced on this, that they kind of figure out what are the key kind of vendors, where they can also play the competition and say, take SAP as an example, yes, of course we want to consume your Joule where there's good value for money, but if you're too greedy, we want to relegate you down to the system of record and use something else for it. What these customers are pressurizing all the vendors on is say, please make sure that we can have these three key vendors we want to have, and they work together. Actually, for the user, they can use either Copilot or Joule or whatever, but then your stuff is headless or the other way around, but it has to be kind of federated.

When we have these discussions, we have a high degree of confidence that there is a seat at the table for us, and that's happening now as we speak. These kind of decisions, architectural decisions, are happening now, and this is why we are also so keen on whenever we see another opportunity to differentiate that offering, to make it even harder to resist, so to speak, like the acquisitions we've done in the more recent past, we think it's important to work on time to market. Because what does it help you if you have a better capability two, three years down the road, but the customer has gone another way? Frankly, that's a little bit the problem SAP has been facing on the data lake stuff, yeah? I mean, we have not invested much, and others have claimed their territory. You know the names.

It's SDK as an SAP system, so we cannot go back and say, can you please write off all your investment with these vendors and go back on something with SAP?" By the way, now we have the capability with Dremio to do that. This is more for greenfield new stuff, because honestly, where the chips are already allocated, you don't just write off your investment in these type of platforms. Now is the time where you want to have the seat at the table. I think the Autonomous Enterprise we have showcased and bring to life as we speak has convinced a lot of customers to go that way. That's the key, and that's everything we focus on right now.

George Webb
Head of European Equity Software Research, Morgan Stanley

I mean, the history of, I guess, software for lots of large enterprises is a very fragmented stack of different applications that are not interwoven together. Is your perspective when you talk about this that the reality or a likely outcome will be that you have your Joule system that sits within the top of your entire kind of suite of products at SAP, and that can handshake off to other parties, agents, and it won't be one layer that necessarily is the dominant force?

Dominik Asam
CFO, SAP SE

I think it will not be a large number. It will be probably not, definitely not more than the fingers on the hand, maybe less, because it's just difficult to train all the users on different systems and so forth, even if they're super intuitive. The integration is something that also these vendors, they will need to spend money building these pipes to the other vendors. We do that with BDC, as you might know, and some other tools, like also on Copilots, and we are the founding partner of the Agent2Agent protocol and so forth, so we promote that. That's the vision we have. I can take one example. Why have we invested so heavily in HR? You've seen most recently TechWolf acquisition, and then we did SmartRecruiters, and we are extremely pleased with the performance, especially SmartRecruiters.

Kudos to Rebecca, who came in with the acquisition and is really kind of breathing a new energy into our HR offering. The good news is, we clearly see huge conversion needs between finance and HR. We think that the transformation we just described here, where you have gigantic productivity numbers, will mean that you need to retrain a lot of people, displace people by tokens, manage onboarding, hiring, SmartRecruiters, but also off-boarding, training. So, in our case, we're talking about five-digit type of employees having to be touched in certain ways, not just doing their work, but being changed in some way. That's an industrialization of that job of HR. Now, as a CFO, I want to have my arms around the controls. I mean, are the authorizations moving the right way with all these people? Are they properly certified to do the job?

Is the money correct they get? The forecasting of the cost, I mean, in our case, three-quarters of the cost is personnel expense in some fashion. If you have to build all these pipes together and the analytics, build all these analytics yourself, it's painful, and we can deliver in one shot the solution to our customers to say, look, here are the dashboards for your enterprise to have full transparency and visibility, and not only get work done, but come to business outcomes. It's nice to process tokens, but in the end, it's about, have we hired the right people? By the way, the whole pitch about TechWolf is that we want to go from just intuition on managing skills to facts and figures-based empiric studies.

If you hire a certain type of profile and put it in that job, how is it affecting productivity, and can you learn from that for the next person you hire? So it becomes much more scientific and facts and figures-driven. That's just one example where I think consolidation is key, and this is why I always jokingly say there's only one thing I really agree with my beloved competitor Workday, which is finance and HR need to go together. Then you can make the same argument with supply chain. So if you want to know when a product is delivered, how to reboot stuff, you need logistics, you need a supply chain. In manufacturing, something has to be changed. So I think the breadth of our portfolio, that kind of suite idea, is helping us a lot in these times.

This is also giving us the big opportunity to up and cross-sell stuff. We showed it also in Sapphire, how much it's actually not only the conversion of installed base in the first step, but then the up and cross-sell later on. Plus then the whole new opportunity of the agentic layer. By the way, that's the other thought we had when we defined this ratio, what is part of the revenues we are going to disclose was, what is actually, in a way, uncharted territory for SAP? Where are we suddenly tapping into value pools that we have been not able to address in the past? So that's where I think, also given the resource intensity, it's a big boys' game. We don't think it's easy for small companies to compete.

I'm not saying that there might be not niches where some people have been much smarter than us. Take, frankly, SmartRecruiters. I mean, there are some agile startups. But the good news is, even if we fail to develop something ourselves big enough, we can absorb them and then put them into that mighty system of a sales machine of everything these enterprises need to then bring that capability to bear on a huge base.

George Webb
Head of European Equity Software Research, Morgan Stanley

Yeah, the strength of that platform and what you can offer, I guess, becomes that kind of strategic advantage. It has been for a while. One of the taglines in this kind of migration that we've become accustomed to as investors and analysts is that two to three times multiplier of value you see on kind of maintenance revenue as customers go on that journey. As we take in many more of these different areas we're now talking about with Business Data Cloud, AI, et cetera, is that framework still useful, or should we be evolving it in some way?

Dominik Asam
CFO, SAP SE

For the time being, I would still stick to that. Now, if really the BAIP is taking off big time, of course, it can go beyond. Let's not skin the bear before we've killed it here. For the time being, when I look at the conversions we do today, it's still at that type of level, and it's normally enough. What I think might change is that you might have a little bit of a higher conversion on the first step because some of the Joule capabilities will be bundled in from the get-go. They have to be part of the first RISE step because, as you said, we are pushing this to be the de facto interface of the customer because it's so much more powerful and gives them such a better experience. I wouldn't now see that this will radically change.

Honestly, this is not a precise number. We always said two to three, and there's a big difference between two and three. We want to also, frankly, because the number doesn't have the same assurance level as the hard debits and credits we have in our accounts. This is a management KPI with some precisions which are not to the last comma, so to speak. From that perspective, it would be too early to just say this is moving up. Of course, what I can say is that you will see a much higher growth in what we call BAIP than you will see in the traditional application business, and that we think that this is the next stage of the rockets that we'll need. Then at some point in time, the conversion will also run out of steam.

Come end of 2030, we are probably by then down to a mid-single digit maintenance base, which is then largely S/4-driven and some customers, for whatever reason, that cannot move to cloud. Then at that point in time, something else has to do the heavy lifting and will—

George Webb
Head of European Equity Software Research, Morgan Stanley

Yeah, presumably AI would be a big part of that story. Maybe if we kind of continue on that AI piece, there's different strands. We've touched a little bit on those across applications, agents, data platform. Can you kind of articulate from your perspective what the simplest way is to think about the different buckets you feel you're playing in at SAP, and where you think the most of the opportunity perhaps sits?

Dominik Asam
CFO, SAP SE

I think it's quite balanced, and I can say that with confidence because we've just finalized, so to speak, the planning for 2027 on how this, the BAIP revenue will be composed. We are not a one-trick wonder here. As I mentioned before, there is a huge opportunity to disrupt the transformation work, the body leasing work by AI on the step of changing the way the customer works. There's a huge opportunity to monetize the value we create in the runtime once the system is installed. There is a big opportunity on the BDC, Business Data Cloud. Some of you heard me comment on, like a couple years back, that each of the AI, the BDC, and also the public cloud GROW opportunity are a billion-plus opportunity in the outer years of our planning horizon.

I think that has nicely played out, and BDC is really the one that has been very fast deployed. I think our leverage in these discussions, having onboarded more partners, is increasing. The leverage in the discussions also increasing because we've now brought Dremio into our own portfolio, which gives us this kind of standard lakehouse architecture on what I'm told by the developers, Apache Iceberg is the kind of de facto standard, and we have now this capability to leverage that. We can always now pitch to a customer that has a new problem to solve with a data lakehouse to say, whatever your prior vendors have, you can bolt it onto that, but here's an alternative, and we can price that aggressively.

All these three buckets, the transformation, the runtime, kind of displacing users, so to speak, or making the users much more productive, and then the kind of integration with BDC, Integration Suite, and BTP, this is the growth engine that will fuel that journey. It's quite balanced, actually. If I put it in numbers, it's probably closer to a third, a third, a third than to something is like 20%, 20%, and then something else is 60%. So it's not exactly a third, a third, a third, but it's quite balanced.

George Webb
Head of European Equity Software Research, Morgan Stanley

Yeah, so a balanced mix. One of the things that I think Christian and Muhammad were quite candid about at Sapphire in May was some of the limitations of the first versions of some of the AI and Joule that you brought out through back end of last year into early of this year. You've rolled out quite a lot of upgrades and extensions of your AI offerings through the summer, as we come out of the summer. How should we be thinking about how that latest wave of innovations has landed? How, as a management team, do you feel about the confidence in what you now have versus perhaps where you stood in May?

Dominik Asam
CFO, SAP SE

Yeah. I mean, first of all, in terms of the fastest uptick, because of the phasing of the tasks the customer has to go through, has actually been in the agent-led tool chain, so that people need to transform, and they need a lot of kind of. They dread the cost, and they buy a lot of these tools to alleviate the cost and pull in the timelines, which is very valuable to them. The other thing that is, of course, now coming to fore, and we see that also internally, is that we have deployed internally Joule Work, and we have also had some customers deploy Joule Work very successfully. Especially where the center of gravity of what that kind of interface is looking into and trying to get done, when that is very SAP-centric, they do see the differentiation in terms of accuracy and performance.

That is the other big lift. I would say, of course, also Joule Studio is super important because the agents that now, still nothing has changed on that, 400 agents on the roadmap to go in general availability by the end of this year. Not all these agents will immediately solve exactly what the customer requires, so there will be some bridging needed. Now you can, for everything you want, call some forward-deployed engineers in and build your own stuff from scratch. There is a lot of companies who make a huge living out of that. Not many, but some. What we are offering is basically a much more competitive solution for more standardized processes where maybe you do not have to reinvent the wheel, and we provide these agents, but then there is some customer-specific extensions required.

Joule Studio is the next leg up, where we basically see a huge opportunity because every customer will need the ability to extend and do that in an ideally no-code way, but then certain enterprise-grade things cannot be done by low code, and then you need still some pro code capabilities. Joule Studio gives you really that capability to push the functionality in a no-code way, and then still allow, on the same platform basically, a developer kind of finalize the tool to make it really enterprise-grade, enterprise-ready from all the authorization compliance topics and so forth you have. So that is another thing. Even if I look at these three buckets, in the AI bucket itself, there is certain diversification because there is some revenue from Joule Work, which is the UI basically. Then there is revenue from Joule Studio. There is revenue from the agents.

There again, we think that the suite approach helps a lot because what we are going to do is to sell credits for all that. These credits are in a kind of commit-to-consume way, so the customer buys a certain bunch of credits. The IT department, this is another advantage we have. The IT department does a project with SAP anyhow. They do not want to come back for every credit every year. They want to put it into that IT budget and then not be bothered with a discussion with a nasty CFO like myself anymore for some time. Then they buy that credit, and then the good news is the risk of the customer not using these credits is much lower than if you are a niche competitor because you have the statistics, the central limit theorem and statistics.

If you have several things you consume, you can, with more accuracy, forecast with a lower error margin, relative error margin, what you need. That is another huge advantage we have vis-à-vis niche competitors, where they can only do that for a very tiny part of the company. This is why we think that is a broad movement, and already 2027 will strongly benefit from that. You will also see in 2026 that also 2026 has benefited from that motion already.

George Webb
Head of European Equity Software Research, Morgan Stanley

Yeah, just on that topic of credits and the idea of pricing models evolving across the software spectrum. Safra, you were talking about the perspective that maybe 10% of cloud revenue that is consumption today moves to at least 30% by 2030. Maybe just taking a step back from that, what is the right pricing strategy for SAP? What do customers actually want to see out of those models? What will they accept? What would they push back against as being maybe a little bit too uncertain for them?

Dominik Asam
CFO, SAP SE

I think there is two key ideas in terms of what they look for. First of all, they like skin in the game, so there has to be a value creation and not just pay tokens, but pay for outcomes. Of course, then they want to contain the value they have to share, so to speak, with the vendor to as little as possible. In that regard, it matters a lot how differentiated is what you do. If you do something that five other guys can do, they will ruthlessly play the competition, and the lion's share of the value will accrue to them, and you will not get much. But if it is a differentiated solution, well, you might be a little bit more demanding in terms of what pound of flesh you take out of that.

The other one is also, how do you manage the dilemma between predictability on spend versus You want to have predictability. On the other hand, it is a little bit of a contradiction, the first thing, because if you do not use it much, for the first thing, you want to have purely consumption-based, and for the other one, you want to have a subscription, basically.

We try to marry that in that kind of credit idea to create a pool, which is like a commitment of the customer to consume a certain amount of credits, and then they can flexibly allocate them, b ecause the variable cost of these solutions is also much higher in some cases than it used to be because you need to run GPUs for them and the more complex the scenarios, it is also hard for us to predict what exactly will be the price of each of them.

What I am told by our pricing specialist is that actually most vendors that are offering these consumption bundles are flexibly then repricing within the bundle. I think it is the tit for tat, you heard me comment on this customer desire to federate key vendors to also play the competition between them, to say, I can shift certain tasks, especially on a trending layer between different vendors to avoid lock-in and being exposed to price increases.

On the other hand, if that is the case, I also say as a vendor, okay, but then I also have to dynamically price my solutions because I need to see how much token I need and how differentiated it is. That is the standard we are going to bet on, that we sell these credits and say, today, there is a conversion of these credits into certain scenarios. You can use so much runtime on that application or Joule Studio or what have you. But we also take the liberty that we reprice that at a certain point in time, and if you do not like it, you probably also have your choices to do something else and develop with some other guy. I think that is the way the market goes.

It is a hybrid model where you have a certain visibility also for the customer because they bought these credits and they know that is the budget I have, and then they can do the same thing we do when I look at my token budget to see how can I optimize within a certain budget envelope and can I displace some tokens by something cheaper. On the other hand, we have a certain predictability that this is a revenue stream we can generate because it is a use it or lose it clause. When you have not consumed a certain tranche in any given year, basically we can catch up the revenues because they cannot use it later. Of course, then there are some customers pushing for some exceptions on that, and then we negotiate that. But that is the way it is going.

It is going into that hybrid model from our perspective today. It is still very emerging and nascent, so I am not sure this is the absolute end game, but that is what 2027 bookings activity will be based on.

George Webb
Head of European Equity Software Research, Morgan Stanley

When you think about that pricing model, I guess a question which often goes quite hand in hand is gross margins. When you think about consumption revenue, what's the inference cost that sits behind all of that? I guess to a certain extent, the models you choose and how you think about that framing is an important component. What are you focused on as CFO? Are you focused on for the next five years for SAP, what happens to the cloud gross margin? Are you more focused on the gross profit dollars? What should we be caring about and focused on most?

Dominik Asam
CFO, SAP SE

First of all, I want to say yes, you're right. This whole gross margin discussion is less stable than it used to be in a classical SaaS business where we basically have reserved instances from hyperscalers, and we know how much terabyte of capacity is absorbed by what type of application. We have a full spectrum. We have some loss leaders. It's not a surprise if we are giving Joule Studio in the first year till the end of this year for free to develop stuff that, of course, that is not having a good gross margin. Yes, the first imperative is to make it super impressive to customers that this thing can do a lot. So you might not be penny smart and dollar foolish on the model, but you want to use a real great model.

You take the other extreme, like I mentioned that we are really having a good take-up and huge growth already on the agent-led tool suite for transformation, which is more query tool, more commoditized in some way, but good enough to be differentiated. Once you have that model up and running with the frontier model, you start testing by de-speccing. By now, we are routing the workloads from that tool to three different low-cost models all the way to some stuff that is internally, we do also open-weight models for development, for instance. So we have a lot of people now figuring out what's the best bang for the buck where, and you see then the gross margin going to levels which are actually higher than what we have in the private cloud business or even the public cloud business.

So it really is a massive spread between loss-making and hugely profitable. There lies the opportunity. I mean, how many customers do you think want to have the world's leading research department on token economics with different vendors? We have 70 large language models on our platform, or even avoiding large language model at all. I mean, it's not very smart to use Anthropic to answer the question, what is two plus two? There's a lot of things we do in our applications which are more of the nature of doing two plus two than being a real super fancy agentic thing. So, just figuring out what problem is solved by what means in the cheapest and most effective way, in the most scalable way, plus the assurance you want to have in these models, that's an art, that's a research, that's a formidable development task.

While I understand that AI is so important that every customer will have certain domains where they differentiate and they want to do that research themselves to differentiate, like a pharma company developing molecules or my former employer, Airbus, developing the next aircraft with the lowest cost and fastest time to market, on record to report, procure to pay, hire to retire, and all these things, do you think it really makes a lot of sense for these customers to do that one by one? No, we do that for them, and as long as our solution is super competitive, I think we are fine because I am not aware of who else of the world is doing what we are doing right now in terms of investing so much money and figuring all that out.

This is why we think we have a seat at the table on the AI domain. Then we have some unfair, or unfair advantage, we are not allowed to say, but then we are developing some proprietary that is much more elegant, things like our SAP foundation models, which are super powerful in the type of SAP context, and to do predictions and stuff. So we blend some very proprietary things into it, plus the application richness. There are so many reasons why we are deeply convinced that, and we see that, we hear that from customers, they are not saying, look, I want to do all my accounting stuff myself now because I can vibe code my own SAP system. Because it is not that easy, and they might be able to do that.

Probably it would not be as effective in terms of the effectiveness of the tool, and I am sure it would not be more economical than what we could do, even if we charge them a fat gross margin, because we do that for so many customers at the same time.

George Webb
Head of European Equity Software Research, Morgan Stanley

Yeah, that is very clear. I want to focus a little bit also just on cloud gross margins more recently.

Dominik Asam
CFO, SAP SE

Yeah.

George Webb
Head of European Equity Software Research, Morgan Stanley

I think in the third quarter, I forget the number, I think it was down 70 basis points constant currency year-over-year. So there's been something that's been a few factors dragging in that mix. As we think forward over the next few quarters, can you just remind us what the kind of drivers will be to eventually get that back on a better track, if you think that's the realistic outcome?

Dominik Asam
CFO, SAP SE

Sure. I think the biggest driver, honestly, first of all, if you look into the individual buckets like private cloud and public cloud, that has been actually not the case. So they have been evolving quite continuously in a good sense. There's one factor is, of course, the lower gross margin of BDC, and as BDC is becoming bigger and we have to pay significant royalties to some partners and other partners, that's a little dilutive. We're also activating a lot on the partner ecosystem when we have these SOLEX models, so there's a bigger share also going to partners. So it's more the kind of mix effect that is a little bit of a headwind, and we try to compensate that as much as we can. By the way, recall we brought that gross margin up from around 70%- 75% over the last four or five years.

Now, we have a little bit of headwind. Yes, there is also the kind of introductory offerings on AI, where I mentioned Joule Studio as an example. Still we are feverishly working on the question, how can we maximize the incremental euros of gross margin that we embark, and also how can we stabilize the cloud gross margin as much as we can? This is why we also need the flexibility on the AI-related models to not be locked in for fixed prices for five years, but adjust that kind of menu at which the customer can convert the credits.

Our conviction is still that this will not be a big headwind going forward from what we see, because we've also seen the great progress we made in our own operations on the development tools to manage the token consumption and manage to have both uptake like hell, effective output, and mitigate the cost. On the other hand, I always insist on the real commitment we give in terms of real commit-to-consume to market is that 80%-90% total expense growth versus revenue growth, and we want to have some flexibility to play within that. So please don't kill us if there's 50 basis points headwind on gross margin. If on the other side I can take it out on some OpEx without harming the business, I think it's okay. That's the key.

If you think about what we can plan and not plan, my biggest uncertainty is not the token consumption on my gross margin in the kind of leverage, operating leverage discussion, the 80%-90%. It is actually more what type of M&A might we do, and we have seen it already this year that we have done some dilutive deals, and then we have to see can we create enough room in the P&L to absorb that. That is something that is much harder to plan because you never know whether you sign a deal or not.

We think that for this time to market reason that I just stressed before, we are going to continue to be active in that space, and these companies on average will not be hugely accretive because they will all be very innovative and research-driven things that where we can play the idea of bringing a head start on innovation to that mighty ecosystem and base of SAP.

George Webb
Head of European Equity Software Research, Morgan Stanley

Maybe that is a good question to end on actually, which is that framework for 80%-90% of revenue growth going into expense growth that gives you that headroom to keep moving the margins higher. Different factors. Yes, you are investing in AI. Yes, there is some gross margin factors. There is also a big and important topic about what does SAP want to do with hiring plans that look very different in the next five years versus history. How do you put those things together?

Dominik Asam
CFO, SAP SE

I think some people have been poking into the fact that our headcount has increased a little bit over the course of the year. There were some acquisitions, but also organically, and I think that kind of episode is over. If I look at the maturity now of the AI journeys and the productivity gains, we see a multiplication of what historically we have been achieving in productivity, lowest hanging fruit being on development. I mean, there we see very significant productivity, but also in admin function, we are going to see some really nice use cases where you can just take people out and get more done by less people. So, I think I would be really surprised if our headcount would not kind of go the other way going forward and then see, if anything, reductions on headcount.

You know that we've also committed to take a couple billion euro of cost plus out. I think we see a productivity boost also in our own operations that is exceeding the work increase we have by virtue of growth. That is why I think headcount should actually trend down over time from here.

George Webb
Head of European Equity Software Research, Morgan Stanley

That's very clear. Great. That's just about the amount of time we have. Thank you so much to Dominik for being here and sharing his thoughts.

Dominik Asam
CFO, SAP SE

Thanks. Have a great day.