Members of the media and the press are not authorized to be on this call. If you are from the media or the press, please disconnect from the call now. The content presented on this conference call is proprietary to and/or subject to the copyrights of Jefferies or third parties. Further, as a matter of legal compliance, we remind you that you must not attempt to elicit from any speaker at this event any material, non-public information, or other confidential information. Accordingly, the speaker may decline to respond to any question in his or her sole discretion. You may not publish or otherwise publicly disclose the name of or otherwise identify the speakers unless Jefferies permits it in writing. By attending this event, you agree to all these restrictions.
Great. I think we can get going with that out of the way. Good morning, good afternoon, everyone. For those that don't know me, I'm Charlie Brennan, and I manage the software and services coverage for Jefferies here in London. Welcome to our Structural Winners Conference series. Today, we're delighted to be joined by SAP. From SAP, we've got Dominik Asam, the CFO. Thanks for joining us today, and thanks for supporting the series.
No, thanks for having us. I really appreciate it.
Just in terms of the format, I'm going to try and run through a series of questions to cover the main topics of strategy, but also some of the key financial drivers across the P&L and the cash flow. If you do want to ask any questions, it's best to Bloomberg me or email me. I'll confess I'm not the best at multitasking, reading, and talking, but if it's appropriate, I'll try and weave any questions into the discussion. Having said that, I think I've got a fair sense of what the key debates are, so I'm hoping to cover off most of the important details. Just for your diary planning, in terms of timing, I'll do my best to try and keep this to 45 minutes. With that, let's get stuck straight in. Maybe I'll kick off with some high-level questions on strategy.
I thought it might be good to start with a simple refresh or reiteration of the core strategy. I used to be able to think of the investment case in fairly simple terms. It was the shift of on-premise customers to the cloud, on RISE contracts, extracting a 2x-3x uplift, and helped by a 2030 end of life of ECC. I actually think in the last six months, most of that's been ripped up. It feels like RISE no longer exists. [S/4HANA] hardly got a mention at Sapphire. The business suite marketing message is back. You've introduced AI agents. You've introduced the Business Data Cloud. 2030 feels like it's softened to 2033. It feels like there's a lot going on. Just in light of that, what do you think the simple mission statement is today?
Is it still a migration story, or is it a sales execution and cross-selling story?
Yeah, Charlie, I think it's really both. I think the reason why we didn't put so much limelight on our S/4HANA product is simply that this has by now turned out to be a blockbuster product and is in full swing. The way we can say that is that if you look at the kind of maintenance base of our customers, and those who are even still paying significant legacy maintenance revenues to SAP, a lion's share of them have already signed up for S/4HANA in some way. The transfer of S/4HANA to the move from legacy to S/4HANA is actually the bigger transformational move as compared to the RISE with SAP journey. The RISE with SAP journey can be as simple as a lift and shift of that S/4HANA system to the cloud.
From that perspective, it's still in full swing, and we want to reemphasize what we did, but what we said at Sapphire that there is still a huge remain to do. Very roughly right now, if you look at the ERP-related maintenance revenues, we disclose maintenance revenues every quarter. It was about $11 billion end of last year. From a run rate, it's coming down very slightly every quarter now. There is about a $10 billion-ish run rate related to ERP. That comprises S/4HANA, but also the older versions of the ERP system. That is only converted in a sense that 1/3 of the customers who are still paying maintenance are also starting to pay cloud. There's still 2/3 remain to do. That's one cut to say, OK, who's already on the cloud? Where they jump off, it can be different.
They can either jump off ECC or can jump off S/4HANA. The other cut is how much is on S/4HANA, and that, as I mentioned, is the majority of our customers. That one is kind of such a given, I would almost say, that we didn't spend much time on it because we thought it's well understood that the themes you mentioned are more incremental stages of the rocket, so to speak, because we heard that some people feel we are kind of a one-trick wonder on that, and there's not enough opportunities beyond that conversion period. While the runway is still very long, let's not forget that of that kind of $10 billion-ish ERP maintenance base, probably the ECC and all the related will run pretty much out by the end of this decade because we are running out of maintenance.
Very few customers will be able to afford or want to afford paying extended maintenance for that. Again, we did not push out the maintenance window here. That was purely so nobody on-prem will be able to get maintenance from SAP on ECC. This is purely a kind of refactoring of ECC on HANA in the cloud, and we can only do that in the cloud to give a little bit of more time for those who have a long transformation window. That story is fully there. Now, what we said at Sapphire is that on top of that, there is beyond the 2x- 3x some cross and upsell. In the first signing of the deals, there is the pure kind of narrow ERP conversion. There tends to be some cross and upsell already in that first signing, and that gets you to this 2x- 3x.
Over time, of course, we want to sell more to customers, and they tend to grow. Some of the metrics, which by far are not all seat-related, are growing at the customer. There are also some top line-related metrics, and if our customers grow, which the good ones tend to do, we have that upsell opportunity to then put some significant upside on top. We were talking of up to 5x, which, by the way, if you dial it back, is not so difficult to demonstrate. We also gave at Sapphire some samples, so to speak, from the kind of cohort 2022 revenues. If you look at that kind of RISE cohort and ERP cohort, and then bring it down to 2024, last fiscal year, we saw a very roughly 50% more revenues from that specific cohort. That's similar for other vintages of RISE.
Please don't misinterpret our story that RISE is out of steam. It's in full swing and continuing. Now, the three things that are now keeping us busy to, as I said, light the next stage of the rocket is for the BDC . It's really a great product. I mean, you know we have that huge kind of five-digit business warehouse customers where we didn't have a strong offering. Now, with the partnership with Databricks, we have been able to really create a product which we believe is super and compelling. It has had a flying start also in terms of customers signing up for it. It's moving pretty much according to plan. We're very pleased with that. It's all about execution. It's not so much about the demand side.
If you look at the increment of revenues from a market point of view, this is a very sizable market. It's probably kind of 1/2 the addressable market we had prior to that launch on top. We can cautiously say that it should be good of a $1 billion+ revenue opportunity. Now, when exactly, it's hard to predict because it's a new product. No matter whether it's year three, four, or five of a planning horizon, it should come to that type of level of revenue. The second is the other big lever on the public cloud, which is now starting to become more meaningful. It's not hugely moving the needle yet, but in a similar kind of time frame, we see it also flip to that type of order of magnitude.
Last but not least, on AI, I know there is an eternal debate about will AI eat into the lunch of the software guys or not. From our perspective, it's an incremental opportunity. The TAM estimates from outside research like Gartner and IDC are humongous. If we cut it down to the ERP markets and those markets we are really tackling in terms of application focus, it's a kind of mid to high triple-digit million software opportunity. We want to really take some our fair pound of flesh out of that because we think we have a pole position with the data, high-fidelity data, the contextual richness of the data, the process intimacy we have with the customer.
We are really sitting at that nexus where we see what the customer needs, where we have now a very powerful data engineering platform to put AI into intelligent apps and sell that in an easy way to customers. From my perspective, that whole discussion which says software is now easier to do for our customers because they can use cursor and other things, yes, that's hopefully true. It kind of makes the assumption that we at SAP are stupid and cannot do the same. I mean, of course, we have to do the same transformation on our side to make our 30,000 developers more productive. Of course, we are ambitious to say, why should we not do that more successfully than our customers?
We can do that at scale, at much larger scale than the lion's share of our customers who tend to have a much smaller number of developers to do that.
Let's just pick up on a couple of those points. Let's do the BDC first. I think if I look back at the past 10 years, it's probably fair to acknowledge that SAP has had a better market share in apps than you've had in data. You've talked about BDC as being one of the fastest growing product launches that you've done. Why do you think that customers are embracing data solutions from SAP now in a way that they were perhaps reluctant to in the past?
I think the very simple answer is that on the business warehouse, which was kind of the equivalent old product of SAP, there was not much investment, frankly. You can now argue, was it a good idea or a bad idea? That's water under the bridge. What we see today is that a lot of customers have already chosen certain data engineering platforms. We perceived, after very thorough due diligence, Databricks to be the prime partner for us to take the first step in integrating our BW with what they do into a very easy, accessible data management platform, which most importantly keeps our SAP data in its original context and therefore ensures the integrity and the non-duplication, what we call a zero copy sharing of that data. We federate SAP data with other data.
We have, on the one hand, people who know Databricks already, who have confidence about the product. We have then those customers who have anyhow a huge part of the data gravity in SAP, who previously were extracting, duplicating data with all the kind of costs and risks and shortcomings resulting from that back and forth all the time. They were bogged down by having high cost on integration of these platforms to create these pipes, which, of course, in upgrades on either side are subject to instability and so forth. These restrictions are basically the complexity of that has been now shifted from tens of thousands of customers who have to do that individually to the R&D department of both partners and solving the problem in a plug-and-play way once for all.
That is such a more economic argument that even if we give a lot of value back to the customer by avoiding that waste, we can still share a very good upside for both partners in this deal. That's the secret sauce, I'd say, on this one. It's the data gravity of SAP, the cost of not federating that data in an easy way with other data. On top of that, the same way our customers can now build AI on top of SAP data federated to BDC with other data, we can do the same.
While we have been providing inside apps on BW before, we now can expand that whole logic not only to inside apps and intelligent applications to solve customer problems on SAP data, but if we need some data from third-party sources, I always bring the example of an average manufacturer on weather reports on how hot the weather will be to use how much beer or forecast how much beer we will need. That type of analysis has become much easier. We will also launch a spree of these intelligent apps so that customers have the choice.
They can either code AI in our AI hub on top of the system. If they continue, if they don't want to use BDC , they can egress data and do the stuff they've done, which we think is not the kind of prime way and the most efficient way to do it. Last but not least, they can buy off-the-shelf solutions because the whole idea when it was founded of SAP was that somebody is looking over the shoulder of the customer. It's not a generic platform. It's really learning from the customer. What problems do they solve day in, day out? Then coding that and not coding that for one single company, but doing that for many. To rather have a bespoke system integrator coding approach to really productize that knowledge into these intelligent apps.
Now we are really opening the data we can use with BDC for our intelligent apps to all the data that might be out there because we can easily plug it into our own AI solutions.
In terms of the revenue opportunity, you framed it as rising to $1 billion over three or four years, five years. Is that the same 2x-3x standard uplift to begin with, with 4x-5x as you layer on richer applications over time, is that the same model?
I mean, honestly, the previous revenues on these types of opportunities were not so huge at SAP. Statistically, the smaller numbers get, the less reliable these kind of X factors are. I think there's also new opportunities where we can reach customers that have not been SAP customers. It is very blurry between what is really conversion and what is net new, I'd say, because somebody could be a customer previously on BW, but they would not have kind of transitioned somewhere else, or they could not be on BW, but they had some Databricks. We go brownfield, so to speak, with that solution. It is a different logic, frankly.
I would rather, for triangulation purposes, say, look, I mean, if you add a kind of, depending on what you look at, 50% more addressable market to our previous market, and then we say over a certain RAM period, which is simply operationalizing that and making it up and running, we grab a $1 billion+ revenues. That doesn't sound very ambitious. It is kind of the crumbs, the breadcrumbs of that huge business that we are tackling. Should there be a very broad success, which we are obviously targeting, that should only be the start of a very good story. This is why we are really excited. I would say that probably in terms of steepness of the ramp, it's indeed probably as steep as AI. By the way, how do you really distinguish pure AI from that platform thing that's also a tool for AI? It's getting very blurry.
Also, if you look at the market research and IDC and Gartner, the further you move out, the more they have a hard time kind of distinguishing what is now embedded in applications, what is kind of an on-top credit sold, what are data engineering platforms that are kind of the picks and shuffles of AI. This is the beauty of it. It is clear with some common sense that given the magnitude of the opportunity, unless we completely fail, these are 1 billion + opportunities, each of them. That gives us the confidence about sustaining growth for longer. You know that we did commit to accelerating our total group revenues, consolidated revenues, 2026 and 2027 versus the prior year, i.e. 2025 versus 2026 comparing 2025 and then 2027 comparing to 2026. Now, the big question is, of course, how high can you go and how sustainable is that?
If you then have a lot of investors thinking about kind of putting that whole growth down to a 3% perpetuity growth rate, we want to say there's some other trends that are of bigger magnitude, which are going against that. The other closing check I always bring is the kind of market we are targeting in cloud is already today running in the more high teens, 17%, 18%, depending on what external researchers you look into. Now, why should there be any attrition on that if you cannibalize, so to speak, labor costs and transform that into IT spending? If you do that job of moving to AI, if you're part of the winners, you should get your pound of flesh, as I said before, on that. That is why we really want to make the point that we are not one-trick wonder.
We have a broad set of opportunities where we feel we are actually very well positioned to benefit from that wave going forward. Sorry, I can't hear you, Charlie. I think you're--
Let's just touch on AI quickly. I think one of the comments or one of the messages that struck me from Sapphire was the observation that in an AI-driven world, the value of the software application layer is going to trend down. The value of the data layer is going to trend up. That almost seems to put fuel on the fire of the investor debate around the potential risks and opportunities of AI. Where do you think there are credible risks to your business coming from AI? Where do you think investors have got the narrative completely wrong?
Yeah, I think where we are best protected is obviously where the quality and the context of the data is a differentiator. That's the strength and that's the importance of BDC in that whole game, is that we have the blessing of having extremely valuable context-rich data, and we can make it easily accessible at low cost and, as I said, federated, leveraging BDC . From that perspective, we think the kind of ingredients for being successful in AI are all there. I think it's much tougher for generic platform providers to have exactly that. I think this is also why we have some leveraging discussions about how to connect systems here because they love to have that. Obviously, we try to benefit and try to extract some benefit for our customers, as I tried to explain before from that.
The other argument is really, why would anyone assume the whole SAP story is individual customers or system integrators competing on a case-by-case basis with some off-the-shelf product of SAP. Yeah. Why do you see SAP doing that kind of digital transformation to leverage AI in coding? Why do you see that should work worse than at our customers, diminishing that competitive advantage of economies of scale? I have my question marks about that. Let's not forget that AI is not only in the benefit for the customer. AI is also helping us in the transformation journey. Transformation journeys are resource-intensive and costly. Oftentimes, the biggest pain of our customers is the kind of J curve they need to run through for the transformation before they can reap and harvest the benefits of AI. Even that one, we can massively accelerate.
Not only do we have the right tools for process mining to really figure out what the customer is actually doing, for enterprise architecture management to really have a complete stocktaking of all the application, including submarine applications in the customer, of having with WalkMe a digital adoption assistant where we can help the customer to take the user by the hand and take them over the hump in that change, which is very tough for many humans to do. We have partners that do all the testing in an automated way, leveraging AI. We are tackling also that cost pool. We have Joule for Consultant, Joule for Developer to make the SAP-specific know-how democratize that and bring the cost down. We are not leveraging AI on a solution. We are leveraging AI on the transformation journey, on the non-recurring cost.
I think there, again, we have that pole position that I would venture to say if you go to most of the SIs and ask them, what's the biggest kind of pool of work you have? Probably SAP is all the way up there. That is kind of the ecosystem with literally millions of consultants and coders that can now benefit from AI tools, which are also, to a large degree, offered and sold by SAP with some proprietary know-how how to do these transformation journeys best and seamlessly integrated with the transformation toolset that is already there. That's the differentiator we have compared to some other more niche and maybe more content-driven AI companies.
I always think there is a difference between the kind of process-oriented AI stuff and the agentic AI and then the content-rich thing where you kind of scrape a lot of data from all kinds of public sources. In our world, reliability, compliance, and so forth has such a high value that that differentiation we think we can preserve. We have a strong position there to make sure we can sustainably benefit from that. If you look at the numbers in terms of what's the cost of these knowledge workers, I mean, I look at SAP itself. 2/3 of our total expenses are knowledge workers, all of which can be significantly enabled with AI to make them more productive and also to enable them to come to faster and better insights. I always talk about the experience of the guy in the shared service center.
Today, they're doing pretty basic work and harvesting the low-hanging fruit. There is high-hanging fruit they have no time to go after. If you bring this high-hanging fruit, like looking back into some old contracts and stuff to verify whether that transaction is really kind of fitting or whether there's some liability claim we can issue and so forth. If you can bring that at the fingertip of the guy in the shared service center, he can even create value. When we talk to customers, and we just last week had a big round of customers, some of the biggest customers we have who always give us advice where we're good and where we have to, where they see improvement potential.
They clearly see that a lot of the AI upside is not only the cost reductions, also to, I always call it, build a ladder to the high-hanging fruit and get some more high-value added, more complicated things done by these frontline workers in shared service centers.
Let's move on and talk about some of the drivers behind some of the key financials. If I go back to the Q2 numbers, there was certainly a nod to some macro uncertainty. As we sit today, do you have any concerns over the magnitude of the pipeline build, or do you feel the uncertainty is only around the timing of contract signatures?
I think clearly the latter. We have seen a pretty steady development in terms of the size of the opportunity, in terms of also really win-lose activity, competitively that's very, very promising, actually, and steady. The real issue is the kind of timing. There is an overproportionate impact on cloud revenues for this year if you have a slippage, say, just illustratively from end of June to end of September. The experience is oftentimes end of June is actually end of September because people then start to prepare for holidays, and then they reconvene on these things in September. You work on these, especially on these white elephant situations, which are sometimes pretty complicated. We have called out two areas, U.S. public sector and then more complex international manufacturing companies with supply chains which are heavily impacted by some tariff discussions.
If you think about the kind of time it takes to activate them, once the kind of contract is signed, it takes a couple of months or so. It means if you do that in June, you still have like very roughly four months of revenues left. If you do it in the end of September, you suddenly have one month of revenues left. That's a factor of 1/4. The important metric for next year is, of course, the CCB growth. While that kind of end of September signing will be in the CCB as early as, or largely as early as end of Q3, and then, of course, be fully embarked in Q4, plus what we do to a large degree in Q4 on the cloud revenues, the impact is bigger. That's the point we wanted to make.
On the CCB, there is no change from what we said at the outset of the year. We said there will be a slight deceleration, including a kind of 1.5% impact when we bring the WalkMe acquisition into the comps of the prior year, which happens actually in Q3. We bought the company in September, I think it was, so in Q3 last year. In the CCB end of September, we are apples to apples. It's including WalkMe on prior year and on this year. That shaves a 1.5% off. Now, we've recently announced the acquisition of SmartRecruiters, which is a small company. That's only very roughly 0.5%. The net is kind of a percentage point or so from this M&A activity. We said that that will be part of that. There's no need to change that.
We still have a consistency between what we see will be the exiting current cloud backlog growth as of today's perspective versus the outlook we gave that in 2026. We want to accelerate revenue growth for the group, total revenue growth for the group. Again, in 2027, do the same. That's the story. To cut a long story short, no change on that front versus what we said on Q2.
I think on the Q2 call, you said something like you were confident of accelerating group growth in 2026 and 2027, even if the CCB decelerated beyond 1.5%.
Yeah.
Should we assume that that's the central assumption, a deceleration that's bigger than 1.5?
No, I mean, look, we have to really wait for Q4. You know, 2/3 of the bookings are done in the second half. In the last weeks of the year, there's a lot of activity. This is a kind of curse that has survived the old license deals. I know in the license business, it was much clearer why you needed that. Now, that psychology that people have a budget flash or they want to do stuff at the end of the year, and our sales force also wants to kind of see what they can do for their target retirements and so forth, that is still there. It's really hard to do forward accounting. I always call it on current cloud backlog growth and say this is exactly what it will be.
Because while cloud revenues are quite predictable in a certain way, because it's, I mean, where we are now in the year, the lion's share of even Q4 is in the books. It's really more about the current cloud backlog, where like last year we had a super strong current cloud backlog that makes the comps also more difficult. We had really a snapback on current cloud backlog growth that was magnificent. It's really the one that is a little bit harder to predict.
Our confidence level is there that we have enough in the tank, so to speak, in terms of pipeline and in terms of extrapolating the conversion experience we currently see, that the current cloud backlog growth, which then can be haircut down to cloud revenue growth by saying, OK, how much transactional dilution there is, and that number we think will be clearly below a percentage point next year, that that kind of logic, we take the current cloud backlog growth end of the year, we haircut it for less than a percentage point of transaction revenues. That should be a first kind of rough estimate of what could happen in 2026. That cloud revenue number that would come from that, we think, is strong enough to just give us that acceleration we continue to believe in.
No change on that guidance that we see an accelerated total revenue top line. Frankly, the mix effect helps us a lot there. I can't, yeah, sorry for repeating, some of you might have heard that before, but I always take this excess spreadsheet exercise where I say, let's take the buckets of revenues of SAP. We start with the kind of lion's share of revenues now being cloud ERP suite. There's an extension suite. Then there is an infrastructure as a service, which is going down. There is the maintenance revenue slowly declining. Then there is the license revenues declining fast. There are services, yeah, which is kind of flattish or slightly up.
If you take the growth rates of 2024 on that revenue base and extrapolate to it right on the constant currency basis, that should give us three years in a row if we theoretically maintain these growth rates, three percentage points of acceleration every year. That is true if you take the 2024 growth rates. It is also true if you take the H1 2025 growth rates. The reasons why we can be so kind of confident about the acceleration growth is really that mixed effect. It gives us enough cushion, so to speak, to say, look, the cloud revenues are always now running far beyond market growth. Right now we have more in the high 20%, so to speak, on that number. They can come down over time and still, we see that acceleration in the top line.
There is a long, still a fat distance between that cloud revenue number and the market growth. While we still are set up to continue to gain market share on cloud for the reasons I mentioned before, I would say there is a kind of closing check. What is the market growth in cloud ERP in the end? That is more in the high teens. We, of course, try to keep that super high cloud revenue growth as high as possible for as long as possible.
I feel like in the last 10 days, investors have really run with your expectations of a modest slowdown in CCB. I'm starting to get some surprisingly bearish feedback. I'm beginning to get people speculate on exit run rates in the 2024s. Can we take those buckets of growth, like the cloud ERP suite accounts for 85% of the cloud? It's been growing over 30%. Can we take that and draw a line under downside scenarios for CCB? Does it mean that arithmetically, scenarios below 25% feel arithmetically unlikely, albeit potentially?
I mean, we said slightly decline and said that 1.5% is from the kind of WalkMe topic. You see, your numbers are kind of consistent with that. Of course, 2024 is already quite significant now because I think we started now, I think it was 28% we jumped off end of last year. It's all constant currencies, by the way. Let's also not mix up the dollar. The dollar has been imploding in the first half of the year. You know that. I'm talking now constant currency on all those metrics. Sometimes I even, you might recall in the Q2, somebody asked me about some nominal decline and said, what's happening with the deceleration? The nominal numbers are currently embellished by the dollar implosion for U.S. competitors in dollars. They're looking worse on European denominated companies like ours.
On the constant currency basis, if you say 24%, that's four percentage points. That's maybe a little bit more than a slight decline, I would say. I want to leave it there.
Yeah, fair enough. Maybe we can just carry on walking down the P&L. I think you've been open that you expect the OpEx to grow at around 80% - 90% of top line growth. For the sake of round numbers, if you're growing at double digits, that implies OpEx growth of at least 8%. One of the most single common asked questions that I get is, why does OpEx growth need to grow at 8%? Like, why can't it be 5% and you give me a big margin beat?
Yeah.
Yeah, can you give us some color on the levers that support that sort of 8% and why we should be anchoring our expectations to that?
Yeah, so it's 80% - 90% of the total revenues. This year, implied in the guidance is a much better fall through because we are still benefiting from the full effect of the restructuring, which was completed end of the year and early in Q1. We are talking now really that 80% - 90% is more of a forward-looking statement for 2026, 2027, 2028, and so forth. For us, there are two main value drivers. One is growth. We discussed it at length already. The fact that you quizzed me a lot on that shows that you're interested in it. If I do a simplistic spreadsheet, discounted cash flow model for the company, and I say if I had the luxury of one percentage point more margin and then the cash conversion with it, or one percentage point top line acceleration, I would pick the latter.
I don't want to underspend. AI is a massive investment undertaking. We want to be not again pretty late to the party like we were in the cloud. The stickiness of SAP has allowed SAP to be very successful. Today, we are actually in line or better in terms of cloud conversion than the overall industry. I just checked it on some outside data that our cloud revenues have just overtaken. They have 50% + now. That's round about what we see in the overall industry. We started quite late. We have been playing catch-up big time on AI. We want to be ahead of the game and really be amongst the front runners, driving solutions into the market and offering them to our customers. That costs money. There is also still a lot of investment in making sure that the platforms speak to each other.
Yes, I do see the most significant opportunities on the selling expenses because we are quite high there. Our selling expenses to sales are not stacking up so favorably against some key benchmarks we consider, especially in light of the fact that we capitalize some of that. By the way, this is why I'm a big fan of free cash flow rather than P&L because that makes everything more comparable. That selling expense topic needs to be activated. We have done a lot of efforts to enable a partner ecosystem also to do the long-tail sales on smaller companies. We cannot afford to send a handful team of SAP experts to a kind of $100,000 ACV customer. That doesn't make sense. There were some things we did in benchmarking on a very granular basis to simply look at what best-in-class SaaS companies are doing.
We had some imbalances there, which are cleaned up by now. We will see the fruit over the coming years to come to bear. There is also a story on the gross margin. I mean, we are pretty pleased with the development of the gross margin. I think it's constant currency, 75.2%, significantly up to the prior year, despite some initiatives that were top-line oriented, like granting transformation incentives. We are very pleased on that. We also highlight that the air is getting a little bit thinner on that level because we are reaching by now also on private cloud pretty healthy gross margins. Yeah, that's the story. The 80% - 90% is really a good estimate. You say maybe you could do better.
I also caution us that if we do these cuts at the detriment of the top line, we do our shareholders and ourselves as a core company a disservice. We are deeply convinced about that. Rest assured that we are kind of chasing every dollar of incremental margin as long as it's not jeopardizing our top line. I think we have a pretty reasonable mix from my perspective between these communicating tubes. In some way you can, of course, yank up the margin by putting the brakes on spending. The question is, how sustainable is your growth? We really want to drive the growth on a sustainable basis. I think we have been super successful in cutting cost more and then creating room for incremental activity. BDC costs money. It's a J curve. This year, it will be a significant burn in the P&L.
We didn't have to touch any guidance we've given prior because we were able to create that upside. There is also for us in that kind of guidance the conviction that sometimes we want to keep that upside to reinvest and accelerate the top line rather than just bringing it down to the margin. This is a constant optimization between margin and growth.
Let's move on and do the cash flow. I think you've been pretty transparent with people that there's a pretty arithmetic framework you can think about. You just tax operating profit and add back the share-based comp adjustment.
Sure.
On the same token, you've been talking about some headwinds from FX, you've been talking about tax, you've been talking about migration credits. I guess the question that everyone's asking me is, is 2026 going to be a year that feels consistent with that standardized framework, or are you flagging up these caveats because it's already evident that 2026 is going to be a sort of below-trend year?
Yeah. I mean, that's indeed the formula. I really like this formula. I always say if we meet in five years down the road and we look at the cumulative operating profit non-IFRS, and we slam the kind of tax rate on that, then we add back the roundabout $1 billion delta we see between what we accrue in the P&L for equity-based compensation and what is really cashed out in cash settled equity-based compensation. That should be what we hit in terms of cash conversion. We also were vocal about 2025 benefiting a little bit on two fronts. I mean, that we still had some very favorable tax position that we in H1 had a very low cash tax rate versus the P&L tax rate.
On top of that, there was, on the other hand, a very massive $800 million-ish or will be an $800 million-ish and some still to come in cash out for restructuring. We also said that we will absorb some more restructuring this year because we are going to do a more kind of continuous improvement, adjusting at a rate of 1% - 2% of our workforce. That's another example where we kind of created some outperformance on the margin that we reinvest. We said we are not even pushing that down to the adjustments because we will probably do that very regularly going forward. In the context of AI and the transformation to leverage AI, we need to take some positions out and then add others and really transform our workforce in a very targeted, systematic way. These are the puts and takes.
This year, the reason why it's still a reasonable cash conversion despite $800 million of restructuring is that we have that kind of positive impact on taxes and some other topics. Next year, you mentioned the transformation credit. That is more of a headwind because the way it works is that when you give that transformation credit, it is actually spread over the life of the deal. Should the customer theoretically call off these credits along the life of the deal, it would be no impact on cash conversion because P&L and cash would be tied with the same formula we discussed before. The thing is that we actually want them to use it upfront to accelerate the transformation. There's a phasing topic. I don't want to declare a defeat on kind of sticking to that trend line yet.
Bear with us really for the planning, the granular planning, because we have also a lot of nice ideas what we can do on cash conversion. I would always say, please don't be surprised if there's a noise of a couple hundred minus- plus every year on that kind of trend line. I don't want to be kind of crucified for that because that's the normal way of life that there are because of these phasing topics. One example, and probably the biggest example, is how the phasing of transformation credits can create ups and downs. While it's neutral over a longer period of time, over the full life of the contract, it creates downside more in the early innings and upside later on. That's the kind of wiggling room I would love to keep and just warn you that you shouldn't extrapolate from every single year.
You could also say, why don't I take the $8 billion Dominik has guided for this year, add the $800 million of restructuring, and off we go, $8.8 billion, and then we grow that whole thing with the margin. Yeah, and that will also be wrong. That's the caveat I want to bring. The trend line is what you summarized nicely. There will be some noise and kind of a couple hundred million either way. By the way, if it's either way, it means in the year-to-year comparison, if you have $200 million more in one year and $200 million less in the other, it's a $400 million delta. Nobody should be surprised at these types of volatilities, given the balance sheet we run and the huge payments we pay, but we also receive. That's, I would say, noise around the trend line.
Can you just talk? I'm going to try and squeeze two more questions in, one on migration credits, one on sovereign cloud. Migration credits, like one of the ways of easing the macro uncertainty and giving customers encouragement to sign is to offer migration credits. Is the envelope fixed in size for these migration credits, and there's a limit to how generous you're going to be, or is there mission creep?
OK.
Is there a chance that you get ever more generous with these migration credits in order to support the CCB?
Yeah. I mean, I mentioned that we would constantly look at the balancing between margin and profitability, i.e., margin and the growth. If we can, I mean, the paradox or dilemma we currently face is that, yes, we do make more money when we bring the customer to the cloud. They know that. They know that for us, it's attractive. It's not only attractive for them, but for us. In a negotiation about moving with a RISE with SAP deal as an example, they, of course, try to extract their pound of flesh and say, OK, what's in for me? Can you help me? Because for me, it's extra cost, and you're making more money, and that's unfair. We accommodate that. Of course, I have in our budget, there's a budget, and I don't want to spend more than that.
If I think about next year, I'm not thinking about crazy things, but I don't want to be boxed into that in a completely static way either. If we have the luxury that we throw off more cash because we are doing a great job on working capital and the margin is coming along nicer, why not accelerate top line more and drive that? It's kind of discretionary. The good news is what you should not forget is that that transformation credit has an economic backdrop, which is that the customer needs to invest in transformation to then come to the benefit. Once the customer has signed that deal, it's actually the other way around. They would have a penalty to switch. The bargaining economics at renewal change.
This is, by the way, why the customers are really so much pushing for extremely long contract durations because they are nervous about this. I think while we have a reputation to be more gentle than many of our competitors in terms of price increases and so forth, they also want to have some confidence on planability and so forth. I just want to highlight that the beauty of the transformation credit versus a discount is that you have then some margin upside for the renewal. Once you have digested that, there is no bargaining chip for the customer to say, OK, give me some transformation credit because I will tell them, why should I pay you the move to a competitor? No, you don't churn that. You keep it.
I know that it's, I compare it a little bit to the discussion when we had been shipping from kind of airlines, aircraft to airlines, and you had a low-cost carrier and they're one aircraft only. Going to a competitor, it's becoming sometimes pretty expensive. I don't want to overemphasize that because, of course, our story should be a story of a big carrot for the customer and not of a stick. We really want to deserve our right. There is a little bit of a kind of ethical or not ethical kind of bargaining dilemma around the fact that for customers to go to the cloud, there's a certain upfront investment. They have also an NPV logic. Sometimes they have restricted budgets.
We make more money in the cloud, and we want to have them in the cloud as fast as possible to really make them enjoy the benefits of the product and the AI as soon as possible. Of course, there is a risk that rather than consuming off the shelf, if they come too late, they might be tempted to try something else. Once they tried something else, they might go a different way. We do have a very rational incentive to be aggressive on that front.
Lastly, just on sovereign cloud, you just announced a big $20 billion investment. We're seeing some of your loose peers being richly rewarded for infrastructure-driven growth. It's amazing how things changed. A few years ago, that was a kind of dirty word, I guess, infrastructure-driven growth. Is there any structural change in your approach to a CapEx investment and data center positioning from SAP?
Yeah, look, I'm personally a fundamentalist on finance. What I care about is what excess return can I make over cost of capital and what investment. The application layer and the platform layer give us tremendous opportunities with quite moderate capital layouts to grow the company aggressively. I'm really pleased that compared to any of our key competitors, we have very fast growth in the cloud. That's super value creative because it's kind of capital-free almost. Actually, it is actually capital-free. I would say the employees with the equity settled stock-based comp, it's actually the shareholders who have to pay for that, and the employees are benefiting from, and the employees are giving us some working capital, so to speak, because they are not getting cash from us for that.
I mentioned with the formula we discussed in cash conversion, we are basically saying CapEx equals depreciation, and working capital is kind of a neutral thing for us. I'll also be a fundamentalist on the infrastructure. I think back on what I learned in business school, which is if I can deploy that CapEx here, and I can make more than cost of capital, I should do that because then I will give some kind of value accretion, NPV, net present value to investors. That's the key question where we still have to be convinced that there is a sustainable opportunity to reap excess returns against the competitive environment, which we have with three large U.S.
hyperscalers who do a magnificent job in terms of productivity, with OCI now jumping in, with Ali Cloud pushing into other geographies now like Europe, with some large enterprises in Germany also trying to get government money and dole that out. We also want to tell customers, look, if you feel a need that you absolutely have to be in a sovereign infrastructure, because otherwise you would not come to the cloud, given what I described before that we make more money in the cloud, of course, I would be willing to do even an NPV neutral thing on infrastructure because otherwise I wouldn't have these guys at all. Now what we did for that number, we basically added up over a 10-year time frame, all the OpEx, all the CapEx. We're going to incur all the R&D on projects which partially already exist, like Delos.
That gave us that very rough order of magnitude. It's a long-term journey. We just wanted to make the point, look, guys, if the push comes to shove and infrastructure is the name of the game and everybody loves it, we will do that, but only if we have a high degree of confidence of earning excess returns. One reason to earn excess returns is if there is no competitive solution because a customer, like take the German Federal Employment Agency, says, we want to have that in a German sovereign cloud. There is nobody else except for SAP who can deliver that. Do not forget that the infrastructure as sovereign is not helping you because if all the applications are not sovereign, people will be able to steal your data left, right, and center.
We are the only ones who can provide in certain jurisdictions a sovereign stack from the infrastructure all the way to the application layer. See it in that integrated offering to say, dear customers, we will not leave you behind. If you insist you want to have a kind of pristine German or European stack, SAP is the way to go because nobody else has both the application and the PaaS and the infrastructure. We will also invest in infrastructure, yes. It would not alter the formulas we have discussed because still there would be a huge share of OpEx and the CapEx I think we can digest within the envelope I have been describing.
Perfect. Good. SAP earning excess returns feels like a good note to end this on. We've overrun the 45 minutes, but thank you so much for your time.
No, always a pleasure. Thanks for the opportunity to discuss. Have a great day. Bye-bye.
Thanks, everyone.