Good afternoon, and welcome to Deutsche Telekom's Third Quarter 2025 conference call. With me today are our CEO, Tim Höttges, and our CFO, Christian Illek. As usual, Tim will first go through his highlights for the year to date, followed by Christian, who will talk about the quarterly performance and our group financials in more detail, and then we have time for Q&A. Before I hand over to Tim, please pay attention to our usual disclaimer, which you find in the presentation, and please also note that this conference will be recorded and uploaded to the internet. Now it's my pleasure to hand over to Tim.
Thank you, Hannes, and welcome to our results call for the first nine months. Amidst various headwinds, we continue to deliver consistent, reliable growth. As usual, I will start with the year-to-date view for the group before Christian will dive into the details of the quarter. In the first nine months of 2025, we delivered 3.7% organic service revenue growth, 4.4% organic EBITDA AL, 6.8% growth in free cash flow, and 9.5% growth in adjusted earnings per share. With these results, we remain on track for the midterm targets of last year's capital market day. Also today, we raise our guidance to reflect T-Mobile's guidance increase. Despite a weaker-than-usual quarter in Germany, we talked about that last time, we keep our full-year DT ex US guidance unchanged thanks to the good developments in other areas.
We made progress with our strategic agenda in the U.S. with our successful acquisitions, in Germany with a record fiber build, new collaborations, and Europe's first industrial AI cloud. Across the Atlantic, with significant progress in AI-driven digitization, and our disciplined financial execution was recognized by Moody's with a credit rating upgrade to A3. Last but not least, the Board of Management is proposing a dividend increase of 11% to EUR 1 per share for 2025. In addition, we plan to buy back EUR 2 billion worth of our own shares in 2026, together amounting to a shareholder return of nearly EUR 7 billion. As you can see on the next page, all our segments are contributing to our EBITDA growth. T-Systems leads the table with 11.7% year-to-date EBITDA growth. DT ex US grew by 2.9% in the first nine months.
Moving on to our networks, where we continue to extend our leadership. In the last 12 months, we passed 3.6 million additional European homes with FTTH. We now nearly reach 23 million homes, of which nearly 12 million coming from Germany. In the U.S., our joint ventures are delivering as expected, and we now have 934,000 fiber customers. Our mobile networks are leading across the footprint, and we are confident to maintain this leadership in all our markets. Let me now dive a bit deeper into a German fiber plan. We made some encouraging progress this year, and I know this is a big question mark for you, what we're doing here and how are we reacting on the developments. First, we passed a record number of homes year-to-date, plus 70%. We connected more homes than ever before, plus 9%.
We achieved this with lower total CapEx, resulting from fiber CapEx savings, minus 9%. We achieved this savings through a variety of measures: AI-powered digitization, I talked about that, process industrialization, more shallow digging, and better purchasing. This slide here is to illustrate that our fiber build has become much more efficient. This is a very important building stone for our strategy going forward. Another important building stone going forward is the tax-benefited grant, which the German government developed over their accelerated depreciation model. We plan to reinvest this benefit into higher CapEx and thereby step up our fiber build-out without any changes in DT ex US free cash flow outlook, as stated at last year's capital markets day. What do I mean by stepping up? First, we maintain our 2.5 million homes passed run rate. That is very important.
We will increase the share of rural homes and the SDUs in the mix. Further, we will accelerate homes connected with regards to the MDUs. We are changing the way how we are building out fiber in our German footprint according to the current developments which we are seeing and the adoption rates of these fibers. With our new fiber strategy, we plan to strengthen our German broadband performance in the medium to longer term, both in terms of value and in connected volumes. The efficiency improvements we have already seen and the tax relief granted us mean that we can deliver a more effective fiber build, increase our fiber CapEx in the coming years, while, and this is very important, confirming our stated free cash flow outlook of EUR 3.6 billion in 2025 and EUR 3.7-3.9 billion in 2027.
At last year's capital markets day, we talked a lot about how AI is accelerating our digital transformation. Throughout the year, we have shown a lot of examples of our AI initiatives. On page eight, you can see that we have made further progress on all initiatives this quarter. We are seeing multiple strong use cases delivering tangible results, and this across the whole value chain of our companies in all our markets. I'm just coming back from a five-day trip to Israel, where I have worked with our partners, with our ecosystem down there, how we can integrate their initiatives into our further digitization efforts. I can tell you the agent model is offering us big-time new opportunities, which we haven't considered yet. At the capital markets day DT ex US, we estimated the financial benefits of around EUR 800 million in cost savings by 2027.
Based on the progress already made, we are very confident in delivering this target and even see more potential, more upside here. Last week, we launched Europe's first industrial AI cloud together with NVIDIA, with a combined investment of EUR 1 billion. This is Europe's largest AI factory to date, by the way, opening up in the first quarter next year already. We also remain in the running for one of Europe's planned AI gigafactories, and we should talk about that later in the Q&A. Our customer growth continues on both sides of the Atlantic. Our mobile customer growth remained very strong with another record quarter in the U.S.. In broadband, we had a steady performance in Europe, while we suffered another small customer loss in Germany. Moving on to ESG.
Despite rising data usage outside of the U.S., we were able to slightly lower our energy consumption in the first nine months. Our ESG commitment has been rewarded with various awards, such as the NetFed Sustainability Award and the Award for Corporate Engagement for our initiative against online hate. Let's now move on to our guidance update on the next page. Our guidance remains based on last year's average of a foreign exchange ratio of 1.08, and as always, in the sum of the guidance DT ex US and for T-Mobile U.S., adjusted by the U.S. GAAP IFRS bridge. T-Mobile once more raised its guidance for customer and financial growth, and we are passing this on in the group guidance today. T-Mobile raised its 2025 EBITDA guidance by $300 million at the midpoint, and its free cash flow guidance by $100 million at the midpoint.
T-Mobile's new guidance now also includes the expected contribution from the recently completed acquisition of Metronet and U.S. Cellular. Our 2025 DT ex US EBITDA guidance remains unchanged at $15 billion EBITDA and $3.6 billion free cash flow. With that, let me now hand it over to Christian for a deeper dive into the third quarter.
Thanks, Tim, and hello everyone. As usual, I'm going to provide you with an overview on the segment performance in the third quarter and then present some selected group financials. As usual, let's start with the U.S.. We have reported their numbers already on October 23, and you know that those numbers include a two-month contribution from U.S. Cellular. Still, the numbers I think are really impressive. You've seen, according to U.S. GAAP, a service revenue growth of 9.1%. If we take a look at the postpaid service revenues, they even grow at close to 12%, and this is coming from volume as well as ARPA growth, and the core EBITDA grew at 5.6%. Where is this all coming from? Take a look at the growth numbers on customers, and I think they're record-breaking.
The postpaid net additions were 2.3 million, which was significantly higher than the consensus, which was 1.6. The postpaid account growth was almost 400,000, which is the highest number ever. The postpaid phone net adds were 1 million, which was also 150,000 higher than consensus and the best quarter since 2014. Finally, the broadband net adds, which also include fiber, grew at 560,000. I think what you've seen is a stunning customer result in the third quarter. The churn rate actually grew slightly vis-à-vis the previous years, but bear in mind it was the lowest one among the three M&Os which we have in the U.S., and the ARPA is now expected to grow at 2% vis-à-vis 1.5% in the previous quarter. Based on these very strong results, T-Mobile once again raised its net add customer guidance.
They now intend to get to 7.2-7.4 million postpaid net adds, which is up more than 1 million at the midpoint relative to the last guidance, and the phone net add guidance has been increased to 3.3, which is also up by roughly 300,000 at the midpoint. A very, very strong third quarter, and now we're getting to Germany, a segment where we have to report some different figures, I would say. If you take a look at the Q3 financials, you see there we're impacted by prior year comps, but also by our cost facing, and I indicated this already in the Q2 call that we have the double WEMI coming from the wage increase, which obviously hit the EBITDA growth.
If you take a look at the headline growth in Germany, it's actually declined by 1.8%, and there are two factors which are responsible for this. In 2024, you had the one-off revenues from the European Championships TV rights, and secondly, we have another, I would say, lower revenue contribution from third-party equipment sales, which are all low margin. If we're taking a look at the EBITDA growth, which is slightly above zero, you can say it's stable and it's the lowest since many, many years. This is coming from a very low contribution from service revenue, where we're getting into, and the double whammy from the wage cost headwinds. You know that we increased the first wage increase in October 2024, and then we had the EUR 190 one-off payment starting from August, and they both collapsed together in the third quarter, impacting the EBITDA quite significantly.
Both of those quarters, the service revenues, especially the comp factor, but also the wage increases will roll over in the fourth quarter. For the fourth quarter, we expect an EBITDA growth of above 2%, of at least 2%, above 2%, 2-2.5%. It is definitely above 2% to reiterate this for this community. Also, if we basically fast forward into 2026, bear in mind that we are not only facing headwinds from the wage cost increases, which are rolling over, we are also facing headwinds from the higher energy cost in 2025. Both of them will obviously roll over in the next year and will help to support a better EBITDA result than the 2025 EBITDA result. On top, we have launched an additional cost savings program, which is targeting only non-personnel cost, which will also support the EBITDA growth in Germany in the year 2026.
Let's move over to service revenues, and you see it on the right-hand side. The service revenue growth in the third quarter 2025 was really low at 0.4%. It was all driven or largely driven by the negative contribution from the fixed line business, and there are two major explanations for this. One is the comp from Q3 2024, which was largely driven by B2B business. You see that we have a very strong growth in the third quarter of the previous years. That obviously has an impact on the year-on-year growth for this year. The second one is we have to actually acknowledge that the German economy is weak right now, and we're seeing this in the number of insolvencies in the German market. I think this is also something which is impacting us in a negative way.
Secondly, the service revenue is impacted by the lower volume trends, which we're seeing in broadband, but also in wholesale, and this basically collapses to that negative growth in fixed line of negative 0.3% in this previous quarter in Q3. For Q4, we have a pretty high confidence that we're going to have a meaningful trend improvement, and for mobile service revenues, we absolutely remain consistent with the guidance which we have given, which is 2%-2.5%. Let's move over to the broadband revenues, and you see that the retail broadband revenues obviously have come down. Also, the wholesale revenues have come down, and this is pretty much driven by negative volume impacts. What you see is that the ARPA up strategy, so the more-for-more strategy is working, the ARPA growth in the consumer space grew by 3.6% in the previous quarter.
Upselling is working and has to contribute a large part of the broadband growth. Despite the volume pressure, you see us discipline on pricing. We maintain our promotional period at three months. You know that we have taken this down since April. It was coming from six months. We have increased front book prices for single play between EUR 2.5-EUR 3 a month. We also, in October, have increased the broadband front book prices by EUR 1. We are playing the value game. We are playing the long game. We are not fighting for every incremental volume, and we hope that the market will stabilize in that sense. On wholesale revenues, what you can see is there is a significant step down relative to the previous quarters.
We are basically flat as we guided it to be at the capital markets day, but we do not expect any significant deterioration in the upcoming quarter. Let's move over to the fixed line KPIs, and you see that the monetization works. Upper right-hand side, 54% of our customer base are now customers with at least 100 MB per second, and you see this continuous trend happening since quite a bit. What you also see is that we're still not mitigating the negative broadband net adds. We're remaining at that 20%-25% trend, and to be honest, we don't expect a significant improvement short term, despite the fact that we're working on quite a significant amount of measures, which is digital retention management, extension of our distribution, and more localized pricing.
What is important is obviously since the market is slowing down and since we are facing ongoing pressures from the overbuilder that we have readjusted our build-out strategy on fiber. We're not only spending more on fiber. We're shifting the mix towards more rural areas and SDUs because we know that the connection rate is coming in much faster than it is with the MDUs, and we're stepping up the initial connections of MDUs. Even if we don't have a customer, at least a home is prepared, and we have a connection there so that we can actually act on customer demand fast. You see that, and I think that is kind of, for me, the bright spot in the quarterly numbers and the customer numbers that the fiber strategy is working off. We added another 155,000 fiber customers.
This is the best quarter which we ever had. That remains the key focus area, and you know that we have given a commitment for 2027 to add a million fiber customers over the course of the full year. Let's move over to the mobile commercials, and you see we're back, as we indicated in the last call. We know that we have elevated competition since quite a bit, but our commercials actually remain strong. You know that in the second quarter, we lost a very large customer. We said we will return back to the, let's say, usual run rate, which is somewhere in between 250-330, and you see us now coming in at 314, and this is also driven by a lower churn rate, which has been reduced from 0.9% to 0.8%. That is pretty much it on Germany.
Not a very good result on Q3, but a much better outlook for the fourth quarter. Let's move over to Europe. Europe has provided another excellent quarter, 31 quarters with consistent EBITDA growth. If we just take a look on an organic perspective, which is the lower part of the chart, you see that overall revenues grew by 2.2%, service revenues by 3.3%, and EBITDA has grown by 4.6%. You see that there is a sequential slowdown in the EBITDA growth, and this is obviously coming from the progressive rollover of inflation-driven price increases in some of the European markets. Moving over to the commercials in Europe, I think what you see is overall very good results across all categories. I think one has to highlight that the mobile net add is actually being impacted by negative cleanup ahead of the Romanian disposal of 60,000.
If you would basically include this, there would be close to 190,000 of customer growth in the mobile space, and you see also steady and strong performance in broadband and TV and in fixed mobile convergence. Moving over to T-Systems, and T-Systems continues to be on a positive track. I'm really happy with the performance of T-Systems. Last 12 months order book is up close to 4%. The organic revenue is up by 3%. It's in the middle column. If you take a look where it's coming from, T-Systems is actually benefiting from the AI-driven digital solutions, and we're talking quite a bit about AI, but also from the sovereign cloud services which we're providing that supported this growth, and they had a very stunning organic revenue growth of close to 23% in the previous quarter in Q3.
Gets us to year-to-date EBITDA growth of close to 12%, and this is coming not only from top-line growth, but also from cost efficiencies. Bear in mind, this is a project-driven business, so there's quite a bit of volatility in there, but I'm completely confident that they're going to make and beat their commitments they have given at the capital markets day. That basically concludes my operational review, and we're moving over to the reported financials. I think first and foremost, we have to acknowledge that we're negatively impacted by a weaker dollar. Last year, the dollar was at $1.10. This year, it's at $1.17, so it's a depreciation of $0.07. That impacts the reported figures.
It's partly mitigated by the contribution from T-Mobile's M&A activities, and you see also that there is some phasing, but bottom line, I don't want to go through all the details here. I would say we're broadly on track with all of the targets, and you're going to see us confirming the CMD targets later on as well. Moving over to the usual Q- over- Q annual comparison of free cash flow and net profit, and I keep it short. You see there's a reduction of 9% in the free cash flow. This is very much driven by two factors.
One is the weaker dollar, which impacts us with negative $500 million and a stronger CapEx volume, and you know that we in the previous quarters have reported kind of CapEx, which was below the average, which you would expect in a given quarter, and there is a catch-up, which you can see here. If you take a look at the year-to-date numbers, they grew close to 7%, so this is very much in line with the increased free cash flow guidance Tim was talking about early on. Same old is true for the adjusted net profit. It grew by 14%, but very much driven by the financial result and despite a headwind from a weaker dollar. Moving to the next page, which is leverage. Overall, the leverage has increased by $5 billion.
Everything was driven and more than driven by M&A activities, which was obviously U.S. Cellular and Metronet, and you see the impact of $8 billion. Nothing of the other contributions to the net debt are a surprise. You see us moving within the corridor, which we indicated we are below 2.75, including leases. Excluding leases, we are 2.23. Sorry for that. I think that also led to the decision of Moody's to basically give us an upgrade in our rating. I can only encourage the other rating agencies to take a closer look at our balance sheet and our financial discipline. Finally, on the last page, the key takeaways, we are confirming our midterm adjusted EPS target, which is around 2.5 by the end of 2027.
Tim was talking about the consistent, reliable growth despite some headwinds, which we are seeing in Germany, but other than that, I think all the other segments are performing well. We have confirmed our targets, which we have given ourselves in the ex-U.S. business, and we have increased the guidance in the U.S. business, and we confirm our midterm CMD guidance. The flywheel works. We are working on expanding our network leadership on both sides of the Atlantic. That drives customer growth, as we have talked about this, and we have a massive initiative running on AI in order to not only drive efficiency, but also top-line growth. I think this is something where you basically should remind us on an every quarterly call. This is kind of what we call a drumbeat when it comes to AI. We are reinvesting, and we were not clear about this in the last quarterly call.
We are reinvesting the German tax relief into CapEx and into adjusted fiber rollout strategies. We remain comfortable with our comfort zone in the leverage, and obviously, I think we have proposed an attractive shareholder remuneration package with an 11% dividend increase to EUR 1 and an up to EUR 2 billion share buyback program for the upcoming year. With that, I hand it over to Hannes.
Okay, and now we can start with the Q&A part and the instructions for that. If you'd like to ask a question via WebEx, please press the raise hand function. If you're required to cancel your question, please press the raise hand again. If you're calling on your telephone, please press star three, unmute by pressing star six, and if you want to cancel your question, press star three once more. I'll announce your name when it's your turn.
As usual, we would be grateful if you could restrict yourself to two questions, and do not forget to mute yourself and unmute yourself when it is your turn. I think we start with Andrew Lee at Goldman Sachs. Andrew.
Good afternoon, everyone. I had two questions. Hi. Two questions. One on capital allocation and one on the U.S.. Just in terms of the capital allocation, you are giving an 11% dividend growth guide for 2026. I do not want to demean that and reloading on the $2 billion buyback. If we look at what that leaves, where that leaves you in terms of your net debt to EBITDA for next year, even if we take out the positive effects on or the reductive effects of FX on net debt, you seem to be leaving yourself with more balance sheet flexibility into 2026 than you did a year ago.
Could you just take us through why is that? What is the strategic flexibility that you need into 2026 that's maybe a greater pull on your balance sheet than last year and where that's coming from? Secondly, on the U.S., clearly that's been, along with Germany, a kind of major source of concern in terms of the sustainability of growth from investors, and that seems to have narrowed down into a major concern around Verizon competitive intensity. The question is just pretty broad and straightforward. What's your take on the degree of change that we've seen in terms of imminent competitive threat in the U.S. market and risk to your to derailing the team's growth story? Thank you.
Let me start with a capital allocation question. Andrew, we actually have a slightly different view.
If the weaker dollar is rolling over, to give you the calculation right now on the leverage ratio, the EBITDA is calculated at EUR 113, and the debt is calculated at EUR 117. If you basically adjust both of them for the same rate, which will happen over time, we are at 2.72. This is one. Second is, look, there's always projects coming up which you don't know. For example, take a look at the gigafactory, which we didn't have on the radar screen, and therefore I want to have some flexibility. Thirdly, I think if you take a look at the shareholder remuneration program, which we have articulated yesterday, it's an 11% dividend increase. It's the highest dividend ever in the history of this company.
The EUR 2 billion share buybacks, if you assume that the EPS is around over the next 24 months at EUR 2.25, EUR 2.30, gets you an 8% yield on that share buyback program. We feel comfortable with the volume, and we do not want to, let's say, be super volatile on this one. We want to be consistent. It is a combination of, first of all, we are not as optimistic on the leverage ratio as your calculation is. Second one, it is prudence. Thirdly, I think it is still an attractive program. Look, we are basically distributing EUR 6.8 billion next year, which is quite a significant number, at least from our point of view.
Just a quick follow-up on that question. The team's buyback is done. Decisions are made on a seemingly quarterly basis. The DT buyback decision is currently on an annual basis.
Can you see a time where that's made more flexibly, i.e., on a half-yearly basis or quarterly, or do you still expect to announce buybacks on an annual basis?
Both have been decided. Both have been decided on an annual basis, but the programs are more flexible in the U.S.. We can course-correct on the program. For example, you know that we stopped the share buyback in the U.S. for quite a bit because we had a leadership change, and therefore we have adjusted the share buyback program now for the remainder of the year. The U.S. has more flexibility because we always have this freeze period or this grace period of 90 days. We have to file 90 days before we actually can execute. We have less flexibility on the ex-U.S. side, but both programs are being decided on an annual basis.
Andrew, I did not want to tell you a fairy tale in investor call, but it's a little bit, let's say, the second question, like the hedgehog and the rabbit. If you look back, the U.S. market has always been very competitive. It is recently very much focusing on device promotions, and that is, let's say, where it's going. We were a share taker in this environment over the years. Now, what we seen this quarter, it was that we were attracting more customers to us, and we had lower churn than before, which is resulting in a much better performance. We did that at the same time with a significant increase on our EBITDA growth with 6%, which is giving us the opportunity to invest into the infrastructure, and our network leadership has improved as well.
Now, it is not only about, let's say, subsidization or the money which you put into the market. It has a lot to do about, let's say, how good you are perceived from a brand, how good your network is improving compared to the others. It is about, let's say, how you enter into the smaller markets or the rural areas, how you're attracting business customers, or even the broadband customer numbers is quite encouraging. We will probably see 1 million next quarter already. This is all, let's say, growth, which is not coming only from price competition, but as well from quality and investments which we have taken before.
Now, coming back to the hedgehog, I think we have heard about, let's say, the new announcement from Dan Schulman, which I know for years about what he's doing, and he has laid out his plans about that Verizon's ambition is to win back shares and moving away reasons to churn and focusing on customer centricity to grow his cash flow and his leadership over the next years. Now, this sounds very reasonable to me, but the hedgehog, he's on a path already. He's already running in another direction. I think they're all well at this point, but I think what Srini and what the team is doing is this way of finding a new digital customer experience, the way of serving customers in a new kind of uncarrier way, this way of surprising customers with new propositions. This is what we are about.
It's not about, let's say, that this is wrong what these guys are announcing, but customers are looking on other things as well and what is new with their brands. I can tell you, I'm very encouraged about, for instance, the efforts with regard to digitizations. It's an outstanding achievement that 80% of the upgrades are done digitally within one year. It shows me even that the skills are within this company to reinvent the way how they're performing. They have cost potentials which they can reinvest. We have a super network leadership which we are strengthening these days. All of this is coming together. We are used to competitive environments. I don't think it's only about money. It has a lot to do with propositions, which we have played out very well in the last years.
We stay, remain focused on a thoughtful balance of commercial and financial growth in the U.S. market.
Okay. Thanks, Andrew. Thanks, Tim and Christian. Move on to Ottavio at Bernstein, please.
Hi. Good afternoon. Thanks for taking the two questions. The first is on the domestic business. You highlighted the good performance on the ARPA. The 3.7% is welcome because that has compensated for your negative net adds. Today, you are somewhat guiding for a continuation of negative net adds, and therefore the ARPA increase will be very crucial going forward. During the call, you attributed the main drivers to the upsellings, not just the price increases. The upselling will be very key for your growth. My question is, how fast you can increase that upselling because it is still running significantly lower than you project for 2027 for around 1 million. You go into rural.
The question is there, how much the increase, because you said that rural would be better take-up rates. Someone would expect that the overall take-up of fiber vis-à-vis the overall base will increase. That should be good for ARPA. Also, you can update on the plans by the German Digital Ministry to weaken the landlord ability to stop in-house fiber rollout. Last quarter, you were very vocal, and I do not know if you see our discussion or any decisions being made because that, I think, will be crucial for you to improve the upselling, therefore, of course, to improve the ARPA trends going forward. The second question, it is going back to the capital allocation strategy, but this time, I would like to do a bit more with numbers. In the CMD, you are talking about a precise number, the EUR 15 billion surplus.
I follow the logic, and at that time, you basically were thinking of unchanged gearing. And Christian, you have been very clear on the fact that 2.64 is misleading because you have to put that and the EBITDA on the same currency. That is fine. If you do, you ended up effectively around the same gearing that you expect to go, 2.75. Therefore, last year, you allocated around EUR 4 billion for additional shares into T-Mobile U.S., and the EUR 2 billion buyback of last year was already included. The additional EUR 2 billion announced today will be against the surplus. Effectively, you used up around EUR 6 billion, roughly, or EUR 6.5 billion of the surplus. My question is, there is still around EUR 8 billion left there.
Now, is the temptation to see T-Mobile U.S. share price relatively low for you to go for more shares in T-Mobile U.S. rather than the buyback considered for last year, or it's still undecided where are you going to do next? Thanks.
Ottavio, complex questions, to be honest. First of all, on the ARPA increase, I think if you basically go back into the previous quarterly reviews, you basically see a linear increase of customers adding faster lines. Obviously, we're working on better monetization, especially when it comes to fiber, because as we're moving into SDUs, we can monetize that fiber much faster. Also, if we connect on the MDUs, that should give a contribution. The second one is related to the price increases which have recently been introduced, right? You don't see any impacts on them right now.
Obviously, we expect a net-net positive effect, which will drive also the revenue. On the volume trends, look, I'm a finance guy. I'm a little bit conservative. Let me put it this way. As long as I do not have line of sight, I do not want to promise you anything other than what you have seen in the previous quarters. Therefore, we do not predict anything as long as we have good evidence that the volume trend is actually moving in the right direction. It is not too far away, to be honest, right? Just let's assume it is 15,000 more. You are getting closer to the 3% right now. I think the 3-4% broadband growth is not being achieved right now. I think on a CAGR basis, we have no indication, or at least we believe this is still an achievable target.
This is the combination on how things are evolving. I don't know whether our signals to the markets, like on single-player and broadband, will be received by others in, let me put it this way, in the right manner. That could help to basically go for a little bit of market repair. We'll see whether it's going to happen. That is kind of the basis we're working on. The biggest lever for us is churn management, right? If we get the churn management on our broadband base just a notch down, that will immediately churn quite a bit on the volume. This is kind of the equation. I can't give you a mathematical equation, but this is the equation on the levers we're working on. Look, let me add update on the plans with German Digital Ministry.
Yes, it is, and I was, by the way, not vocal last week. I was vocal this morning already again in the press. It cannot be that we are paying the bill of building a fiber network not only on level four, but even, sorry, level three, but even level four in the houses, that we pay tons of money for connecting the country into the next generation infrastructure, and that then the landlords are sitting there and asking for a revenue share or asking for an installation fee for the apartments. If Germany really wants to get digitized, they have to support the environment. By the way, the German Digital Minister is already on our side. He had put a paper, Eckpunktepapier, into the discussion, which is clearly enabling and accelerating the build-out in the multi-dwelling units, which is the main part of it.
To be honest, I cannot tell you when they're coming to a decision. I have the feeling that the German government is under a lot of pressure and taking a lot of decisions every single day. Next week, on Tuesday, there is the digitization summit with Chancellor Merz and with Macron and a lot of, let's say, other players. I promise you, I will address this topic there again. I have the feeling that there is a big understanding. The problem here is that Vodafone is trying to defend their position in the houses with their coax, which is nothing else than CAPA. This is not fiber. I think more and more people tend to understand that, that they are not investing into the next generation fiber, that they're just trying to defend their position here.
I make sure that we will, by the way, with the other fiber investors in Germany, fight for these initiatives. Let me make a general comment at the end because Christian laid it out. Guys, since the last quarter, three months, we have worked intensively, intensively on reshaping the way what we are doing with the fiber rollout. We have laid it out in the presentation today that the CapEx per connection has gone down. We expect further reductions being possible. Second, that we are now changing the rollout areas. Thirdly, that we are building more fiber, multi-dwelling units, homes connected, and more homes connected with regards to the SDUs, so in the rural areas and single households, because we see here a higher acceptance rates. Thirdly, to stop that the Altnets are eating our cake.
Fourthly, we have a total new go-to-market with regards to, additionally to the ranges, we have now enabled our sales organization, our retail organization, to meet people at their homes. Fourthly, we have allocated additional people to this one. On top of that, we have a new churn program, as Christian laid it out, and so on and so on. We were very unhappy, and we are unhappy with negative net adds. This is not acceptable. Therefore, I can promise you that there is a big program up and running within the financial commitments which we have given to improve the situation here in Germany.
On the capital allocation, let me try to answer that question at least partly. First of all, we want to keep the flexibility.
If you're taking a look at what I said earlier on, the share buyback program, which we decided, or which we consulted with the Supervisory Board and obviously then decided on later as a board, will give us an 8% return, which is obviously a pretty good return relative to other means. Secondly, what we have not decided yet, we want to keep that flexibility for good reasons on whether we should basically continue with the share buyback program beyond the EUR 2 billion we have just announced or basically put everything into the T-Mobile U.S. shareholding. Look, none of us in the summertime would have estimated that the share is going down to $203, right? None of us. I think this flexibility is prudent to have, and obviously, it has to be taken into account for.
The third one is, if you take a look at the current run rate, we indicated 46% EBITDA growth. We're around 4.4, so we're not at the midpoint. Obviously, that surplus is also coming down. Therefore, we take it, let me put it this way, one year for another and explain why we're doing what we're doing instead of giving you a midterm outlook of what we're planning to do with the surplus.
Thanks, Christian. Thanks, Tim. Next, we move, I think, to Robert Grindle at Deutsche Bank.
Yes, thank you very much. Sorry, no video. It's usual WebEx versus DB issue here. Two questions on the increased attention to fiber in rural areas and stopping the Altnets, eating your cake, as you say, Tim. Are you thinking more greenfield sites here or looking to defend in areas already under threat from the competitor build?
Secondly, you acquired a call option over 10 million TMUS shares owned by SoftBank last month. Is there an ongoing cost to that? Have you thought about buying out their residual stake? Thanks.
Greenfield or overbuilt? It is both, is the answer. There is greenfield. I mean, basically, we are looking at areas which are most likely to be overbuilt, and then we build there. That is kind of a big part of this change in mix. If it involves overbuilt of an existing plant where we feel that we have an attractive interest business case, then it will involve overbuilt.
With regards to the SoftBank question, at the beginning of October, SoftBank granted DT 10 million call options in TMUS that can be exercised at market price until, and now listen, April 2029. This is a very, very long-term option.
You know that we have a very good partnership with these guys, which has worked even without buying them out. There is no read across to our target TMUS stake. What I can tell you, this is more a sign of the partnership, which we have built for a much longer-lasting relationship. That said, as stated at the capital markets day, TMUS stake increased to remain one of the preferred users of any surplus capital alongside M&A and DT level share buybacks. Therefore, there is nothing to say. The only thing is, I think there is no need to make any kind of short-term speculation on activities here.
Great. With that, I think we move to Paul Sidney at Berenberg.
Thank you very much for taking the questions. I had two.
Firstly, we've seen more and more European telcos announcing their AI initiatives, talking about data centers, etc., yourselves, you're partnering with NVIDIA, going live in Q1 next year. I was just wondering, is it possible to put some numbers around this opportunity? I'm not looking for specifics, but just in terms of what the opportunity could be for Deutsche Telekom and perhaps the industry. Secondly, you're one of the last European telcos to report. On our calculation, European service revenue growth has worsened versus Q2. I was just wondering, what do you think yourselves and your peers need to do in Germany and the rest of Europe to maintain healthy service revenue growth and keep it in positive territory? We talk about value over volume, playing a value game, prices are going up, but a lot of this stuff just doesn't really seem to stick.
I just wondered, it'd be great to get your views on what you think the industry needs to do or change. Thank you.
Yeah, okay. Okay. I think, first of all, there is no single answer for each country. I think you have to take each country country by country. Let me start with the largest one. Look, the indication, Paul, which we have given to the market is we would appreciate market repair in broadband, right? We are doing this both on single play, where obviously we do not have a lot of competition, to be honest, but also in broadband. We have to see whether actually the other guys are following this direction, yes or no. We do not know. We cannot influence this.
I think in a slow growth market like the broadband market in Germany, that is the only way that you either upsell or that you have market repair on the overall market. I think as the market growth rates, especially in broadband, are coming down, it's not true for every Eastern European market, by the way. I think I would clearly favor value over volume. In mobile, it's a different answer. You've seen the net adds from our two competitors, -1 and + 157. Our segmentation is working. The conjunction of B2B vis-à-vis B2C and the separation between, let's say, single households, which are predominantly addressed by Congstar and family plans, which are predominantly addressed by the first brand, it's working out just fine. This is why we're growing, where the others are not growing.
Therefore, I think we do not see any necessity right now to basically change that proven model, especially given the fact that we have our Network Optimization Program, NEMO, which gives us the capacity to actually fuel those future demands. I think you have to answer this country by country. To be honest, I would not be in the position to give you an answer on every European country. Overall, I think it is only working if the market, let's say, environment is also reacting in a rational manner. Because what you are seeing in many markets is that GDP growth is much faster than, for example, mobile service revenue growth. I think that should not be, given the importance of that service which we are providing, that should not be the case.
We have to work on, and especially market leaders have to work on repairing the market and actually getting the right value from the service. It's all about value-added service, especially in the B2B space, right? It's, for example, adding security on top. Security is a massive issue across the board, especially for smaller companies because they don't have the capabilities to basically have an own staff which is dealing with that security by adding IoT services on top, where we're seeing quite a significant volume impact here. These are the things where I would say in our core business, I was just talking about mobile and fixed, how you basically put additional and adjacent services on top.
Tim? I'd like to address your AI question.
Look, by the way, the first one is you mentioned that telcos are going into AI on gigabit or in AI initiatives. Yes, that's true. You mentioned NVIDIA as an example, which is data center capabilities. In this case, I would say no because Deutsche Telekom is the trailblazer here in this industry. I do not know where the others are following, but we are the early mover in this environment. There is not a single other telco who had made a commitment, our partnership with NVIDIA, committing 10,000 GPUs being available from first quarter 2026 for the industry here in Europe already. Let me talk about the AI gigabit factory for the first step. I think this is where we are unique.
Maybe Telecom Italia is a little bit comparable here because they have this governmental commitment that all the data is moving into their inference centers, data center infrastructure, but no GPUs so far as I know it. We have now the partnership with NVIDIA where we started with 10,000 GPUs. I think Jensen and NVIDIA selected it very wisely because they are going to the industrial core of Europe, which sits in Germany. They are going to us and with us, with the biggest market access to business customers. We can offer a sovereign solution, network infrastructures, this high 400 gig connectivity coming from us. The data center infrastructure is something which we know already because we are running 186 data centers across the globe. I will go into that in a second again.
The sovereign cloud, which we are offering already, is now almost a decade in the market, so well known to a lot of, let's say, classified services. Our partnership with SAP on the B2P side is enabling the customers with their applications to go into this ecosystem. NVIDIA is providing their latest Blackwell chips into this industrial environment in Germany. I think this is a great opportunity now for us to see how the industry 4.0 is becoming real in automation and digitization here. For us, this is a kind of good learning case for the next step, which is the AI gigafactory. Together with our partner, Brookfield, we have submitted a consortial bid for this Europe's planned AI gigafactories. The size of the location is expected to be around 100,000 GPUs.
We expect any investments here together with them off balance, but nevertheless, the distribution and the go-to-market will be facilitated by our T-Systems arm. Therefore, this is a big opportunity for us even to participate in this new high compute and digital ecosystem. We call it physical AI, or we call it, let's say, participating in this environment of robots and the industry 2.0, however you want to call it. Look, we do that step by step. We do that with strong partners. It is not that we are going alone here in a big risk. This is a big opportunity. Now, you can judge what you get for 10,000 GPUs on the market price today for revenues. That would give you an indication about how much money we earn with that.
We have a deal with NVIDIA that almost 50% is getting invested from them, 50% from us, and we have a revenue share model established so that we are cautious with regard to all the upfront investments here. I think this is a unique proposition which gives us as well credibility for the second step, which is the gigafactory. On top of that, by the way, maincubes, and sometimes we always forget that, maincubes is with over 200 MW of capacity in operations or in development in Frankfurt, in Berlin, and GreenScale, it's another subsidiary of Deutsche Telekom and is with 170 megawatt project in Ireland and a 300 megawatt project in Norway on its way. This is something where I think Deutsche Telekom is building on their infrastructure experience, something new where we have a lot of, let's say, competencies already, scaling it up.
We only scale with commitments from customers. That is the good thing in this industry. It is not that we have to build a mobile network first and then we will see whether we get customers. We will learn on the run. I think, yeah, that is an opportunity for our T-Systems business. Yes, we want to do that as cash cautious and CapEx cautious as well for our business. Nevertheless, we want to, under the frame of building sovereignty for Germany, scale that here in our industrial environment.
Thanks, Tim. To be clear, maincubes and GreenScale are held through DTCP, right? Thanks for the questions. We move on with James at New Street, please.
Yes, thank you, Hannes. Good afternoon, Tim and Christian.
Actually, the first question I'd like to ask is to follow on precisely, Tim, from what you actually just talking about there. I'm excited to learn more about the kind of NVIDIA project, which a financial analyst would like to just go a bit further on the numbers from what you said just now. I think the initial project with NVIDIA, you've said it's about EUR 1 billion, of which maybe now Deutsche Telekom is going to be putting in 50% of that. You've said you could scale up with Brookfield now to 100,000 GPUs. Could we take that as saying that if that's successful, that becomes a EUR 5 billion investment we see from Deutsche Telekom? Would therefore love to just also understand a little bit more about some of the specifics about how you see the return on capital on that project.
Then the second question I had, maybe one for Christian. Christian, in one of your answers earlier, you seem to link the EUR 15 billion of surplus capital that could be used to the EBITDA growth of 4%-6% range. I suppose the question is, if actually EBITDA ends up being at the lower end of that range of 4% growth, what does that imply for the EUR 15 billion of surplus capital? Is there actually a commitment that all that money would be spent somehow by the end of 2027? Thank you.
You want to start it? Fine.
By the way, I'd like to start with the first question. Again, these are two separate projects. The first project is 10,000 GPUs. It's going to be a data center being based in Munich. It is using an existing facility, which we have renovated.
It's three, four floors under the city. It is using 100% renewable energy and cooling from water, which is available. This 10,000 GPUs is something which we have in our planning, in our financial envelope which we have laid out. No additional funding or concerns which you should have with regard to the envelope which we have laid out. This project is now the first step. That is, by the way, 100% on balance because this is a project which we run out of T-Systems. The project number two is the planning and the preparation for the AI gigafactory, which is a European RFQ for six data centers across Europe, where the U.S. is committing to a certain utilization of their public domain data in this environment. In this case, we are planning an off-balance solution.
In this case, we are not planning automatically, let's say, a high ownership on the infrastructure investments because we have said that we are going to take Brookfield as a partner into this ecosystem who is taking, let's say, a significant portion of the investments. We might even consider other partners who are building this infrastructure. It is too early to give you now the financial construction about how that is taking place, but the infrastructure will be built off balance. It will be supported with public sector money or utilization, which is helping that. We are now in the selection of the real estate. We are in the selection of where we are building it. We are in the selection about how this consortium would look like. There is an application which is taking place in January.
There is the decision from the European Commission who is taking the offer. We will see whether we are successful or not. Until then, we will decide on the financials, which is something we then have to release at a later stage, but it is too early now. Let's focus on the Munich site firsthand.
James, without declaring the detailed numbers, what is our planning assumption? Obviously, we can assume that we have not built on the low end nor on the upper end on the EBITDA corridor. There are several factors which are basically impacting the surplus. The second one is obviously our adjusted EPS because that impacts our dividends. The adjusted EPS is very much driven and impacted by the U.S. dollar.
At the time where we have given the capital market stay, we said we do not see any auction in the U.S. in the foreseeable future after the One Beautiful Bill Act. Obviously, there will be spectrum made available in the U.S. You see there is quite a bit of activity also on the satellite side from SpaceX. These were things which can also be used for that surplus. In that given chart which I presented, I said it is predominantly meant to be used for either share increases or buybacks on the DT side, share increases on the U.S. side or buybacks on the DT side. We also want to have strategic flexibility in terms assumptions are changing. Especially when it comes to U.S. spectrum, I think I would say we do not see a spectrum auction up until end of 2027.
I would be less optimistic that this is going to happen given what we know right now. Therefore, this is how we want to use the surplus, and this is why we are vague in how we want to use the proceeds.
Thanks, Christian. With that, we move on to Polo at UBS, please.
Yeah, hi. I've got two questions. The first question is Rodrigo Diehl has taken over as CEO of Germany, but can you comment on how the strategy for the German unit is evolving and what are Rodrigo's priorities? You've obviously already flagged a change in terms of the German fiber strategy. What else is changing in the German unit? The second question is actually just on Starlink. Investors have had a number of questions on how Starlink will impact both broadband and the mobile markets in both Europe and the U.S..
I'm just interested in your perspective. Do you see Starlink as complementary, or do you expect Starlink to take share? Thanks.
Look, I'd like to start with Rodrigo. I told you that we're going to see a reinovation of our team within Deutsche Telekom over the next years. That is taking up here. I have to say I'm very, very happy how the first weeks with the new team is being at Srini now in the U.S. with Illeks' experience and his track record in Europe and Germany, plus his insights into fiber. Being at Rodrigo now in Germany, taking over the lead. He's, by the way, hiring a new B2C head who is there, the former Congstar manager, which we have seen.
We have Abdu, who is the new CTO in the group, another young man with a lot of experience running or being in charge for the infrastructure and the network before. We have a new CIO in the group, K.D, who is coming from India with all his experience about using AI for software development and accelerating this business. There are a lot of people who are now trying to build their own legacy, and that is definitely something which is very encouraging. What we have talked or discussed today about the new direction with regard to fiber is definitely Rodrigo's work. He has intensively spent the first weeks on looking what is working, what's not working, how can we improve the homes connected, how can we improve the take-up rates on the numbers.
I do not want to repeat all the initiatives which we are driving here these days. That was, I think, a tough start for him. He is as well focusing on B2B and the capabilities of stepping up in new services beyond connectivity because traditionally this market is somewhat competitive on the pricing side on the connectivity. The third thing is he is very much focusing on culture in the organization. The way of becoming more uncorporate, this element about becoming more collaborative across the teams. The third one, digitizing the efforts, digitizing the organization, using AI, modernizing the way of how we are doing things, learning from the U.S., by the way, in this regard. This is something which he is driving actively at that point in time. I think these are already four big initiatives which he is on.
We will bring him up into one of the investors' call next year to get to know him. I gave him some relief to work first on the operations and on his team before he's coming here and committing. What you see, what we are announcing today is already his work. With regard to Starlink, to be honest, we can now highly speculate about what's going on there and what is Starlink doing and where is he going to. The first thing what I want to say is that Starlink, and for us, very much relevant is the direct-to-cell connectivity. This is definitely a very attractive complement to our wireless service. Because in the U.S., in large parts of the country, there are no mobile infrastructure. There are no emergency calls possible. For this service, Starlink entirely makes sense.
That is why we made that deal and why we are collaborating with them on the Gen One. They're using our spectrum in this regard. That is then possible that you have an immediate connectivity in these areas. I think that's very important to know that this has to play on the same bands as the bands which you're having in the phone. Otherwise, you have a very complicated switch and a complete registration service. The second is Starlink has now stepped up by buying Dish EchoStar Spectrum. For Gen Two, my understanding is this will not be deployed before 2028, 2029 with new satellites. With this, they might have a different position to play because they have more spectrum. We should not forget that satellite providers are fighting with some technical issues as well.
The first one is that there are limitations with regard to the capacity. Look, we have today 350 MHz of spectrum while these guys are coming with 40 MHz-50 MHz of spectrum. Second, they have latency issues. Thirdly, they have disruption caused by weather or line of sight issues compared to the networks. In the cities where you have this dense traffic, it's very hard to substitute our services. I see that as a very logical adjacency for telecommunication operators. We are very interested to further collaborate with Starlink as we did in the past. In Europe, the situation is, and by the way, whatever we are talking about is very much U.S. because the spectrum which he has now is very much American spectrum. It's less of really globally used spectrum.
The one which globally is available from EchoStar is for renewal in 2027, at least for a lot of European markets. There are regulatory discussions coming up. With regard to the rest, European, I think the homes in Europe are much better served by terrestrial services than in the U.S. The substitution risks to a fiber line from satellite, I do not see that. It is only for houses which are really, let's say, rural, unconnected. In this case, Starlink might make sense. If you have a fiber or a 5G coverage at your house, I do not see a big risk on this one. On top of that, spectrum for Europe is limited in this regard as well. It is not that they can have unlimited spectrum for satellites. I would say the market potential in the U.S. is in these very uncovered areas.
It is an adjacency to mobile communication services. In Europe, I really see that as a niche play.
Okay. Thanks, Tim. Now we move on to Josh at BNP, Exane BNP Paribas.
Thank you, guys, for the questions. The first was just on the updated fiber strategy and the second on fiber CapEx. On the fiber strategy, it looks like you're playing a mixture of offense in the MDU areas and defending more in the rural areas. Is that a fair characterization of how this new strategy has evolved? Perhaps to help us think about the impact of this, could you maybe give us a bit of a steer on what your market share in MDU areas is, what your market share in some of the rural areas you're now targeting is, and how that compares to your nationwide broadband market share would be very helpful.
Secondly, on the fiber CapEx, I know you have not quantified this explicitly, but I think you are due to receive a tax benefit of about EUR 500,000,000 over the next three years from these fiscal rule changes. Is that the right proxy for how we should think about the increased fiber CapEx? Would you go above that tax-saving envelope, as it were, to do more fiber if you needed to? Beyond 2027, should we now be thinking of EUR 100 million-EUR 200 million higher German CapEx as a fair run rate, or is this really just a pull forward of more expensive homes that you would have gotten to later in the decade anyway? Thank you.
Start with the second question. I will never call this a pull forward if the build-out is not ready by 2030.
What kind of pull forward are we looking for then? I would say an indication of around 200 a year is, I think, a good indication. I would use this as a proxy. We have not finalized our planning session yet completely, nor have we discussed it internally. I think that is, so the 500, maybe 550 something, around 200 is the right indication for an annual, let's say, increase of the envelope. It is not going to be a pull forward because that program is running for so long that I would not call this a pull forward. On the other hand, tax benefit from the accelerated depreciation comes to an end in 2028. From 2028, the corporation tax rate in Germany will come down progressively by 5 percentage points, which is also then resulting in tax relief, in progressive tax relief.
Therefore, there is a longer timeline for this equation that we have outlined today, although we have basically been specific on the next three years. Since we're playing ping pong here, I think we're hopeful that this accelerated depreciation will be extended, especially if you see that the money which you basically get is being reinvested into Germany. We can prove that. I think that's a good argument to basically make this like the immediate expensing in the U.S. a more permanent vehicle or tool.
Look, the answer to your first question is you're right. In the rural areas, we have to defend our position. If you look, they are very stable and gaining market share from us where we are not covering. In these areas, we have traditionally high market shares. We want to stop that bleeding by building out in these rural areas.
In the MDUs, we have a lot of MDUs where we have homes passed, but we have no homes connected. If you ask me about, let's say, the market share in MDU areas, it's traditionally very low because this is Vodafone area and the cable area. Therefore, we have their opportunity to grow market share. If you ask me about, let's say, where can we invest in these areas, I would call the mix would be with this additional money 50/50 in MDUs connected and in rural areas as well. I've just looked up the numbers here, so it's around 50/50 if that helps you. I think that is the new allocation of the additional money. What we urgently need is definitely this kind of getting access to the apartments and to the houses. That's definitely something where we need the political support.
Otherwise, these investments are very difficult to monetize. Anyhow, we should give you an update about all the details when implemented. I do not want to release all details here because that, from a competitive angle, is as well something relevant for us that we have a little bit surprise factor as well.
Okay. With that, we move on to thanks, Josh. We move on to Carl at Citi, please. Carl, can we have your questions, please?
That's great. Thank you very much. That's great. Thank you very much, Hannes. Two questions, please. Firstly, in Germany, on the wholesale access revenues, what drove the slowdown in Q3 or recognizing that your CMD guidance was stable? Maybe the better question is, why was the wholesale access revenue growing faster than anticipated in the first half?
Secondly, I was wondering if you can talk a bit about T-Systems, both the growing disparity between public sector and corporate revenue growth rates, and also EBITDA growth. I'm used to talking about margin dilution in enterprise telecoms divisions. Can you talk a bit to the margin growth you're seeing there? Year to date, margins have improved by 100 basis points. Is that just phasing, or are we seeing a structural shift in T-Systems' margins going forward?
Thank you. Look, on the wholesale side, our capital markets guidance was for stable wholesale access revenues for the period of 2023 to 2027. That is what we always have said. So far, we have outperformed the guidance, but now we're seeing volume losses overcompensating upper growth in these areas. That is mainly coming from the weakness of our competitors in the broadband area.
It's a little bit, let's say, the indirect impact of the development of the retail broadband situation here in Germany. In the third quarter 2025, our wholesale access revenues were essentially stable. We are expecting somewhat a similar picture for the next quarter. Here, we are focusing on monetizing fiber footprint with our partners as well. What we are doing for us should be, let's say, accessible and available for our wholesale partners as well. Telefonica or 1&1 or like. We are discussing now how they can improve their fiber utilization as well. So far, I think, or I know that we are in line with our expectations here.
Carl, let me try to give you an answer. I'm not sure whether I'm satisfying or whether you're going to be satisfied with the answer.
Look, first of all, we have a mix of different businesses within T-Systems. You have infrastructure-like business, like the cloud services business or the road charging business, which is obviously very much depending on the capacity utilization of a given infrastructure. The second one is digital solutions, which is predominantly driven by utilization and rate card performance, right? How good is your pricing lever you're providing to your customers? Completely different businesses. The third one is the team around Ferry is laser-focused on efficiencies. He is probably one of the hardest guys when it comes to cost reduction because he knows that his margins are razor sharp and thin. Therefore, he has to prepare also for quarters where things are not happening the way how he wants to see it. The fourth topic is the nature of projects.
Look, first of all, it's the mix I was talking about, whether it's infrastructure-led or more digital solutions-led. Obviously, digital solutions come in with lower margins relative to the infrastructure. The second one is, do you have a lot of A deals, which are very large deals, or do you have a contribution from smaller deals who usually have a better profitability? This is why I was causing you. Don't read too much into that 23% EBITDA increase because there's volatility coming from different angles, and you don't know how the business mix is going to look like in the upcoming quarters. It is much, much harder to predict relative to the infrastructure business, which we're running outside T-Systems. What I'm seeing is, look, we're coming from negative cash contribution from T-Systems, and we are now in positive territory. The operating free cash flow is actually growing.
I think this is where I'm saying, as a finance guy, I don't expect you to give me 10% of the overall pie on EBITDA, right? I want to see a continuous trend improvement so that we don't have to discuss T-Systems as a financial, let's say, challenge. They're helping us in kind of pull through by selling other businesses because they're solving complex issues, especially with the public sector. For example, remember the COVID app, which was basically being built between T-Systems and SAP that helps you in those sectors. This is kind of a pull-through effect, I would say, you're going to have from the infrastructure business.
This is why I'm happy, but I can't give you kind of an equation, whether it's accretive or dilutive, because it depends on the mix of the business, which is coming in every quarter, and that changes.
Okay. Thank you. Thanks, everyone, for the Q&A, which is now coming to an end. I think Tim would like to make a few closing remarks, and then I take back from you.
Look, guys, thank you for the questions. Look, my summary of this quarter and even looking for the end of the year is this is everything is well on track with regard to the overall capital markets targets. We had this concern about the German broadband market. We have a great plan now worked out, which is in execution. We have a good team, which is now pushing for that one, young, fresh leaders here.
On top of that, we are able to increase our dividend to EUR 1, which is another commitment. It is the highest dividend ever paid in the history of Deutsche Telekom. On top of that, we are committing to the share buyback, which was highly and well received from the market environments. All the acquisitions are well on track. No kind of negative surprise. The opposite is the case. For instance, with U.S. Cellular, we have a very good development as lighting out one issue. We have cleaned up the portfolio again because after a long, long painful period, Romania is out of the portfolio, which has now resolved as well. Deutsche Telekom is quickly taking the opportunity of the sovereignty discussion here in Europe, where we see big opportunities. There is definitely the AI factory, which I want to mention here.
We were able to develop this whole concept to implementation, ready to use within six months. 10,000 GPUs. That is the biggest GPU in Europe at one single place. On top of that, it is increasing the capacity of GPUs in Germany by 50% in one single step. I can tell you this is giving us huge credibility, not only in the public environment, but as well for business use. We are going into the defense sector, both on the T-Systems side and as well on the DTCP side, which is helping. The last thing which I want to mention is expect more from us with regard to AI and the AI implementation. Great ideas in the organization, agent models enabling new opportunities here for us, which we are evaluating.
Strong momentum here in our company, good use cases and success cases as well from the U.S. now swapping over here to Germany and other markets. Next year, it's going to be an AI year. That is something which is helping us to not only increase our customer orientation, but as well our efficiencies here, which is well on track. I'm overall very happy with the situation here. We will do everything to improve the financials, not doing the stupid and ridiculous things here. We'd like to thank you for your trust and have a nice day, guys.
Thank you, guys.
Thank you. If you would like to ask further questions, please contact the IR department. We look forward to hearing from you again and see you soon. Thank you very much. Bye-bye.